Académique Documents
Professionnel Documents
Culture Documents
and WiMAX
Comprehensive Report
Presented by the
International Engineering Consortium
Copyright 2004 by Professional Education International, Inc. All rights of reproduction, including that of
translation into foreign languages, are reserved. Requests for republication privileges should be addressed to
Publications Department, International Engineering Consortium, 300 West Adams Street, Suite 1210, Chicago,
Illinois 60606-5114, USA.
All opinions expressed in the Broadband Wireless and WiMAX comprehensive report are those of the authors and
are not binding on the International Engineering Consortium or Professional Education International.
ISBN: 978-1-931695-30-X
International Engineering Consortium
300 West Adams Street, Suite 1210
Chicago, Illinois 60606-5114, USA
+1-312-559-4100 voice +1-312-559-4111 fax
publications@iec.org www.iec.org
ii
iii
This past year, the University Program has provided more than 250 grants to professors to
attend the educational opportunities offered at IEC Forums. Because of their participation in
IEC Forums, professors created or updated 400 university courses; more than 10,000 students
were impacted by these new and upgraded courses, and more than 5,000 students benefited
from improvements made in university laboratories. Since its inception in 1984, the University
Program has enhanced the education of more than 500,000 students worldwide.
University of Glasgow
Howard University
Illinois Institute of Technology
University of Illinois at Chicago
University of Illinois at
Urbana/Champaign
Imperial College of Science, Technology
and Medicine
Institut National Polytechnique de
Grenoble
Instituto Tecnolgico y de Estudios
Superiores de Monterrey
Iowa State University
KAIST
The University of Kansas
University of Kentucky
Lehigh University
University College London
Marquette University
University of Maryland at College Park
Massachusetts Institute of Technology
University of Massachusetts
McGill University
Michigan State University
The University of Michigan
University of Minnesota
Mississippi State University
The University of Mississippi
University of Missouri-Columbia
University of Missouri-Rolla
Technische Universitt Mnchen
Universidad Nacional Autnoma de
Mxico
North Carolina State University at
Raleigh
Northwestern University
University of Notre Dame
The Ohio State University
Oklahoma State University
The University of Oklahoma
Oregon State University
Universit dOttawa
The Pennsylvania State University
University of Pennsylvania
University of Pittsburgh
Polytechnic University
Purdue University
The Queens University of Belfast
Rensselaer Polytechnic Institute
University of Southampton
University of Southern California
Stanford University
Syracuse University
University of Tennessee, Knoxville
Texas A&M University
The University of Texas at Austin
University of Toronto
VA Polytechnic Institute and State
University
University of Virginia
University of Washington
University of Wisconsin-Madison
Worcester Polytechnic Institute
Consortium-Affiliated Universities
The International Engineering Consortium is affiliated with the following leading high-tech universities:
Columbia University
Cornell University
Auburn University
Drexel University
Lehigh University
VA Polynechnic Institute
and State University
University of California
at Berkeley
Marquette University
Polytechnic University
Georgia Institute of
Technology
Northwestern University
Purdue University
University of California,
Santa Barbara
Howard University
Massachusetts Institute of
Technology
Rensselar Polytechnic
University
The University of
Massachusetts
University of Southern
California
Worcester Polytechnic
Institute
Stanford University
Clemson University
University of Illinois at
Urbana/Champaign
Syracuse Univesity
University of Tennessee,
Knoxville
McGill University
The Queens University of Belfast
Technische Universitt Mnchen
Universidad Nacional Autnoma de Mxico
Universit dOttawa
University of Glasgow
University of Southampton
KAIST
University of Edinburgh
University of Toronto
The IEC wishes to thank the following publications that support the Consortium and its initiatives
throughout the year by participating in the Media Partners program.
1909 Avenue G
Rosenberg, TX 77071
877-588-1649
www.privatebroadband.com
Table of Contents
About the International Engineering Consortium ........................................................................................iii
University Program and Sustaining Sponsors.............................................................................................. iv
Consortium-Affiliated Universities and International Affiliated Universities.............................................. v
Media Sponsors and Partners....................................................................................................................... vi
Table of Contents by Author......................................................................................................................xiii
Acronym Guide......................................................................................................................................... 483
The Business of Wireless
On the Selection of the Optimum Emerging Wireless Broadband Technology ........................................... 1
by a Mobile Operator
Maria Afioni, M.Sc., 3G Telecommunications Engineer, Switching and Network
Management Department, BSS and New Technologies Division, 3G Technology
Section, COSMOTE Mobile Telecommunications S.A.
Petros Lekakos, M.Sc., 3G Telecommunications Engineer, Switching and Network
Management Department, BSS and New Technologies Division, 3G Technology
Section, COSMOTE Mobile Telecommunications S.A.
Nikolaos Lepidas, Dr.-Ing., 3G Telecommunications Engineer, Switching and Network
Management Department, BSS and New Technologies Division, 3G Technology
Section, COSMOTE Mobile Telecommunications S.A.
George Lyberopoulos, Dr.-Ing., 3G Telecommunications Engineer, Switching and
Network Management Department, BSS and New Technologies Division,
3G Technology Section, COSMOTE Mobile Telecommunications S.A.
Fiber Optics for Wireless Telecommunications.......................................................................................... 21
Ken Chauvin, Marketing Manager, Technology, Corning Cable Systems
Emerging Wireless Technologies ............................................................................................................... 55
Another Telecommunications Technology Disruption
Samuel Hughes, Senior Manager, National Advisory Services, Ernst & Young LLP
Trent Tishkowski, Manager, National Advisory Services, Ernst & Young LLP
Integration of Wireless Access with Wireline Networks: OAM&P Support.............................................. 59
Architecture with ITU-tML Technology
Dr. Wei Liu, Adjunct Professor, Shan-Tou University, and Senior Consultant,
BellSouth
How to Make Money in Broadband Wireless............................................................................................. 67
A Pragmatic Guide to Operator Profitability
R.J. Mahadev, Co-Founder and Executive Team Member, EuroWireless S.A
Converged Public and Enterprise Wireless Networks ................................................................................ 79
Sunil Mahajan, Chief System Engineer, Hughes Software Systems, India
vii
viii
A Business Justification: WiMAX Service Providers and Security Investments ..................................... 189
Ralph P. Martins, Jr., Senior Consultant, Booz Allen Hamilton,
and Graduate Student, George Washington University
WiMAX: The Next Generation of Wireless Communication?................................................................. 197
T I M Shaniur Nabi, Research Student, Centre for Advanced Technology
in Telecommunications, RMIT University
Richard Harris, Professor and Director, Centre for Advanced Technology
in Telecommunications, RMIT University
WiMAX Promises a New Era in Telecom................................................................................................ 207
Athena Platis, Wireless Industry Analyst, National Telecommunications
Cooperative Association (NTCA)
WiMAX: Outlining Business Strategies ................................................................................................... 211
Kotni Mohana Rao, Senior Software Engineer, Wipro Technologies,
Bangalore, India
WiMAX: Final Destination or Path .......................................................................................................... 219
Amit Rawal, Engineer, Himachal Futuristic Communications Limited
Simulation of 802.16a Deployment Scenarios and Their Performance Analysis ..................................... 227
Dr. Daniel Rodellar, Project Leader and Telecommunication Engineer,
Swisscom Innovations
M. Eng. Ludovic Fournier, Information and Communication
Technology Engineer, Swisscom Innovations
Dr. Christian Fischer, Senior Research and Development Engineer,
Swisscom Innovations
Wi-Fi and WLANs
Wireless Networks for Real-Time Multimedia Communications ............................................................ 239
Celal Ceken, Lecturer, Electronics and Computer Education Department,
Kocaeli University, Turkey
Ismail Erturk, Lecturer, Electronics and Computer Education Department,
Kocaeli University, Turkey
Cuneyt Bayilmis, Research Assistant, Kocaeli University, Turkey
Cellular and WLAN Convergence............................................................................................................ 259
Chandrakantha T.N., Software Engineer, Global Edge Software Ltd., India
Adaptive Antenna Arrays for WLAN Communication Systems .............................................................. 267
Dr. Sathish Chandran, Chief Telecommunication Consultant
and President, International Union of Radio Science (URSI), Malaysia
Wireless LAN: Security, Reliability, and Scalability ............................................................................... 275
Vikas Koul, Technical Leader, Hughes Software Systems
Prashant Vashisht, Senior Technical Leader, Hughes Software Systems
Masood Ul Amin, Project Manager, Hughes Software Systems
ix
xii
xiii
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
To propose and apply a methodology, which may facilitate mobile network operators to select the
optimum emerging wireless broadband technologythat best serves their business strategy
taking into account a variety of technical and nontechnical parameters.
I. Introduction
Mobile operators are increasingly faced with the dilemma of selecting the optimum emerging wireless
technology, such as WiMAX, Mobile-Fi, UMTS TDD, and HSDPA. The crucial question is, Should an
operator introduce a new wireless technology earlier and thus enhance its service offerings disregarding
possible interoperability issues, or should follow a standards-proposed evolution path (e.g.,
3G/WCDMA3G/TDDHSDPA) despite (possible) late availability of commercial systems/terminals?
Each technology exhibits certain advantages; however, the operators final selection should be in
accordance with its overall business strategy.
In this paper, initially, a comparative study is conducted, revealing the current standards/systems
development status of the emerging wireless technologies. As a second step, to facilitate a second- or
third-generation (2G/3G) mobile operator deciding on the optimum future wireless technology, an
evaluation methodology is proposed and applied. The methodology employs an algorithm, which takes
into account both technical and business-oriented parameters such as investment, risk, profitability,
fulfillment of customer/market needs, and marketability. Indicative evaluation results for two operators
that exhibit different business strategies (conservative vs. aggressive) are presented in section III, while
conclusions are drawn in section IV.
II. Emerging Wireless Technologies: A Comparative Study
A. WiMAX (IEEE 802.16)
Overview of WiMAX
While wireless local area network (WLAN) technology is designed to add mobility to private wired
LANs, WiMAX is designed to deliver a metro area broadband wireless access (BWA) service combining
nationwide coverage with high bandwidth. WiMAX promises to open new, economically viable market
opportunities for operators, wireless Internet service providers (WISPs), and equipment manufacturers.
The flexibility of wireless technology, combined with the high throughput, scalability, long-range and
quality of service (QoS) features will help fill the broadband coverage gaps and reach millions of new
residential and business customers worldwide.
WiMAX Standards: The 802.16x Family
The 802.16x standards genealogy [2] is shown in Figure 1, while a comparison study between the 802.16
versions is depicted in Table 1. As far as the WiMAX systems availability is concerned, we may note
that the following is true:
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
802.16
(Dec 2001)
802.16c
(2002)
802.16a
(Jan 2003)
802.16REVd
(Q3 2004)
802.16e
(2005 exp.)
802.16a/REVd
802.16a: Jan 2003
802.16e
Standards Status
Dec 2001
Airwave
Spectrum
1066 GHz
< 11 GHz
< 6 GHz
802.16REVd: Q3 2004
Channel BWs
Modulation
Same as 802.16a
Channel
Conditions
LOS Only
NLOS
NLOS
Typical Cell
Radius
25 km
25 km
Mobility
Fixed
Fixed
Pedestrian Mobility
Regional Roaming
Bit Rate
Up to 75 Mbps at 20 MHz
channelization
Up to 15 Mbps at 5 MHz
channelization
Applications
Point-to-Point Applications
Point-to-Multipoint Applications
Point-to-Multipoint
Applications
802.16a CPE: External box connected to PC with outside antenna; 802.16REVd CPE: External box connected to
PC with built-in antenna; 802.16e CPE: PC card
3
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
The BS uses an outdoor antenna (sectored/directional or omnidirectional) to send and receive high-speed
data and voice to CPE, thus eliminating the need for extensive and expensive wireline infrastructure and
providing highly flexible and cost-effective last-mile solutions. A fixed CPE typically uses directional
antenna while mobile or portable CPE usually uses an omnidirectional antenna.
Mobility will be supported by 802.16e and will target only urban usage, with up to 60 km/h
vehicle speed to maintain optimum throughput performance
802.16REVd applications:
Indoor broadband access for residential users (e.g., high-speed Internet and voice over Internet
protocol [VoIP])
802.16e applications:
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
Terminals Availability 2
802.16a chipsets are currently available
802.16REVd chipset3, will be available by the mid2005
802.16e chipsets embedded in laptops and later on in other devices, enabling mobility (portable
Internet) will be available in 2006
Deployment Status [1]
Currently, a number of trials are in progress:
TowerStream and Bellsouth (New York, Chicago, and North Carolina for Nextel [USA])
BT (U.K.)
AT&T
Verizon (Oregon, USA)
In China, WiMAX will be deployed in six cities while Malaysia, Japan, New Zealand, and parts of
Australia are willing to introduce WiMAX technology in their networks. Finally, Wi-Lan in France is
testing Libra 3000, a fixed wireless broadband technology that is offering a 802.16a network.
B. Mobile-Fi (IEEE 802.20)
Overview of Mobile-Fi
The next generation of mobile connectivity beyond WiMAX will be based on the (pending) Mobile-Fi
standard, which is being developed from the ground up as a mobile specification (see Table 2). The
physical (PHY) layer (i.e., radio front end and baseband signal processing sections) and data-link layer
(i.e., MAC protocol) are specifically designed for mobile requirements, such as adaptive antenna arrays.
Mobile-Fi Technology (IEEE 802.20)
Airwave
Channel BWs
Mobility
New PHY & MAC optimized for packet data and adaptive antennas.
Packet-oriented architecture.
Technology
Service Provider
End User
2
3
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
Pure IP
Architecture
(IEEE 802.20)
RADIUS
Foreign IP
Network
RADIUS
RADIUS
Mobile
Station
IP Base
Station
Home Access
Provider
Network
Home IP Network
SS7
Network
VLR
HLR
Broker Network
IP
Network
Home IP Network
RADIUS
Home IP
Network
RADIUS
Radio
Network
IP Network
Mobile
Station
BTS
BSC
RADIUS
Broker
Network
PDSN
Home IP
Network
The air-interface shall support different modes of mobility from pedestrian (3 km/h) to very high
speed (250 km/h)
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
Deployment Status
Currently, a number of trials are in progress:
C. UMTS TDD
3GPP Standards and TDD Technologies
The 3GPP family of standards consists of three alternative air-interface standards, which all share higher
layer protocol stack and core network architecture. These air-interface standards are UMTS frequency
division duplex (FDD) (or more commonly wireless code division multiple access [WCDMA]) and two
variants of UMTS TDD. These two types of UMTS TDD technologies are characterized by different chip
rates and different bandwidths. More specifically, the chip rates are 3.84 Mcps and 1.28 Mcps, with the
corresponding bandwidths being 5 MHz and 1.6 MHz. Accordingly, the two TDD types are referred to in
literature as follows:
High chip rate (HCR) TDD and low chip rate (LCR) TDD or as
Time divisioncode division multiple access (TDCDMA) and time division-synchronous code
division multiple access (TDSCDMA)
Source: Mark Klerer, former chair of the 802.20 working group and an executive director at Flarion, the standards
main technical contributor
7
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
UTRA FDD
UTRA TDD
DSCDMA
TDCDMA
TDSCDMA
3.84 Mcps
3.84 Mcps
1.28 Mcps
Carrier Bandwidth
5 MHz
5 MHz
1.6 MHz
Frame Length
10 ms
10 ms
10 ms
Modulation
QPSK
QPSK
QPSK
Spreading Factor
Up to 256
Up to 16
Up to 16
Common antenna with UMTS FDD system (via the use of appropriate UMTS FDD/TDD duplexer)
New Node-B
New RNC (dual-mode RNC [supporting both TDD and FDD] is expected)
Existing GPRS/UMTS FDD core network infrastructure
From a standards perspective, UMTS TDD intersystem/intermode handovers to/from GSM/GPRS/UMTS FDD are
supported. Handovers between UMTS TDD and FDD systems, however, have not been commercially
performedin trials only.
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
TDD is harmonized with FDD at the chip rate level, making TDD functionality a highly costeffective complement to existing FDD designs.
TDD is fully and seamlessly integrated with FDD in the UMTS air interface standards6 and
provides an opportunity for operators to make the most of their allocated 3G unpaired spectrum
while exploiting the inherent integrated user mobility and service roaming features of FDD and
TDD.
TDD is cost efficient for network deployment (improved spectrum efficiency, no need for softhandover and sharing of higher layer protocol stack and core network architecture with FDD). It
offers scalable capacity for hot spots, where combined voice and data traffic must be served
efficiently, through a tiered architecture, which may include macro, micro, and picocells.
TDD can provide an attractive wireless data solution for wide area coverage and islands of
coverage where there is a need for wireless asymmetrical digital subscriber line (ADSL) services
and outdoor WLANlike services.
TDD can adopt techniques such as channel sensing and adaptive antennas in order to improve
performance, coverage and capacity.
TDD is 3GPP standards compliant. TDCDMA was introduced in R99 while TDSCDMA was
in R4.
Generally speaking, the combination of FDD technology and TDD technology in third-generation UMTS
can result in significant technological and financial benefits, including improved return on investment
(ROI) for network operators, equipment manufacturers and application developers. TDD can leverage the
infrastructure of a first-wave FDDonly rollout of 3G, which can further reduce the cost of deployment,
by being a part of a multilayered hierarchical deployment strategy. National regulatory bodies have
endorsed the proposed three-tiered network deployment and have allocated paired frequency bands for
FDD and unpaired licensed frequency bands for TDD as well as unlicensed bands for TDD.
Handovers between TDD and FDD systems, however, have not been commercially performedin trials only.
9
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
Terminals Availability
TDCDMA (packet data implementations):
Personal Computer Memory Card International Association (PCMCIA) cards and desktop
modems are commercially available
Chipset is expected to be available in 2004
TDSCDMA:
Australia: IQ Networks
Germany: Airdata
Malaysia: Atlas One
New Zealand: Woosh Wireless
South Africa: Sentech
United States: Maui Sky Fiber
Hong Kong: PCCW
Also, a number of TDCDMA (packet data implementations) and TDSCDMA trials are in progress:
D. HSDPA
Overview of HSDPA
HSDPA is based on WCDMA evolution and is standardized as part of the WCDMA 3GPP Release 5
specification. HSDPA is a packet-based data service in WCDMA downlink with data transmission up to 8-10
Mbps over a 5MHz bandwidth. This means that HSDPA can coexist on the same carrier (5 MHz) as the
current 3G services, allowing mobile operators to introduce greater capacity and higher data speeds into
existing 3G networks. The most important thing with HSDPA is not the peak rate but the throughput capacity,
which increases significantly. This leads to more users being able to use high data rates on a single carrier.
HSDPA introduces several new key techniques, such as the following [17]:
The service offerings are provided to customers via PCMCIA cards and/or desktop modems.
10
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
HSDPA is based on a new distributed architecture enabling low delay link adaptation, channel quality
feedback and HARQ processing. This is accomplished by incorporating many of the key scheduling and
control processes at the Node-B, as opposed to the RNC, and thus closer to the air-interface. Specifically,
the MAC functionality, which fully resided in the RNC in R99, is split between the RNC and Node-B in
Rel5. In Rel5, most of the key MAC functions critical to delay and performance are defined by the
MAC-HS, which is located in the Node-B [15].
HSDPA can be introduced where the WCDMA RAN has been deployed. This means that it will make use
of existing infrastructure in all parts of the network. It is envisaged, however, that initially, high-speed
services will be offered in a small part of the network, for example, in areas with heavy traffic, such as
city centers, office areas, and airports.
Although the main emphasis in air-interface optimization can be seen in the area of downlink high-datarate support, the uplink also needs attention. HSUPA will be standardized as part of the WCDMA 3GPP
Release 6 specification, which is ongoing in 3GPP and will be finalized by the end of 2004. Enhanced
data rates in the uplink will benefit the end user (e.g., in file transmission or when such office applications
as NetMeeting are used). It is anticipated that many of the same techniques used in HSDPA will be used,
but these still need to be formalized. Also, an optimized uplink will be introduced (comparable to that in
the downlink) to support lower terminal output powers.
Standards Availability
WCDMA 3GPP Release 5 specifications released in March 2002. HSDPA is one of the key Rel5
features that provide significant spectral/network efficiency, performance and functionality
advantages over the R99 and Rel4 standards.
WCDMA 3GPP Release 6 specifications will be available by the end-2004. The HSUPA will be
included in these release standards.
11
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
Consequently, network operators that implement HSDPA will achieve a lower delivery cost per bit.
HSDPA/HSUPA will benefit end-users by providing the following:
Siemens and NEC are shipping their second-generation Node-B. For support of HSDPA, the U.S.
PCS 1900 MHz band capable Node-B requires software-only upgrades which can be downloaded
remotely when they become available in 2005. The first commercial HSDPAcapable terminal is
planned to be a PCMCIA form-factor data card, which will support the market trials and commercial
deployments expected to take place in 2005.
Nortel Networks has been shipping its HSDPAready Node-Bs for sometime now and all Nortel
Networks Node-Bs currently deployed across the globe will support HSDPA. Nortel Networks
expects HSDPA trials to take place later this year, with commercial deployments in 2005.
Ericsson successfully demonstrated a radio network emulator at the CTIA Wireless 2003 trade show.
Customer trials will start in 2004 and commercial systems will be available in the second half of 2005
globally.
Nokia Research Center Dallas completed its first HSDPA demonstration in January 2003. The system
was demonstrated publicly in Cannes at the 3GSM World Congress in February 2003 as well as in
New Orleans at CTIA in March 2003. Nokias current commercial UMTS system can be software
upgraded to commercial HSDPA, which will be available in the second half of 2005 timeframe.
NTT DoCoMo has indicated that they intend to have HSDPA live by the second half of 2005.
12
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
In Europe also, 3G mobile operators are now whispering that they are prepared to bring their HSDPA
rollout forward considerably, and word coming out of the vendors supports this. Even O2, which is
not the habit of trailblazing technology, has publicly stated that it is targeting HSDPA in 2006.
E. Comparative Results
See Table 4 for a comparison.
Emerging Wireless Technology Evaluation
WiMax (802.16)
802.16 & 802.16a:
completed
Standards
Availability
802.16REVd: Q3/2004
802.16e: within 2005
System
Availability
Terminal
Availability
Spectrum
Requirements
UMTS-TDD
TDCDMA: 3GPP (R99 &
onwards)
TDSCDMA: 3GPP (R4 &
onwards)
TDCDMA (packet data
implementations):
Commercial
TDSCDMA: Trials
TDCDMA (packet data
implementations):
Commercial (PCMCIA
cards, desktop modems),
chipset (to be available in
2004)
TDSCDMA: Prototype
(single-mode) handset
available commercially
available handsets by mid2005
HSDPA
Licensed
Licensed
Licensed
802.20: end
2004
Not available
Not available
Other:
Cell Range
LOS/NLOS
Data Rates
Mobility/Roaming
Handover
QoS
Security
Applications
Mobile-Fi (802.20)
Max range: 15
km
Data rate: up to
16 Mbps
802.20: NLOS,
PTMP
applications,
full mobility,
roaming, QoS,
security
Interworking with
3G: Handover,
Mobility/Roaming,
Impact on Existing
Network, MultiMode Terminals
Supported:
Interworking
with 3G,
Handover,
Roaming to be
supported
Impact on
Network: new
Base Stations
Interworking with 3G
(FDDWCDMA):
Performed in trials only (no
commercial support)
Impact on Existing
Network: TDCDMA
(Packet data
implementations) [new
Node-B, new RNC (dual
mode FDD/TDD is
expected)], TDSCDMA
(new [smart] Antenna, new
Node-B, enhanced
[upgraded] RNC)
Trials in progress
Not available
Interworking with
3G (FDD
WCDMA)
Impact on Existing
Network: Node-B
hardware/software
upgrade, RNC
software upgrade
New terminals
13
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
Investment
Investment
Payback
Payback
Period
Period
Implementation
Implementation
Timeframe
Timeframe
Profitability
Profitability
Competitive
Competitive
Advantage
Advantage
Outcome
Regulatory
Regulatory
Requirements
Requirements
Optimum
Technology
Companys
Companys
Brand Name
Brand Name
Customer/
Customer/
Market
Market
Needs
Needs
Response to
Response to
Competition
Competition
Risk
Risk
Marketability
Marketability
Investment. The introduction of a new wireless technology needs certain investment. Investment
is affected by both CAPEX9 and OPEX10 costs.
Customer needs. Operators that will manage to satisfy market needs (such as, real-time multimedia
services, service personalization and customization, low tariffs, service differentiation, single terminal
equipment for all environments) will dominate the wireless market.
These factors are related to technical (see previous sections) and nontechnical parameters (risk, regulatory).
Investment required for the introduction of new elements (e.g., base stations, switches), as well as existing network
(2/2.5/3G) infrastructure upgrades (e.g., Node-Bs, BSC/RNC).
10
Investment required for spectrum acquisition (if applicable), network operation and maintenance, customer
acquisition/advertisement campaigns, transport network upgrades, handset subsidies, etc.).
9
14
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
Response to competition: Mobile operators may adopt a new technology simply as a response to
competition.
Competitive advantage: A technology that enables the operator to offer service differentiation
(mobility, handover, roaming) and QoS, enhancing thus the operators service portfolio, business
role and brand name may provide a competitive advantage.
Marketability: A new technology supporting seamless interworking with the existing network/services
(2/2.5/3G)via multimode terminalswhile allowing the introduction of brand new applications
(e.g., VoIP, real-time advanced multimedia services) may offer marketing opportunities.
Risk: Business, cost, and companys size risk in conjunction with the business model and the
marketability of the services on offer should be also considered. Risk is also affected by the
payback period. The possibility of making an overestimated investment (including new
recruitment) and finally offering a cannibalized service should be eliminated.
Implementation timeframe: This factor corresponds to both (commercial) system and terminal
availability.
In order to simplify the evaluation process, we map the decision-making factors onto five more generic
categories, which are given below:
Financial
(Q1)
(Q2)
(Q3)
Investment
Payback period
Contribution to profitability
Marketing
(Q4)
(Q5)
(Q6)
(Q7)
(Q8)
Regulatory
(Q9)
Regulatory requirements
Risk
(Q10)
Implementation
(Q11)
Implementation timeframe
B. Algorithm Description
The general guidelines of the proposed algorithm [5], [6], [7] are given below:
15
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
Tradable quantities are added together, so that it is the sum of the question responses (see Note 2)
that contributes to the overall figure of merit.
Not-tradable quantities (see Note 2) are multiplied. Not-tradable criteria contribute independently
to the quality of the triggering event. Multiplying independent criteria (or independent groupings
of tradable criteria) emphasizes the importance of each one individually and minimizes the
potential selection of triggering events that are extremely poor performers.
To weight the criteria so that they reflect the desired emphasis, tradable question responses in the
same added grouping are each multiplied by a weighting factor to reflect their relative level of
importance. Multiplied question responses in the same grouping are each raised to an exponent to
reflect their relative level of importance. Weights within the same (added or multiplied) group are
then independently normalized (see Table 5).
Note 1: The quantity 1 should be abstracted to shift the range from zero to four, and then the entire
expression is divided by four to scale it between zero and one.
Note 2: As shown in Figure 5, the tradable quantities are (payback vs. profitability) and (competitive
advantage and marketability vs. risk).
Decision-Making
Parameters
Criteria
Q1
Company Name
Q4
Customer Needs
Q5
Response to Comp.
Q7
Regulatory Reqs
Q9
Impl. Timeframe
Q11
Q2, Q3, Q8,
Cost
Rest
Q6, Q10
Weight.
Factors
Scenario #1:
Scenario #2:
Conservative
Aggressive
VALUE
NORMALIZED
VALUE
Payback
Q3, Q8
Q2
Risk
Q6
Q10
0,217
0,048
w12
0,174
0,095
w13
0,174
0,238
w14
0,043
0,143
w15
0,130
0,048
w16
0,130
0,238
w18
0,130
0,190
21
w21
0,556
0,750
w22
0,444
0,250
w31
0,556
0,600
w32
0,444
0,400
w41
0,625
0,200
w42
0,375
0,800
w51
0,625
0,500
w52
0,375
0,500
9
Profitability
Competitive Adv.
Q3
Q8
8
Group 1
Group 2
Q2, Q3, Q8
Q6, Q10
16
VALUE
9
Marketability
NORMALIZED
w11
23
Prof/Comp. Adv.
VALUE
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
Investment
Cost
Company
Company
Name
Name
Payback
Payback
Customer
Customer
Needs
Needs
Resp.
Resp.to
Competition
Compet.
Regulatory
Regul.
Reqs
Reqs
MarketMarketability
ability
Profitability
Profitability
Competitive
Competitive
Advantage
Adv.
Risk
Risk
Implem.
Timeframe
Not-Tradable
Tradable
The second scenario (aggressive approach) describes a 2G/3G operator who does the following:
Takes risks in order to increase competitiveness and gain more market share (by introducing new,
bandwidth-consuming applications), disregarding interoperability issues11
Endeavors to cover all customer/market needs as soon as a (new) technology enables it12
Finally, Table 6 presents the overall scoring based on the values given to the evaluation criteria as well as
on the weighting factors values.
Note 1: The values assigned to weights reflect the different operators business strategies (see Table 6).
11
Examples might be those (2G/3G) operators who introduced i-mode technology, which necessitated the utilization
of new terminals while (currently) no interworking with 3G is supportedat least on the terminals side.
12
Note that the introduction of a new technology may be dictated by the competition itself (churn reduction,
alternative to a service offered by a competitor, etc.).
17
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
Note 2: The values given in Table 6 are the same for both scenarios. Note that the overall scoring (see
formula above) depends (also) on the values given to weighting factors (see Table 6) that differentiate the
operators strategy.
Note 3: The values of the decision-making parameters were given, taking into account a three-year
perspective on system implementation availability.
Scenario #1: Conservative
Financial
#
TECHNOLOGY
Cost
Marketing
Payback
Contr. To
Period
Profitability
Company's
Image &
Risk
Customer
Market-
Response
Compet.
Regulatory
Needs
ability
to Comp.
Adv
Reqs
Risk
Impl. TF
Implem.
SCORING
Timeframe
WiMAX (802.16)
Brand Name
3
0,5103782
Mobile-Fi (802.20)
0,6826044
UMTS TDD
0,5434879
HSDPA/HSUPA
0,7820767
Risk
Impl. TF
TECHNOLOGY
Cost
Marketing
Payback
Contr. To
Period
Profitability
Company's
Image &
Customer
Market-
Response
Compet.
Regulatory
Needs
ability
to Comp.
Adv
Reqs
Risk
Implem.
SCORING
Timeframe
WiMAX (802.16)
Brand Name
3
0,4613516
Mobile-Fi (802.20)
0,7212362
UMTS TDD
0,5275592
HSDPA/HSUPA
0,7139285
For conservative operators (Scenario I), HSDPA technology is the optimum technology to be
followed, assuring smooth network evolution and seamless service provisioning while needing
minimum investment and infrastructure upgrades/changes. Such operators should continue rolling out
3G and start building out HSPDA when available (on existing 3G sites) and finally, migrating highend users to HSDPA.
For those operators, the aggressive ones, who take risks on customers satisfaction, who want to be at
the forefront of the technology, who are eager to capture the critical mass of early adopters, should
(also) consider Mobile-Fi as an option. We should stress however, that the final decision would be
greatly affected by the system/terminal commercial availability (Mobile-Fi vs. HSDPA/HSUPA)13.
IV. Conclusions
It is envisaged that either standards-driven wireless broadband technologies (e.g., UMTS TDD, HSDPA,
WiMAX, Mobile-Fi) or proprietary ones (FLASHOFDM, iBurst, etc.) will create new market opportunities
for mobile operators, while, on the other hand, will enable end users to experience true real-time multimedia
services (e.g., voice/video/videoconferencing over IP, video/audio streaming, etc.).
As a result, the dilemma of selecting the optimum emerging wireless technology, i.e., the one that best
serves operators strategy, has already started torturing mobile operators. The crucial question is, Should
we introduce a new wireless technology earlier and thus enhance service offerings disregarding possible
13
18
At this stage, though, it seems premature to suggest certain implementation timeframes, as standardization and/or
trials are in progress. That is why we have assigned the same value for the implementation timeframe parameter
for all technologies.
Maria Afioni, M.Sc., Petros Lekakos, M.Sc., Nikolaos Lepidas, Dr.-Ing., and George Lyberopoulos, Dr.-Ing.
We have proposed an evaluation methodology, which enables an operator to select the optimum
emerging technology, by suggesting certain parameters so as to reflect its business strategy. The
methodology takes into account a variety of technical and nontechnical parameters (such as, risk,
marketability, profitability, and payback).
We have applied the proposed methodology for two scenarios exhibiting different operators
business strategies. The first scenario represents a conservative approach, while the second an
aggressive one.
Aggressive operators should also consider Mobile-Fi as an option. The final decision, however,
will be greatly affected by the system/terminal commercial availability (Mobile-Fi vs.
HSDPA/HSUPA).
In any case, the introduction of an emerging wireless technology will bring innovative, advanced realtime multimedia applications (VoIP, video over IP, videoconference over IP, audio/video streaming, etc.),
and as such, it will contribute to churn reduction, generate of new revenues streams, and strengthen the
companys position in the market.
V. References
[1]
[2]
Caroline Gabriel, WiMAX: The Critical Wireless Standard 802.16 and Other Broadband Wireless
Options, Monthly Research Report, BluePrint Wi-Fi, October 2003.
[3]
Mark Klere
[4]
Caroline Gabriel, Examining the Hotzone: The Impact on Business and Community
Communications, Monthly Research Report, BluePrint Wi-Fi, January 2004.
[5]
Anne DePiante Henriksen and Ann Jensen Traynor, A Practical R&D Project Selection Scoring
Tool, IEEE Trans. on Eng. Management, Vol. 46, No.2, May 1999.
[6]
C.N. Konstantinopoulou, K.A. Koutsopoulos, G.L. Lyberopoulos, and M.E. Theologou, Core
Network Planning, Optimization and Forecasting in GSM/GPRS Networks, SCVT 2000, Leuven.
19
On the Selection of the Optimum Emerging Wireless Broadband Technology by a Mobile Operator
[7]
[8]
Relative Assessment of UMTS TDD and WLAN Technologies, UMTS Forum Report 28 by
TDD Ad-Hoc Group.
[9]
WiMAX, making ubiquitous high-speed data services a reality, Alcatel Strategy White Paper.
[10] 3GPP TS 25.855, High Speed Downlink Packet Access (HSDPA); Overall UTRAN Description.
[11] 3GPP TS 25.856, High Speed Downlink Packet Access (HSDPA); Layer 2 and 3 Aspects.
[12] 3GPP TS 25.876, Multiple-Input Multiple-Output Antenna Processing for HSDPA.
[13] 3GPP TS 25.877, High Speed Downlink Packet Access (HSDPA) Iub/Iur Protocol Aspects.
[14] 3GPP TS 25.890, High Speed Downlink Packet Access (HSDPA); User Equipment (UE) Radio
Transmission and Reception (FDD).
[15] 3G Americas, The Evolution of UMTS 3GPP Release 5 and Beyond, June 2004.
[16] Stefan Parkvall, Eva Englund, Peter Malm, Tomas Hedberg, Magnus Persson, and Janne Peisa,
WCDMA Evolved High Speed Packet Data Services, Ericsson Review, No. 2, 2003.
[17]
WCDMA Evolved. The First Step HSDPA, Ericsson White Paper, May 2004.
[18]
[19] http://www.commsdesign.com/showArticle.jhtml?articleID=17300672
[20] http://www.tdscdma-forum.org/
[21] http://www.umtstdd.org
20
21
signal processing for satisfactory performance. Even then, copper communications wires are subject to
the effects of interfering electromagnetic sources such as radio, television, cell phone, and air traffic
control broadcasts.
As more bandwidth-intensive end-user applications continue to drive increased throughput requirements
for the various wireless access technologies, the spacing between the repeater points for copper-based
support systems must decrease in order to maintain the same aggregate data rate capacity. Contrast that to
fiber optic systems that can support ten Gigabits per second (Gbs) transmission rates over hundreds of
kilometers without the need for signal regeneration. Also, the diameter and weight of fiber optic cables
are much lower than comparable copper cables, resulting in lower associated materials, installation, and
maintenance costs for fiber optic systems.
2.1.1 Architectures
Hybrid fiber-wireless systems take many different forms depending on the environment and application.
Figure 1 illustrates the most common wireless networks in use today.
Notes:
(1) Base Station Hotel Remote Antennae Host Platform
(2) Telecommunications Enclosure
(3) Remote Sites for Distributed Antenna Network (DAN)
(4) Corporation Business
(5) User Premises
Internet Access
VLAN Telecommuter
(6) Mobile User
(7) Remote Tower Antenna
(8) Airport/Hotel Hotspots
All figures provided courtesy of Corning Cable Systems LLC and Corning Inc.
22
Ken Chauvin
23
limitations on effective coverage. Current large antennae structures are not economical (high initial and
recurring costs) or aesthetically pleasing, rising well above surrounding landscape. Such systems have
to account for the following:
By comparison, DAN antenna sites (e.g., telephone poles, light posts, buildings, etc.) are readily available
and accessible, and installation and maintenance issues are comparatively very minor. The next sections
describe the fiber optic components needed to support the various emerging wireless technologies.
24
Ken Chauvin
3. Passive Components
3.1 Optical Hardware
Optical hardware serves many purposes in a telecommunications system, primarily:
The hardware for the hybrid fiber-wireless system is best described by looking at the inside and outside
plants separately. For simplicity, the traditional outside plant long haul and metropolitan
telecommunications systems will not be discussed in detail as they are usually already in place, and the
wireless systems is simply connected at an appropriate location to the existing fiber optic plant.
3.1.1 Indoor Optical Hardware Wireless Local-Area Network (LAN)
The hardware components used in indoor systems are optimized for ease of system management and to
provide connection to the intrabuilding electronics serving the LAN. Indoor applications include
corporations, airports, hotels and the like where access to a private LAN (e.g., virtual LAN) or to the
Internet is needed for business or personal use. In indoor applications with an existing fiber infrastructure,
many of the network components may be in place and have sufficient capacity and space to handle the
additional needs of an intrabuilding wireless network. Fiber cables may be used to connect directly to the
antenna transceivers in the horizontal, but these wireless network access points may also be connected via
a copper cable run back to fiber fed electronics. See Figure 4.
25
connect that serves each area. The main and horizontal cross-connects are linked via the intrabuilding
backbone cable, which is also called the riser cable in multi-floor buildings.
The foundation of the indoor optical hardware plant is the connection housing as shown in Figure 5,
which provides the connectivity for the various system components. Depending on the system specifics,
several different types of housings may be used including; connector housings, splice housings, or
combination housings that can serve multiple functions. Accessory components such as jumper panels
and slack storage housings may also be appropriate. The hardware connection housing is the flexibility
point that provides for ease of system management.
(a)
26
Ken Chauvin
(b)
Figure 5: Indoor Fiber Optic Housings Rack and Wall Mount
Housings can be mounted to free-standing frames or wall mounted with accessory brackets. The key
considerations for ordering optical hardware are connector type, connector capacity, and dimensions for
mounting. Removable adapter panels located in the housings provide for the mating and alignment of the
27
connector pairs. Housings can also support: splice trays (separately, or in combination with connector
panels), splitters, couplers, wavelength division multiplexing (WDM) modules, and attenuators.
3.1.2 Outdoor Optical Hardware
The outside plant is, of course, much more diverse and severe than the climate-controlled indoor
environment. Products placed outdoors must be able to withstand the rigors of the environment, such as
solar radiation, temperature extremes, wind, rain, and snow. In some applications, other potential hazards
such as gnawing rodents or shotgun blasts must also be considered. The primary components for the
outside plant are fiber optic cables, optical closures, passive components, and cable assembles. Figure 6
shows common closures for outdoor applications.
(a)
28
Ken Chauvin
(b)
Figure 6: Outdoor Fiber Optic Closures
Recent fiber optic product advances for the outside plant include a whole series of products specially
designed to improve the speed and efficiency with which hybrid fiber-wireless systems are deployed.
Factory-terminated drop cables can be pulled through existing conduit, buried, or strung aerially and
plugged into closures (terminals) specifically designed to accept rugged, environmentally secure threaded
fasteners without even having to enter the closure itself. These special fiber optic splice terminal closures
can also be used to extend the reach of a DAN system by providing ready access back to a distant base
station, central office, or other metro fiber system access point. Overall, pre-connectorized solutions
greatly simplify the placement of fiber network and provide significant labor savings. Figure 7 shows two
pre-terminated outdoor terminal solutions and a companion pre-terminated cable.
(a)
29
(b)
(c)
Figure 7: Pre-Connectorized Fiber Optic Terminals and Cable
Note that when ordering pre-connectorized cables, it is important to ensure ordered lengths are correct to
prevent delays in system start-up, and to allow for the protection of the connector ends during installation
with fitted pulling bodies. Working with a reputable manufacturer or supplier with experience in
designing telecommunications products will ensure that the ordered product is appropriately and correctly
designed, built, and installed.
3.1.3 Fiber Optic Connectors
Fiber optic connectors are devices that are installed (terminated) upon fiber ends to allow for the
interconnection and re-mating of optical fibers. The ferrule of each connector houses the fiber and is the
specific part of the connector aligned in a connector pair. When two connectors are mated, the fiber end
faces are mechanically aligned and held in physical contact to achieve the requisite insertion loss and
reflectance performance. The connector also allows the fiber to be easily mated into the input and output
ports of transceivers. Figure 8 shows commonly used fiber optic connectors.
30
Ken Chauvin
31
32
Ken Chauvin
Finally, as described earlier, cables (indoor and outdoor) may be pre-connectorized on one or both ends,
which can greatly simplify the overall system installation. Customers employing factory-terminated
cables do so because they find value in the relatively lower installation times and associated labor cost
savings, as well as the guaranteed factory-polished performance. For direct termination in the field,
connectors can also be installed with an epoxy and polish procedure similar to that used in the factory. In
any event, the performance distribution of field-terminated products is not as good as products terminated
by factory-automated means, nor are the quality checks performed usually as detailed. Field installed
connectors can be installed successfully in a wide range of applications but may not be appropriate in all
cases. The success of such products also depends upon the system optical loss budget (or specified
acceptance criteria) and the experience of the installer.
A more advanced method of field termination utilizes no-epoxy, no-polish connectors. In these products, an
optical fiber stub is cured into place in the fiber ferrule and polished at the factory. For installation in the
field, the field fiber from the cable is matched up to the fiber stub inside the connector body and a
mechanical or fusion splice is made as appropriate. These connectors are typically very quick to install and
require few consumables. Although single- and dual-fiber no-epoxy, no-polish connectors are the most
common types used, 12-fiber quick-mounting connectors are also available for use on optical fiber ribbons.
3.1.4 Hardware and Connectivity Checklist
The following table is provided as a guide for the key fiber optic hardware components used in a hybrid
fiber-wireless system. The list is by no means inclusive and is intended only to provide wireless carriers
with talking points when discussing options with a fiber optics manufacturer or provider.
Product Selection Considerations
Mounting Hardware
Distribution Frame
Wall-Mount Brackets
Jumper Management
Dimensions (Height
and Width)
Housings
Rack or Wall Mount
Termination Capacity
Connector Type(s)
Splice Capacity
Adapter panels
Simplex
Duplex
Miscellaneous
Labels
Cable Straps
Cable Trays and Conduit
Jumper Panels
Jumpers
Simplex or Duplex
Length(s)
Connector Type
UPC or APC
Fire Performance
Connectors
Field Installed
Factory Pigtails
Pre-Connectorized Cables
Connector Type
Passives
Coupler-Splitters
WDM Modules
Splice Trays
Splice Protection
short in duration. Other external stresses, such as those experienced over the life of the installation due to
the surrounding environment, are more long term in nature. In addition to protecting the fiber, the cable
structure must also organize the fibers and support the identification of individual fibers via printed
legends, color coding, groupings, or some combination of the three.
Most environments will fall into one of three main categories that are generally referred to as indoor,
indoor/outdoor, and outdoor. For each environment there may be numerous other specific considerations
that are germane to a particular application, such as sewer or shipboard environments, or to an installation
method, such as aerial self-supporting or jetting into miniature (micro) ducts. As with the optical
hardware, there are a number of unique cable designs optimized for various combinations of applications
and installation methods. The performance criteria for the most common designs are also specified by the
various telecommunications industry standards forums.
In wireless access system applications, one of the biggest issues is the ability to connect to base stations
or to existing fiber optic systems in densely populated areas, where construction costs can run high and
the time needed to secure access rights can significantly delay system start-up. However, lengths of fiber
optic cable can be installed employing a number of different methods including aerially, trenched or
buried in the ground, pulled or blown into ducts, placed in narrow grooves cut into road surfaces, and
potentially even placed into water or sewer pipes. These varied placement options can reduce the number
of system components, thereby making the overall installation of optical fiber based telecommunications
systems more efficient. The relatively small size of fiber optic cable also saves on valuable conduit space,
especially when considering some emerging micro-cable designs that are specifically engineered for use
with air-blown, or air-assist, installation techniques into very small micro ducts only about one centimeter
in diameter.
Another advantage of optical fiber and fiber optic cable is the inherent flexibility in design, allowing for
the development of innovative products for specific applications. Since optical fiber is a man-made
composite glass structure, it can be custom designed to meet optimal cost/performance targets in any
number of specific applications. As it does not conduct electrical current and is not affected by
electromagnetic interference, fiber optic cable can be made all-dielectric, thereby making it the ultimate in
electromagnetically compatible transmission media for wireless communications support systems. This
eliminates such issues as the effects of spikes or surges from the cycling of electrical equipment, and
requirements for separate conduits for metallic conductors. It also improves the security of controlled
transmission systems as it is much more difficult to tap into a fiber line than with copper systems.
Even though the details of a particular cable design depend on the application environment and
installation method, there are a few generalities. Various applications expose the cable to different
mechanical and environmental conditions that have the potential to stress the fibers contained inside. As
optical fiber technology continues to mature, end-users are becoming more concerned about expected
fiber optic cable lifetimes, especially where early generation cables have already been in place for more
than 20 years. To date, no absolute performance model exists to accurately predict all potential failure
mechanisms for every application or environment, so the best gauge of the health of a system is typically
the optical properties of the cabled fiber.
The ultimate performance of cabled fiber depends on a number of factors, including the cable environment,
the type of installation, cable design, and the cable materials themselves. There are multiple factors that may
affect certain installations: structural cable damage caused during or after initial installation, improper cable
design, the quality of the splicing and closure work, exposure to periods of severe environmental loading or
certain chemicals, or exposure to temperature and humidity conditions outside of specified operating
performance ranges. Evidence of performance degradation can be elusive and is largely dependent on the
specific characteristics of each installation and the type of cable design deployed.
34
Ken Chauvin
installation along high-tension lines), or resistance to various chemicals or other organic hazards such as
gnawing rodents.
As fiber connections get closer to the remote DAN antennae sites, cable designs are changing to adapt to
this specific environment. Cable fiber counts for such applications are typically very low (two to four
fibers) and the distances involved are much shorter than for traditional outside plant cable installations
(hundreds of feet versus of hundreds of miles). Cables for this environment are usually termed drop
cables. Although the installation methods are generally the same as other outdoor cables, the placement
of drop cables often involves the use of manual labor to install, obviating the need for heavy equipment
and high tensile loads.
These cables can either be buried from the curb to the antenna electronics, pulled or blown into conduit,
or can be hung aerially from pole to pole and directly secured to the wireless electronics housing. This
environment allows for innovative cable designs and materials selection that can support lower cost
product solutions, which is critical in enabling the deployment of affordable cost-effective DAN systems.
3.2.1.3 Indoor/Outdoor
Indoor/outdoor applications involve cables which typically run short distances to make interconnections
within and between adjacent buildings. These cables are fully resistant to the typical outside plant
environment, but the overall performance requirements are typically not as stringent as those for longlength outside plant communications cables. These cables are used primarily to avoid the need for a costly
extra splice point where the telecommunication system cabling transitions from outside to inside, and thus
are employed in only relatively short lengths.
3.2.2 Installation Methods
There are a number of diverse installation methods for the various spaces in which fiber optic cable is
used. The installation methods discussed below are primarily associated with outside plant applications.
The relatively benign environment of the inside plant means that installation considerations are somewhat
less critical, although that is not to say such are unimportant.
3.2.2.1 Direct Buried
One of the most common and widely used installation methods is the placement of the cable directly into
the ground, which is known as direct burial. The traditional methods of direct burial are trenching and
plowing. Newer installation methods designed specifically for the types of crowded metropolitan areas
familiar to wireless providers include placing cable into narrow grooves cut into hard surfaces, such as
roadways or parking lots (also known as road cable). The obvious advantage to these types of installations
(i.e., buried) is the inherent measure of safety and stability provided by the cable surroundings if left
undisturbed. Buried cables are subject to damage from dig-ups due to subsequent construction, lightning
strikes, chemicals in the soil, gnawing rodents, and ground heaves. Protection against these dangers is
somewhat accounted for in the design of the cables used, which typically includes the incorporation of a
layer of protective metal armoring for most direct-buried applications without a duct.
For cables designed specifically for placement in road surfaces, such as asphalt or concrete, the process
for installation involves placing the cable in a narrow groove (about one inch wide) ground into the hard
surface of a road or parking lot. The cable may be secured in place with special hold-down spacers
pressed into the groove on top of the cable. The remainder of the groove can then be filled with bitumen
a common road repair materialto restore the road surface (see Figure 10). Since the process can be
done as a continuous operation, it minimizes the disruption to traffic flow. This method also boasts a
minimal impact to the useful life of the roadway because the groove is very narrow. Conversely,
traditional roadway trenching methods are comparatively more disruptive, costly, and also can
significantly reduce the life expectancy of the roadway surface, not to mention the repaired sites can be
36
Ken Chauvin
aesthetically unappealing. Many local governments have strict regulations covering roadway access, but
the groove method used with special road cable can obviate some of the more severe requirements
associated with obtaining installation permits.
3-to5-ingroove dept
h
(surface dependan
t)
Bitumen
Rubber Hold-Do
wn
Foam Spacer
MCS Road able
C
37
cable jacket to drag the cable along. These air-assisted installation techniques, sometimes referred to as
jetting, may also involve pushing the cable with a tractor mechanism or pulling with tape while blowing
compressed gases into a pre-installed duct or miniature duct at a high speed. This technique allows the
cable to essentially float inside the duct during installation, minimizing the forces placed on the cable
due to friction between the cable jacket and the duct wall. Because of the limited stress placed on cables
when utilizing this installation method cables can be much smaller due to the lower tensile strength
requirements, thereby saving valuable duct space.
3.2.2.3 Aerial Installations
The common methods for placing cables aerially include lashing the cable to a pre-installed steel
messenger wire or employing the use of cable specifically designed to be self-supporting. Self-supporting
fiber optic cable designs are usually based on one of two primary designs. Figure 8 self-supporting aerial
cables consist of a fiber optic cable core and integrated messenger, typically made from steel. Both the
cable and the messenger share a common outer jacket resulting in the characteristic shape that gives these
cables their name. All-dielectric, self-supporting (ADSS) aerial cables contain no metallic elements and
are usually round in shape, although many smaller drop cables have a flattened shape to support the use if
inexpensive wedge clamp (also known as P-clamp) supporting hardware.
Aerial applications for fiber optic cable offer unique challenges. Planning for aerial cable installation for
wireless networks must include taking into account proper clearances over hazards, the cable types and
physical properties, and the mechanical stress loading placed on the cable over the life of the installation
due to the environment. Proper planning requires knowing the sag and tension characteristics of the cable
once installed. Sag is the amount of vertical distance the cable will hang, or droop, from an imaginary line
drawn between successive attachment points. The tension is the force the cable will experience under
various installation and environmental conditions at the attachment points. Understanding the stresses
placed on a cable installed aerially is important to ensure that safe operating conditions are not exceeded,
cable is not lifetime shortened, or that dangers from material failures do not occur.
There are also a number of other potential hazards that must be accounted for when installing cables
aerially because of the exposure. These include shotgun damage, lighting strikes, gnawing rodents, and
even the effects of high electric field potentials in installations where the cable is located near hightension power lines. Dielectric cables can minimize the effects of lightning or high field potentials, but do
not render such cables impervious to their effects. Armoring can improve the survivability of cables to
mechanical damage from gunshot and rodents, but adds considerable weight, which can limit achievable
span distances.
3.2.3 Cable Types
Combining the various attributes discussed to this point enables the construction of a wide range of cable
types for various applications. The most common cable designs are described in the following sections,
and the design of choice for a particular application depends on a number of factors. The primary
considerations are fiber count, installation method, and the environment. Specific designs based on the
basic constructions discussed below (i.e., stranded loose-tube, central tube, and tight buffered), but which
are specialized for a particular application, are discussed in Section 3.2.3.1 through 3.2.3.3. Common
fiber packing options for the primary cable design families are shown in Figure 11.
38
Ken Chauvin
Ribbon
Matrix
Optical
Fibers
(a)
900 m
Buffer
TB2 Coated
250 m Coated
Optic Fiber
(b)
Optical Fiber
(250 m Coated,
Colored)
BL
WH
GR
SL
Filling
Compound
OR
BR
Buffer
Tube
(c)
39
Stranded (buffer tube) loose tube cables are the most common design of fiber optic cable and were also
the first ones in widespread use. The primary benefits of these designs are the large operating temperature
window, the ability to access fiber within a short length of cable (mid-span), and the relatively small size
at moderate fiber counts when individual colored fibers are used. Conversely, the main benefit of singletube designs is that they can have the smallest outer diameter size for a given fiber count. At counts above
12, the fibers in single (or central) cable designs must be grouped into ribbons or bundled together with
colored threads to support the identification of individual fibers. See Figure 12 for common outdoor
stranded and central tube cable designs.
OptionalRPE
T Outer Jacket
Ripcor
d
Dielectric
Stren
gth M em bers
W ater-Sw ella
ble ape
T
Buff
er ube
T
Fibers
W ater-Sw ella
bleYarn
(a)
PE Outer Jacket
W ater-Sw ell
able Fiberglass
Dielectr
ic Stre
ngth M emb er
Buff
erTube
Fibers
(b)
40
Ken Chauvin
PE Jacket
Stranded Steel
M essenger
PE Outer Jacket
Ripcord
Dielectric Strength M em bers
(as required)
W ater-Sw ell
able Tape
Fill
ed Buff
erTube
Fib
ers
W ater-Sw ell
able Yarn
Dielectric CentralM em ber
(c)
PE Jacket
SteelM essenger
PE Jacket
Buff
erTube
Fibers(1-12
fibers)
(d)
Figure 12: Common Outdoor Cable Examples
41
Tight-buffered cables, shown in Figure 13, were the first generation of premises cables and are still the
most commonly used due to the ease of connectorizing and handling such. The name tight buffered is
derived from the layer of thermoplastic or elastomeric material that is applied directly over the fiber
coating. The buffer coating on the fiber makes it easier to handle and supports the direct termination with
fiber optic connectors in the field, without the use of furcation kits.
Fiber (250 m)
Thermoplastic
Buffer
(900 m)
PVC Outer
Sheath
Dielectric Strength
Member
(a)
PVC Outer
Sheath
Dielectric Strength
Member (Aramid Yarn)
Fiber
(250 m)
Thermoplastic
Buffer
(900 m)
(b)
42
Ken Chauvin
Ripcord
Dielectric Strength Member
TBII Buffered Fibers
Dielectric Strength Member
Dielectric Central Member
(c)
(d)
43
Outer Jacket
Dielectr
ic Stre
ngth M emb er
12-FiberRibbon
(e)
Figure 13: Standard Indoor Fiber Optic Cables
A number of specific cable designs, based on the basic constructions listed above, are also available for
certain specific applications. Some of the most common types are described below along with their
potential uses.
3.2.3.1 Self-Supporting Aerial Drop Cables
Standard terrestrial cables, shown in Figure 12, are often used for aerial applications, but in such cases
they must be lashed to an existing aerial support messenger in the field, which increases installation time
and cost. Conversely, aerial self-supporting cables, as seen in Figure 14, can be installed much more
quickly and efficiently since the tensile bearing elements are integral to the cable design itself. Selfsupporting cable types are usually a variation of the loose tube design containing individual colored
fibers, though other variations such as ribbon designs may see limited use.
44
Ken Chauvin
PE Outer Jacket
Ripcord
Dielectr
ic Stre
ngth M em bers
(as re
quired)
W ater-Sw ell
able Tape
Fil
led Buff
erTube
Fib
ers
W ater-Sw ell
able Yarn
Dielectr
ic Centra
lM em ber
(a)
PE Outer Jacket
Ripcord
Strength Members
Water-Swellable Tape
Dielectric Central Member
Filled Buffer Tube
Colored 250 m Optical Fibers
Water-Swellable Yarn
(b)
45
PE Outer Jacket
Corru
gated SteelTape Arm or
Dielectr
ic Stre
ngth M em bers
W ater-Swell
able Tape
W aterSwell
able
Tape
Ripcord
Fill
erRods
W ater-Swell
able Yarns
Buff
erTubes
FiberRibbons
Dielectr
ic
Centra
lM em ber
(c)
PE O uterJacket
Corrugated Steel
Tape Arm or
Ripcords
Steel Stre
ngth M em bers
W ater-Sw ella
bleTape
Filled
Buff
erTube
O pticalFiberRibbons
(d)
Figure 14: Aerial Self-Supporting Loose Tube Cables
46
Ken Chauvin
PE Outer Jacket
Dielectr
ic Stre
ngth M em bers
(>12Fibers)
Buff
erTube
Color-Coded BinderThread
FiberBundles
PE Outer Jacket
Dielectr
ic Stre
ngth M em bers (>12Fibers)
Buff
erTube
FiberBundles
Color-Coded BinderThread
Fillin
g Com pound
47
utilizing a central metallic tube to house the optical fibers. The tube is then jacketed with standard jacket
materials to complete the cable.
Outer Jacket
CopperCentra
lTube
Sheathing Options
Fiber Packaging
Fiber Type
Miscellaneous
Comments
Indoor
Outdoor
Direct Buried
Conduit Pulled
Conduit Air Blown
Aerial self-Support
Aerial Lashed
Other
Jacket Material
Armors
Loose
Ribbon
Tight Buffer
Single-mode
Multimode
Packaging
Test Equipment
Aerial Span Distances
Weight and Dimensions
48
Ken Chauvin
Coating
Single-mode
Cladding
Core
Multimode
Coating
Cladding
Figure 17: Optical Fiber Profiles
When designing and specifying a wireless telecommunications system the key factors to consider are the
transmissions speeds (i.e., bandwidth requirements), operating distances and wavelength(s), and system
installed costs, all of which play a role in choosing between the available options. It is also critical to
consider the flexibility and adaptability of the system, and thus its ability to support upgrades with
minimal expense (i.e., future-proofing). Conversely, the expansion of an existing infrastructure requires
that newer products also be backwards-compatible. With respect to what type of fiber to select, it is
usually easy to decide between multimode, which is typically used for intrabuilding networks, and singlemode, which is primarily used in the outside plant. The following sections provide general information
which can be used to better understand the primary types of fibers used today.
3.3.1 Single-Mode Fibers
Single-mode fibers have historically operated in one of the primary wavelengths regions around either
1310 or 1550 nm, although applications that take advantage of the extended operating spectrum in
between are already commercially available. The introduction of such equipment was slowed somewhat
early on in the new millennia by the burst of the telecom bubble which began at the end of 2001.
However, newer wireless architectures employing various wave division multiplexing schemes are
emerging to take advantage of this available space.
49
The strength of single-mode fibers lies in their inherently low attenuation and dispersion characteristics at
the primary operating wavelengths. As a result, single-mode fibers are used in applications to transmit
high data rates over distances from tens to thousands of kilometers, where the spacing between costly
signal conditioning and amplification equipment needs to be maximized. Single-mode fibers can be
generally categorized into two different types that are best differentiated optically by their dispersion
characteristics, and physically by the shape of their refractive index profiles as seen in Figure 18. Note
that under visual inspection, or without the benefit of complex test equipment, it is impossible to tell the
difference between any of the fiber types discussed herein, were they to be laid side by side.
Coating
Cladding
Core
N
D
Step Index Core
D
Segmented Core
50
Ken Chauvin
In the early to mid-1990s, a better understanding of fiber non-linear effects resulted in the development of
non-zero dispersion-shifted fibers. These are fibers with segmented core profiles, but for which the zero
dispersion wavelength is shifted such that it falls outside of the intended operating window, or windows.
The amount of dispersion is controlled by managing the zero dispersion wavelength and characteristic
dispersion slope through the careful manipulation of the fibers complex index profile. Although some
amount of dispersion (positive or negative) is desired to minimize non-linear effects, the effects cannot be
eliminated entirely. Moreover, too much dispersion is not desirable either as it results in excessive pulse
spreading, which can limit achievable distances or data rates. The latest generation fibers are also
designed with a larger mode-field diameter to lower the transmitted power density, and thus further
reduce non-linear effects.
Table 3 summarizes some of the parameters for the various single-mode fiber types commonly used today
(typical values for cabled fiber are shown).
Non
Dispersion-Shifted Non-Zero Dispersion-Shifted
Single-Mode
Single-Mode
Parameter
Mode Field
Diameter
8.6 9.5 m
9.6 11.2 m1
8 11 m2
Numerical Aperture
0.14 (typical)
Cutoff Wavelength
< 1260 nm
< 1480 nm
Cladding Diameter
125 .0 1.0 m
125 .0 1.0 m
Coating Diameter
245.0 10.0 m
245.0 10.0 m
Attenuation
1310 nm
1550 nm
NA
< 0.3 dB/km
Dispersion
O-band
C-band
NA
0.1 - 6 ps/nmkm 2
1.
2.
1310 nm
1550 nm
51
center, which then falls off approximately in a parabolic shape to where the value is that of the pure silica
glass cladding at the core-clad interface. The primary core sizes for graded index multimode fibers in use
today are 50 and 62.5 m. Other larger-core multimode fibers saw limited use early on in
telecommunications applications, but such are rare today and current large core multimode fiber designs
find service primarily in very-short-length, low-data-rate applications such as automotive, medical, and
similar uses involving imaging or low-intensity lighting.
50/125
62.5/125m
Core Size
50 3.0 m
62.5 3.0 m
Numerical Aperture
0.200 0.015
0.275 0.015
Cladding Diameter
125.0 2.0 m
125.0 2.0 m
Coating Diameter
245.0 10.0 m
245.0 10.0 m
Attenuation
850 nm
1300 nm
OFL BW
850 nm
1300 nm
EMB
850 nm
1.
2.
NA
New 50/125 m fibers designed and specified primarily for use in 10 Gbps serial systems laser-based
systems, but can also be used for legacy applications
Minimum bandwidth values vary from application to application.
52
Ken Chauvin
Summary
As the wireless access industry grows and matures it will continue to reach further into our everyday lives
whether at work, school, home, or riding in a car, and bringing with it a wider variety of services and
improved quality of life. For outdoor applications specifically, distributed antennae networks are an
attractive technology that benefit greatly from the reach and bandwidth capability of fiber optics. In order
to properly design, specify, order, install and grow a hybrid optical-wireless system, wireless carriers
should work with reputable and knowledgeable industry fiber experts to ensure their systems are viable
and efficient to maintain and upgrade for current and emerging applications.
References
Hecht, Jeff, Understanding Fiber Optics, Fourth Edition, Prentice Hall, Columbus Ohio, 2002.
Commercial
Building
Telecommunications
Cabling
Telecommunication Industry Association, 2001.
Standard,
ANSI/TIA/EIA-568-B,
EIA Standard Fiber Optic Terminology, EIA-440-A, Electronic Industries Association, 1989.
Optical Fibres Part 210: Product Specifications Sectional Specification for Category A1 Multimode
Fibres, International Electrotechnical Commission, 2002.
Optical Fibres Part 250: Product Specifications Sectional Specification for Class B Single-Mode
Fibres, International Electrotechnical Commission, 2002.
Recommendation G.651, Characteristics of a 50/125 m multimode graded index optical fiber cable,
International Telecommunications Union, 1998.
Recommendation G.652, Characteristics of a single-mode optical fiber cable, International
Telecommunications Union, 2000.
Recommendation G.653, Characteristics of a dispersion-shifted single-mode optical fiber cable,
International Telecommunications Union, 2000.
Recommendation G.655, Characteristics of a non-zero dispersion-shifted single-mode optical fiber cable,
International Telecommunications Union, 2000.
53
Trent Tishkowski
Manager, National Advisory Services
Ernst & Young LLP
Wireless technologies have become an alphabet soup that is extremely difficult for consumers and business
end-users to navigate. Even in the much more mature wireline broadband marketplace, many customers are
not fully aware of the fundamental differences between various flavors of Internet service, let alone the
meaning of acronyms like DSL. Imagine, then, the confusion a typical wireless user confronts when hit with
terms like wideband code division multiple access (W-CDMA), general packet radio service (GPRS), wide
integrated digital enhanced network (WiDEN), and a host of other mysterious codes.
As Figure 1 illustrates, this alphabet soup can be broadly classified into four major buckets of second
generation (2G), 2.5G, third generation (3G), and fourth generation (4G), each aligned with the applications
that are enabled by their respective download speeds. To date wireless applications have been limited by
download speeds to basic, but highly utilitarian, applications such as text messaging, and the slightly more
advanced applications offered by services such as multimedia messaging service (MMS).
Technologies
Avg. Download
Applications
4G
4G
Flarion
WiMAX
3G
3G
CDMA 2000
UMTS/W-CDMA
~380-400 Kbps
Video (low-quality)
Web browsing (brisk)
Medium file downloads (e.g. MP3)
2.5G
2.5G
CDMA/1x
GSM/EDGE
GSM/GPRS
WiDEN
~35-115 Kbps
2G
2G
CDMA
iDEN
GSM
TDMA
~10-20 Kbps
Voice
SMS (e.g. text messaging)
55
3G technologies will deliver much faster downloads, but still fall short of the much more rapid speeds
offered by wireline broadband services such as cable Internet or fiber-to-the premises (FTTP). This
performance gap has fueled the idea of a two-line world, wireless for voice and narrowband data needs,
and wireline for broadband Internet and video services.
4G wireless technologies hold the potential to overturn the two-line hypothesis. These emerging
technologies may potentially deliver end-user download speeds comparable to those offered by cable
modem service and faster than the DSL services that have been broadly deployed by the fixed-line
telephone incumbents. More importantly, these wireless technologies may require considerably less in
capital expenditures than wireline broadband.
As Figure 2 illustrates, 4G technologies such as worldwide interoperability for microwave access
(WiMAX) and Flarion are positioning to occupy a new space in wireless access networkingthe
metropolitan area network (MAN). These MAN offerings could fundamentally change the
telecommunications business model, presenting a new challenge to broadband wireline infrastructure.
802.16a enables
non-line of sight
transmission from
transmitter to
multiple users at
~1 Mbps average
speeds -- targeted
as a fixed wireless
solution rather than
a mobile
technology
802.16e, under
development,
enables portable
usage
Enables data
connections between
electronic devices in
the unlicensed 2.4
GHz spectrum band
Designed to replace
cable or infrared
connections for short
distances (~5-30 feet)
Metro Area
Network
(MAN)
802.20
(Flarion)
802.16
(WiMAX)
Local Area
Network
(LAN)
Specifically designed
for high-speed mobile
usage
End-user data
throughput speeds are
expected to be
comparable to
DSL/Cable modems
(~1.5+ Mbps)
Flarions solution
utilizes an all-IP
architecture for both
voice and data
802.11
(WiFi)
Personal
Area
Network
(PAN)
802.15
(Bluetooth)
56
The scope of the market disruption created by an economic wireless MAN solution would not be limited
to the fixed-line telephone incumbents; all wireline and wireless incumbent service providers could face
dramatic impacts. As Figure 3 summarizes, the nature of the impact to incumbents is driven by two
fundamental variables: the download speed and mobility of a next-generation wireless offering. Scenario
1 achieves high mobility, but with the trade off of slower speed; scenario 2 has higher speed, but
sacrifices true mobility (while still supporting portable applications).
SCENARIO 2
~2-5Mbps sustained
average data speed
Portability, but limited
mobility (similar to
WiFi -- ability for endusers to walk
around)
Traditional
Wireline Telecom
Traditional
2.5G/3G Wireless
Modest impact
Leapfrog technology
provides superior data
transfer speeds at
equivalent (or superior?)
cost economics
High impact
High impact
Modest impact
VoIP deployments
negatively impact access
revenues and other enduser call settlement
charges
57
wireless operators might face only a modest impact in this scenario, since the new technology wouldnt
compete against truly mobile applications.
These potential impacts must be tempered against the fundamental questions that remain regarding the
marketplace viability of next-generation wireless technologies. A renewed--and healthy--focus on return
on invested capital (ROIC), rather than revenue growth at any cost, has forced wireless carriers to cut
through the technology hype and determine whether these emerging technologies can be implemented to
serve a sufficiently broad set of paying users to generate positive economic returns. Definitive analyses
required to answer this question have not yet occurred in the following five key areas detailed in Table 1:
economics, mobility, performance, scalability, and regulation.
Economics
Mobility
Performance
Scalability
Regulation
The views expressed herein are those of the authors and do not necessarily reflect the views of
Ernst & Young LLP.
58
tML (telecom markup language for TMN) starts with a lower case to distinguish it from the TML (telephony
markup language).
59
Integration of Wireless Access with Wireline Networks: OAM&P Support Architecture with ITU-tML Technology
an innovative management architecture framework that adopts the latest technical standards to support
telecom network and services management across wireline next-generation network as well as broadband
wireless access.
2. TMN Evolution
TMN [8] was originally developed due to the emerging of packet network technologies, as a supplement
to the transport circuit technologies. The goal was to provide a standard OAM&P infrastructure with
benefit of flexibility for growth and changes.
The original TMN data network, however, was only limited to a cumbersome OSI framework to provide
the necessary application layer service. As technologies progressed, new architecture framework
eventually replaced the OSIbased TMN architecture while reusing much of its management capabilities.
2.1 OSIBased TMN
The OSIbased telecom management architecture included the structure of management information
organized as a management information base (MIB). To specify the MIB objects and operations on them,
it supplied the guideline for definition of managed objects (GDMO) and abstract syntax notation number
1 (ASN.1). Only five operations (create, delete, get, set, and report/notify) were supplied utilizing the OSI
remote operation service between a manager and an agent, although any operations maybe emulated by a
combination of those primitives. There are techniques to deal with large-scale operations and efficient
information processing such as scooping and filtering functions.
The OSIbased common management information protocol (CMIP) and management framework
architecture has been successful in providing yesterday and todays network management solutions and
some service-management capacities. The latest application of the general definition of managed objects
(GDMO)/ASN.1/CMIP to the major telecommunications industry included the PIC model, which allows
the freedom of switching long-distance carriers [2]; and the local number portability (LNP) model, which
allows change of wireless carriers and wireline providers while retaining the same telephone number.
Due to new technology advancements, the telecom industry realized that the OSI/CMIP TMN framework
is no longer the best technologies in terms of cost, efficiency and availability of skilled resources. After
last implementation of CMIPbased PIC and LNP models, the TMN applications are no longer relying on
GDMO technologies.
The key alternatives are CORBAbased [11] or XMLbased TMN framework to be summarized in the
following subsections. While SNMP [7] enjoys successful applications in Internet network managements,
SNMP based management has not been approved in the telecommunications services management arena.
Thus, this paper intentionally excludes the SNMP as a major telecom management framework.
2.2 CORBABased TMN
When the Object Management Group (OMG) completed the CORBA software architecture [11] that enables
the distributed software to interact regardless of location and implementation, the telecommunications industry
took notice and eventually recommendations were made at the industry level to start adopting the CORBA
technology as an alternative to the CMIP technology.
Following the ITU-T tradition to provide a uniform approach to define various TMN application
interfaces, the standard organization led by the efforts of the committee T1M1 took the initiative and
developed the framework and IDL models [10]. A generic (TMN) managed object was defined in IDL
with native CORBA network management interfaces. While native CORBA services (such as
notification) were used whenever possible, basic TMN IDL models reflected much of the traditional
60
61
Integration of Wireless Access with Wireline Networks: OAM&P Support Architecture with ITU-tML Technology
Fault Management
Trouble resolution, testing, and tracking are important functions performed by fault management. Both
internally and externally generated troubles are handled in the fault management functional area.
Accounting Management
The accounting management functional area associates event data with the appropriate billing and/or
accounting system.
Performance Management
Performance management monitors and logs the performance of the entire network architecture. This
functional area helps ensure that the network and systems in place are performing at an expected level and
meet performance requirements.
Security Management
Security functions deal with access control, resource authorization, and implementation of security
policies from the network-element level to service-request level.
The above management functions are essentially important to the OAM&P for network-access and
service-management capabilities. The key management functions in broadband wireless access include
the following:
Preordering functions to determine the wireless local loop status or status of access points
Ordering and configuration of the wireless access as part of the tall service ordering
Provisioning and status updates in both access and PVC or SVC end to end setup and OSS
updates
Controlling and coordinating of the service view across the (wireless) access domain and the
(wireline) transport domain
Monitoring and assurances that a specific service class is performing according to specified
requirements
Providing network diagnostic and performance capabilities to the access network or even to the
end-user customers premises in a customer service domain
Managing end-to-end network views of each node and allowing integration wireless access
services in operation centers
Provisioning and modifying of network capabilities for the customer network management
Maintaining interface configuration and dynamic reconfiguration capabilities in a customer
network management environment
Maintaining statistics and history pertaining to sessions or connections across domains
The OAM&P functions are essential for broadband-wireless and broadband-wireline integrations. It is
imperative that a new unified architecture framework can provide the necessary management functions
while allowing integration of different types of network technologies.
4. New Architecture Framework and Technology
Figure 1 illustrates an innovative architecture for OAM&P functions in management of wireless
broadband access with integration to wireline-management capabilities.
62
Broadband
Wireless
Access
Fixed
Wireless
Access
NETWORK
MIB
MIB
MIB
Management
center
gateway
gateway
tML CORE
VoIP
Business Rules
Rules, Objectives,
Guidelines for tML
SLA
Monitoring
tML Schema
Messaging
Security
tML Document
Infrastructure Profile
CS
M
New
Services
ITU-tML Framework
CN
Multimedia
Self-Setup
middleware
SelfManagement
Figure 1
4.1 Standard-Based tML CORE
At the core of this architecture lays the fundamental model for standardized TMN in a next-generation
environment. Access business rules are to be defined with international standard vocabulary. Service
agreements will conform to standard tML language and associated management processes.
The ITU-tML will be the benchmarks and framework for guidelines, objectives, and rules. The
architecture framework for wireless and wireline integration extends the wireline-based tML rules [9] to
include wireless local loop and mobile broadband access networks.
A minimum set of infrastructure requirements will ensure the interoperability in messaging and encoding.
The infrastructure requirements also ensure security with universal and uniform access. Furthermore, they
are important for guaranteed reliability and performance of the user services such as voice over Internet
protocol (VoIP) and multimedia information-exchange applications. The implementation profiles and
rules of conformance provide a uniform approach to enable future integration of wireless access with
wireline networks.
As an example, consider the IEEE802.16defined access interface [6] and an asynchronous transfer mode
(ATM) based core network [5] being supported by service providers. The key 802.16 architecture
includes the air interface for 10-to-66-GHz devices, the air interface for licensed systems below 11GHz
and coexistence of broadband wireless access. The ATM architecture consists of the user-network
interface (UNI) and network-network interface (NNI). ATM also provides for cell relays with
corresponding transport encapsulation functions.
63
Integration of Wireless Access with Wireline Networks: OAM&P Support Architecture with ITU-tML Technology
The IEEE 802 management group defines MAC layer MIBs, primitives and actions. The 802.16 protocols
also define MAClevel support for service capabilities such as QoS by being able to allocate channel
capacity to satisfy service demands. At the service management layer, however, there is no standard MIB
or OAM&P functions spelled out for broadband access standards; and eventually, a network- and servicemanagement layer management functions and management information have to be defined and
standardized. The approach can be similar to the Third-Generation Partnership Project (3GPP)
management solutions sets TS-32-101 to 111 specifications [1]. Furthermore, tML framework is followed
when defining those fault management, configuration, and other management functions.
The ATM Forum has completed the management models (for wireline telecom transport) in ATM Forum
M4 Interface. They contain the MIBs and management interface functions in CORBA IDL. In the
example being considered above, the UNI management model can be extended to include the broadband
wireless access management capability. The extension may be converted to the core tML models together
with the access management models, so that all can be incorporated or integrated for uniform processing
within this architecture.
With both the access management and ATM transport management converted to a uniform tML model,
the core management rules now can support single schema of service providers networks view to include
both wireless and wireline subnetworks. The uniform tML model will also implement simple object
access protocol (SOAP) to determine equipment status and trouble-ticket diagnostics; it will facilitate
network configuration, maintenance and monitoring network partitioning; and it will allow service
providers to resell bandwidth and services and allow a private network owner to share control of the
network functions either within a single wireline/wireless component domain or between different
network and service domains. Service providers can partition the network according to technology type
(i.e., wireless, IP, ATM) so that they can differentiate their OSS interfaces while maintaining a single
external management view.
4.2. Customer Network/Service Managements
Well-defined function blocks and interfaces unite diverse underlying networks and provide uniform
interfaces for customer network management (CNM) and customer service management (CSM).
CSM and CNM enable self-setup and proactive service-level monitoring. Additional self-management
functions include trouble ticketing, repair status, usage collection, and dynamic reroute or reconfiguration.
The CSM/CNM capabilities are derived from the TMN base functions. The TMN functional blocks cover
trouble ticketing and fault resolution, configuration management, billing and accounts, performance
threshold and reporting, as well as security management. Those TMN OAM&P functions were described
in the previous section.
By following the core support architecture, CNM and CSM solutions are also compliant to the tML
framework rules and guidelines. The benefits of compliance are interoperability across the wireless and
wireline domains; productivity improvement in reusing the industry practices, as well as data and
software portability in adopting XML.
In order to support various network technologies across multiple domains, a number of (tMLenabler)
middleware and gateways will provide the interworking between the management entities. For
examples, a self-provision service will interface with the wireless access local manager to setup the
required channel capacity while it interfaces to negotiate the traffic rate for ATM admission control.
Provision status shall be in a total view to include the edge as well as the transport for both the CNM
user and the management center.
64
The management center is mainly for integration of views from network operator and service providers,
while CSM/CNM allows users to configure, provision, and monitor their portion of access and
networking resources.
The entire internal management center and the external customer network and service management share
and reuse the tML core management information. Data consistency and security are preserved by
operational policies as well as system supports.
5. Conclusions and Future Research
This paper considered the management capabilities to integrate wireless access to the core wireline
network. We summarized the evolution of TMN technologies from OSIbased GDMO/CMIP
framework, to the CORBAbased IDL framework, to XMLbased tML framework. Following the
technology trends and the needs of the industry, our research produced an innovative architecture for
telecom management applicable to networks in the wireless, wireline as well as integration of both
wireless access and wireline network.
The core of the architecture adopted the latest tML standards from the ITU while the overall applications
beyond the core could use other technology for transformation purpose. The new management
architecture model could supply OAM&P supporting functions for data/voice/video and multimedia
services. The presented architecture framework has incorporated the new notions of CNM customer
network management and CSM customer-service management.
Future research will incorporate traditionally nontelecom management domains (especially the SNMP
based IP network management). Additional development efforts are also required in order to provide a
complete set of service models as well as performance testing tools.
65
Integration of Wireless Access with Wireline Networks: OAM&P Support Architecture with ITU-tML Technology
6. References
[1]
[2]
ANSI, T1.246 Standard, W. Liu (technical editor), OAM&P Model for Preferred Inter-Exchange
Carrier (PIC) CARE, a GDMO/CMIP Model, 1999.
[3]
ANSI, T1.256 Standard, W. Liu (technical editor), OAM&P Model for Electronic Access Ordering
(EAO) Inquiry Functions, a CORBA/IDL Model, 20002001.
[4]
ATIS, www.atis.org
[5]
ATM Forum, AF-NM-002, 0058 & 0185: ATM Forum M4 Interface Requirements, 19942002.
[6]
[7]
[8]
ITU M.3010 Standard TMN Principles, a part of TMN Standard Series, 2000.
[9]
[10] ITU X.780 Standard, TMN Guidelines for Definition of CORBA Managed Objects, 2001.
[11] OMG, www.omg.org
[12] W3C, www.w3c.org
66
67
Broadband Ecosystem
Vendors
Providers
End Users
Sector
BBW
Focus
BBW
Rev.
Components
No
Low
Equipment
Yes
Low
Systems
No
Low
Applications
No
Low
Service
Yes
Low
Business
No
Low
Residences
No
Low
Emergence of killer applications. Applications like mobility can draw end users away from
DSL and cable modems. Most successful telecom technologies only experienced widespread
end-user penetration after a common end-user value proposition was created. This requires
standardization of the product offering across markets and service providers and greater
dialogue among industry players.
Entry of the major players. The entry or creation of major operators will act as the catalyst for
established vendors, system, and application providers to enter this technology. For this to
happen, the service capabilities have to be core to the operators value proposition. Unfortunately,
non-3G broadband wireless technologies do not integrate well with the incumbentsregional
Bell operating companies (RBOCs), Post Telephone and Telegraph Administration (PTTs), and
mobile operatorseither strategically or operationally. This requires the creation of a new class
of carriers that may include Internet service providers (ISPs) and satellite providers. Rural
markets and developing countries do not offer a large enough scale to attract significant
investments and create a sustainable broadband wireless ecosphere.
Because of these issues, the best strategy for any broadband wireless operator may be to get in early to
build technical, operating, and market competencies, but delay full deployment until a technology and
Motorolas advanced mobile phone service (AMPS) platform and Qualcomms code division multiple access
(CDMA) technology are examples of this.
68
R.J. Mahadev
market direction appears. This creates the need for a phased deployment that balances the benefits of
early-market entry with the risk of making the wrong decisions.
Typical Deployment Milestones
A clear definition of the deployment phases will allow management to optimize resource allocation
and minimize external funding requirements. The table below contains an example of suggested
deployment phases.
Pre Formation
(6 months)
Key
Milestones
Other
Activities
Resources
Funding
Form company
Procure licenses
Finalize business
plan
Work on business
case and operating
plan
Identify merger/
acquisition partners
Finalize
compensation and
ownership plans
Core executive team
(three to five
resources)
External advisors and
consultants
~$500K in expenses
Necessary bank
guarantees
Phase III
Achieve positive
cash flow and
continued
performing
according to plan
Sign up initial
customers
Acquire equipment,
backhaul, and other
key n/w components
Build customer and
network operations
Implement
operating plan to
support businessas-usual operations
Expand into
additional markets
as desired
Full management
team and operations
team
Changes in
ownership and
management based
on exit event
25% of overall
funding
Remainder of
external funding
Phases I and II represent the most critical stages for the company. The Phase I proof of concept will
enable management to validate the technology and the market proposition. However, it is during Phase II
that the company will test its ability to scale and develop into a profitable business entity. The timing of
these phases also needs to coincide with achieving clarity around the technology.
Of course, developing an effective strategy requires a clear understanding of the critical success factors.
Considering what has caused operators to succeed or fail in the past will allow new entrants to position
themselves for future success.
Critical Success Factors for Broadband Wireless Operators
The way to make money in broadband wireless is like in any other business, charge more than it costs you
to provide the service. The complexity arises from the fact that revenues are earned over time and have a
tendency to decrease over time. Costs, on the other hand, are not directly driven by the monthly revenues,
i.e. the network requires money to be spent up front and operating expenses are driven by network costs
69
and usage, neither of which may be directly driven by revenues. Broadband wireless success requires a
company to address five critical issues. Optimizing the interplay between these areas has spelt the
difference between success and failure.
The Right Markets
Selecting the right markets is the most important requirement for the business. The business should be viewed
in the perspective of both a one-year and a five-year horizon. Adequate customers should exist initially so the
company can begin to offer services and generate revenues. However in five years time, the company needs to
be a significant player with a large enough customer base for the investment to pay off.
The Best Technology
The technology platform should at best offer a competitive advantage and at worst be comparable to the
best in the market. Good marketing and excellent customer service cannot compensate for a weak
technology over the long term. Given the high bandwidths and the bundling possibilities offered by DSL
and cable, the wireless operator must find USP to compensate for this.
Strong Systems
This area is often overlooked by most telecom operators and affects the operators ability to scale
effectively. Poor systems can quickly result in poor network quality and unhappy customers due to its
impact on network operations and customer service. Systems and processes should be scaleable and
should allow the company to migrate from a semi-automated solution that can be quickly implemented to
an automated solution that may require a larger investment but offers a lower per-customer cost.
Adequate Cash
Since the telecom battle is won over the long term, the company needs to have enough cash to survive
industry downturns. This requires an extremely sound understanding of the business case and financial
metrics so the company can optimize where funds are invested and ensure promised returns.
Strong Management
Telecom continues to be a dynamic environment that requires management to optimize across markets,
products, technology, and operational areas. Hiring a strong management team will allow the company to
build the operational competencies required to establish a strong competitive advantage. The risk of the
wrong decision far outweighs the cost of a good manager. It is also important to realize that the best and
the brightest can make expensive mistakes; hence the need for constant scrutiny and comparison to the
business plan.
A company that has the optimal combination of these five critical success factors has a good chance to
establish a strong market presence and a profitable business. As mentioned previously, articulation of a
clear strategy starts with a clear understanding of the market need and developing an appropriate product
strategy. Business profitability will continue to flounder until broadband wireless operators are able to
create a USP that draws customers away from DSL and cable modems.
Suggested Product Strategies
A clear articulation of the product strategy will provide the roadmap for the technology selection and
business case. This in turn depends on the dynamics of the specific market and the competitive space
available. Key questions to answer include the following:
70
Does the market have enough room for the incumbent and at least two other players today? As
the history of competitive local-exchange carriers (CLECs) and data local-exchange carriers
(DLECs) has shown, underestimating the incumbent is a surefire recipe for disaster. Always
R.J. Mahadev
factor in a strong competitive response from the incumbent at some point. Stealing market share
is never easy and often impossible. Focus on markets that are growing.
Is there a window of opportunity within which the company can establish a presence before a
strong competitive reaction? Being first to market is a significant advantage as long as the
company is able to consolidate its position.
Will the product offer be competitive? Obviously the company will have a difficult time
attracting customers in a market where the competition is offering lower prices or more
comprehensive bundles. Mixing voice and data will give customers what they want today and
will provide profitable growth areas for the future.
Cost parity with DSL and cable modemsan end-to-end capital per subscriber of around $400
today. Actual pricing will be based on what the market will pay.
Capabilities that incumbents are weak onportability, managed quality of service (QoS),
immediate activation, self-install, etc.
Strategic partnerships with at-risk incumbents like ISPs and satellite providers.
Focus on segments and verticals that value the unique capabilities offered by broadband wireless.
This includes the typical mobile segments like the construction and real estate industries, small
businesses, and professionals. Targeting segments like gamers, media streaming users, and music
download users is also profitable.
As complete an application bundle as possible. Voice over Internet protocol (VoIP) and streaming
media applications are a basic requirement.
Target both wholesale and retail customers. Wholesale customers will greatly increase network
utilization while retained customers will ensure higher average revenues per user.
More basic applications like Internet and low-cost voicekeep the product offering simple.
Offering speed and service options will confuse the market and delay adoption. Wait for the
market to mature and users to become more knowledgeable before trying to segment users.
Price to optimize profitability. The operators in these markets have a great advantage in being
able to charge more due to initial price inelasticity. This needs to be managed effectively; too
many operators drop prices to drive penetration, squeezing their margins without appreciable
revenue gains. One good strategy is for operators to focus on building a smaller network,
71
charging higher prices, and maximizing utilization so that the network costs are recovered in 12
18 months. Operators not following this path should be prepared for a long wait.
Offer a retail-only product initially to discourage wholesale entry and share loss.
Over time, as the market develops and competitors enter, a wholesale product should be offered
to increase network utilization.
Once the product strategy has been determined, the process of technology selection can truly commence.
Given that the network technology is the key determinant of future success, a number of strategic and
technical factors need to be analyzed prior to technology selection.
Selecting the Technology
The number of frequencies and technologies available further complicates vendor selection. Luckily the
frequencies dictate so much of the network behavior that starting with the frequency selection greatly
simplifies the process. The use of a technology selection matrix and a qualified consultant can help speed
through this learning curve.
Figure 2 represents a typical technology selection matrix. Comparisons to incumbent technologies like
DSL should be made to allow competitive benchmarking. Key issues to remember about the various
frequencies are as follows:
72
The sub six gigahertz (GHz) licensed frequenciesmultichannel, multipoint distribution system
(MMDS) licenses in the US, 3.5 GHz licenses in other parts of the worldoffer a number of
benefits over alternate frequencies, including licensed, interference-free operation, and non-line
of sight functionality (self-install, portable operation) at comparatively lower costs.
While a number of WISPs have selected to go with unlicensed frequencies, the significant
interference issues coupled with the bandwidth and power constraints limit the economic viability
of these networks.
R.J. Mahadev
Technology
Issues
Licensed
< 6 GHz
License acquisition
Licensed
>10 GHz
LOS requirements,
cost
Unlicensed
>6 GHz
3G Mobile
DSL
Portability
CPE Cost
Capex/Sub
Operational
Complexity
QoS
Capability
Install
Cost
Interference,
security, capacity
Capacity, cost,
speed
Deployment
issues, QoS, NMS
Legend:
Good
Medium
Bad
Selection of an all IP network reduces the operational complexity and the level to which company
operations are tied to a single vendor.
Look for software defined radios that can be upgraded less expensively and write key upgrades
into the vendor contract, along with penalties for non-compliance.
Almost all vendors are paying lip service to WiMAX. Although this standard has yet to be
finalized and is not comprehensive enough, also write WIMAX compatibility into the vendor
contracts to increase the number of options available.
Dont be fooled by terms like fourth generation (4G). Nobody really knows what this means, and
no commercial roadmap exists for getting there.
Most mobile operators have swapped out their networks at least twice in the last 7 years.
73
Figure 3 contains the important questions that need to be considered in selecting the technology platform.
Each of these issues could make or break the company and require in-depth discussion with prospective
vendors and their customers.
ATTRIBUTE
STRATEGY
COST
CAPABILITY
SCALEABILITY
METRIC
IMPLICATION
)Fit
)Capex/sub
)Install cost/
sub
)Reliability &
service quality
)Ease of
management
)Ease of
Operations
)Interoperability
)Scale ability
)Future proof
Developing a scaleable solution that allows for a low cost initial implementation without
constraining growth.
Provisioning and network management systems that allow for seamless service activation and repair.
Web-based customer care that minimizes support costs and enhances customer lifetime value.
The systems area also separates the ISPs from the telcos. ISPs often choose a low cost solution that allows for
fast entry but limits future growth due to high cost operating costs at scale. On the other hand, telecom
approaches have high upfront costs and often force operators to define business operations before they fully
understand the customer requirements. The savvy operator should select an initial solution that will get them
going and yet should be prepared to completely change their systems once they hit a certain scale.
74
R.J. Mahadev
Figure 4 contains the key systems components that need to be addressed by the broadband wireless
operator to support basic services. Services like VoIP and vertical applications will require additional
hardware and software.
Device Polling
DNS
Event Management
IRC
Event Correlation
Instant Messging
Ticketing
VOIP Gateways
SIP Servers
Shared Hosting
Maintenance Planning
Billing/Invoicing
Payment processing
General ledger
Workflow/Order Mgnt
Messaging/Colaboration
Fin. Reporting
C us tom e r S up po r t
Usenet News
Authentication Services
OSS
Serv ice s
Shared Information
Customer E-mail
SFA
Customer SelfCare
Availability Checker
Web Based Ordering
Web Based Configuration
Web Based Reporting
SLA Reporting
Network Visualization
Auto Provisioning/Config.
Inventory
Usage Accounting
Estimate pricing levels based on existing prices for broadband services, trends from comparable
countries, and operator pricing strategy.
Develop revenue projections considering the expected product mix and the diffusion of the
various customer segments.
The costs for operating the network and services customers are likely to be the next major
expense items.
Other capital expenditure includes licensing costs, data center, and IT networks.
Other operating expenses include the rental costs for site locations, salaries, and support expenses.
75
Identify how sensitive the business case is to pricing, market uptake, and technology choices.
Figure 5 provides the high level metrics of what a broadband wireless operator can expect to see. The
business modeled here covers approximately seven million people in a dense urban environment. As
discussed earlier, the business must focus on building and achieving a certain scale if it is to succeed.
Year 1
Subscribers
# BTS
% Seen/Sold
# Employees
Revenues
Operating Expenses
Capital Expenditure
Net Cash Flow
Cumulative Cash
ARPUs
Mo. Opex/Sub
EBIDTA Margin
Capex/Sub
Compounded NPV
Compounded IRR
Year 2
Year 3
Year 4
Year 5
Year 6
Total 6 Years
1K
6
20
11 K
26
11%
36
35 K
45
11%
56
87 K
47
11%
108
161 K
93
11%
149
236 K
146
11%
170
236 K
146
11%
170
$0M
($ 3 M)
($ 9 M)
$6M
($ 6 M)
($ 6 M)
$ 17 M
($ 11 M)
($ 9 M)
$ 42 M
($ 19 M)
($ 12 M)
$ 70 M
($ 33 M)
($ 27 M)
$ 91 M
($ 44 M)
($ 27 M)
$ 226 M
($ 115 M)
($ 91 M)
100%
-51%
-40%
$ 10 M
$ 17 M
($ 0 M)
$ 17 M
$
48 $
43
$ 22/Mo.
$ 21/Mo.
53%
52%
$
394 $
385
Operating Cashflow Breakeven
Net Cashflow Breakeven
$ 17 M
7%
($ 11 M)
($ 11 M)
$
NA
NA
$ 102 M
72%
($ 6 M)
($ 4 M)
$ 11 M
($ 17 M)
($ 21 M)
($ 10 M)
67 $
60 $
53
$ 64/Mo.
$ 40/Mo.
$ 24/Mo.
5%
33%
55%
$
1,312 $
685 $
419
External Funding Required
($ 21 M)
External Funding Period
36 Months
49%
20 Months
42 Months
Limiting initial capital expenditure to the absolutely necessary elements like the spectrum
licenses, sites required for the proof of concept, and a rudimentary back office system.
Delaying large-scale network deployments to match the market adoption and technology stability.
Focusing on driving cell site saturation in the early years to really understand how to maximize
network utilization.
Managing the seen-sold ratio (cell site over subscription) to conserve scarce radio and IP resources.
The costs to operate the network will overtake the capital costs very quickly. Key components to this are
as follows:
76
Costs to provide and manage the network, including labor costs and backhaul/IP costs. Effective
systems are essential for managing these expenses.
R.J. Mahadev
Due to high labor costs, selective outsourcing of non-core functions will reduce the companys
exposure to market fluctuations.
Minimal revenues should be expected in the first year, due to issues with ramping up the sales channels
and establishing customer and network operations. Even a relatively large number of customers may
result in low revenues due to service credits and billing issues. Wholesale revenues should be treated as
just as important as retail revenues since many ISPs are interested in alternative broadband connections.
A well managed broadband wireless networks should break even within 20 months and pay back the
investment in five years. A large portion of the value of the business will lie in its terminal value and the
ongoing cash flow expected. The key to maximizing profitability includes the following:
Maximize network utilization by offering off-peak packages (revenues with almost zero
incremental costs), tightly managing subscriber over subscription, and employing location-based
sales models.
Having a strong set of operating metrics such as per cell-site profitability, per customer revenues
and costs, and employee efficiency metrics.
Gating capital expenditures based on milestones such as meeting the business case, proving in the
technology, creating a competitive advantage, etc.
As discussed earlier, the development of a robust business case will allow for improved decision-making
and strategy development. This will provide a strong foundation for rapid deployment of the network.
However, it is also worth remembering that no amount of planning can compensate for a high enough
concentration of bad luck.
77
With the convergence happening in public networks and public network operators offering both voice and
data services, an enterprise can get both services from a single operator. Both PSTN (fixed network
operators) and mobile operators offer voice and data services, such as digital subscriber line (DSL)
connections, general packet radio service (GPRS), and third generation (3G) networks. Moving forward,
these networks will offer both services on a single unified network, and this network as per current
evolution is an IPbased packet network offering both voice and data services. Enterprise networks can be
expected to parallel the convergence happening within the public network. Within enterprise networks the
voice and data services will converge to a single unified IPbased packet network. In fact, some of the
other communication needs within an enterprise, such as announcement systems, will also converge with
IPbased networks.
The future convergence of enterprise and public networks will be based on single end-to-end packetbased networks with the public network operator offering both voice and data services to both public and
enterprise users.
There are a few other important requirements for the convergence of public and enterprise networks. A
true converged network will give a uniform feel to the end user and should also provide seamless roaming
between the two. Therefore, any services being offered by a public network should also be available to
enterprise users. A few examples of the existing services include short message service (SMS) and
multimedia messaging service (MMS) services available on the public mobile network. Another
important aspect is that the mobility of the user within the public network (including roaming) should be
available to enterprise users as well. This implies that any new service launched by the public network
operator should be instantly and simultaneously available to both public and enterprise users. However,
this excludes the set of services that are specific to one network or the other; for example, an enterprise
user may want an operator interface or front-desk key board system, which is not a requirement for the
public network user. Another example could be video on demand service, which may be of interest to the
public network user while not required by the enterprise user. One of the easiest ways to meet some of
these listed requirements is through convergence in the end-user device. This means that the device used
by the end user to access the services while in the public network and the device the end user uses to
access services within the enterprise should be unified to a single device. This device might support
different interfaces as required by these two networks, for example wireless fidelity (Wi-Fi) within the
enterprise and Global System for Mobile Communications (GSM) in the public network.
A true converged network will have many more requirements; however, convergence can also be
achieved through a smaller set of requirements, and then later such networks can move toward a true
converged network definition. In fact, a sudden change in the user interface or its behavior toward the
network will also inhibit deployment of such networks. Hence, such networks will slowly move toward a
real or true converged network.
Whats the Need?
The need for any such technology can be analyzed from two different perspectives. One is the pushing
factor, which considers whether there are any services or requirements that are not met by an existing
solution or whether the new technology offers additional gains over and above the existing technology.
This basically translates to ease of use or end-user savings along with gain to the public network carrier.
End user in this context connotes both users of the service and the enterprise as a whole. Some of the
push factors include the following:
80
More and more services getting added to the public mobile networks, such as SMS, MMS, PTT,
and many more, today are not available to enterprise users. Some of these services alone can
really enhance a users productivity and hence can become push factors.
Sunil Mahajan
Another example is user mobility within an enterprise, while remaining reachable, which is
directly related to improved productivity.
Currently within public networks a user can roam from one network to another, or from one city
to other, but does not maintain the same identity when moving from one enterprise office to
another. Convergence can provide seamless, consistent identity.
Other than the above mentioned push factors, there are many other gains or advantages that can help us to
save moneyor earn moreand can help increase productivity. A starting point is an analysis of the
benefits of a real or true converged network with respect to the various entities involved in it. Various
entities, which will gain from such convergence, include enterprise management, enterprise IT and
telecom departments, end users, and public network carriers.
Gains for Enterprises
With every converged technology there are some inherent gains attached, which also comprise part of the
convergence solution. When various network services are converged, such as voice and data on the same
network, less equipment and equipment management is required. This means that both capital expenditure
and operational expenses come down. Capital expenses are reduced as less equipment must be acquired
and operational expenses reflect this reduction. In addition, less space, less power, and fewer people are
needed to manage networks.
In this solution as the network is operated by the public network operator, enterprises need not
spend money on system acquisition and maintenance or on network maintenance and operations.
If the service providers monthly charge is higher than the operational and capital expenses, then
this is not a viable solution, unless it provides other benefits that justify the additional cost.
Another important gain related to system maintenance is the cost of equipment upgrade. With
each additional new feature, enterprises will need to spend additional money for system upgrade,
whereas with a converged solution, this expense is borne by service provider. Any new service
deployment by the service provider will also be instantly available to enterprise users.
Any expansion in the enterprise user base is also taken care of by the network operator, which is
again capital expenditure saving.
If an enterprise has multi-site offices, then connection across these enterprise offices is also taken
care of by the network operator. Enterprise users need not spend money on equipment, network
connection, or capacity expansion.
When uniform service is experienced, end users enjoy a high degree of satisfaction, resulting in
higher productivity.
When the end user can use the same device to access both public and enterprise networks, a lower
investment is required than when two separate devices are needed.
If enterprise network access is based on a wireless network, which supports mobility, reachability
is much higher and in turn results in higher productivity.
81
Public network service providers can increase their revenue by either increasing the subscriber
base or by increasing the average revenue per user (ARPU). With a converged network approach
the service provider can achieve both.
A larger subscriber base is available as the service provider can offer services to enterprise users,
which are not otherwise targeted by public network operators. The call volume generated by these
users is typically higher than that of network users.
The addition of new services in the network, which are typically used by users in public networks,
could now be accessed by enterprise users as well, which in turn will result in higher ARPU.
Service providers will also save on software upgrade costs, as the same set of services will be
made available to both public network users and enterprise users via a single network software
upgrade. Providers need not invest in two different systems for upgrade.
Service providers can also define new sets of services for the converged networks that are
typically not possible in two disparate or separate networkssuch as follow me across the
public and private networkto increase ARPU with additional charges for these added services.
A uniform service experience achieves higher end user satisfaction, which means higher productivity.
With a single device to access both public and enterprise networks, end users and enterprises face
lower costs.
A greater number of services are available to the end user, as all the services that can be accessed
from the public network are also available within the enterprise or vice versa.
However, as it is not easy to build a true converged network to meet all of the requirements previously
listed, most of the deployments will take a step-by-step approach. There are many challenges to be
overcome and many problems to be solved before reaching a true converged network. Industry players,
consortiums, and standard bodies are currently working on these issues.
82
Sunil Mahajan
Technical Solution?
In order to achieve a real converged network, there are multiple requirements that a service provider must
meet. Additionally, a converged network also provides the option to enterprises or users to use only a
portion of the available services, for example, an enterprise might only be interested in converged voice
services and, for security reasons, does not require a data network to be integrated with the public
network. Another enterprise might only be interested in voice services, but over existing circuit networks
and not converged voice services on packet networks. Consequently, a service provider should be flexible
in its offerings.
An evaluation of some of the technical solutions that are available and possible for such a converged
network follows.
There is a requirement for the converged network to offer mobility to the enterprise users, which
implies that the access network within the enterprise needs to be wireless based.
There is a requirement from the perspective of the service provider network to do local switching
of calls within the enterprise for all local terminating calls, so as to save bandwidth or spectrum.
There is a requirement to save on equipment costs, so deployment should not require too much
additional equipment in the network.
There is a requirement to provide seamless roaming between the public network and enterprise
networks.
Such network deployment should not require too much infrastructure deployment within the
enterprise, such as additional wiring or replacing wiring, either by the enterprise or by the
service provider.
New services added to the public network shall also be available to the enterprise users; this shall
not require any upgrades of enterprise equipment.
The new converged network shall add new innovative services to enhance user experience.
In order to meet some of the basic requirements of a real converged network, service providers need to
rely on wireless access within the enterprise at the low cost. Wi-Fi or 802.11x standard can very well
satisfy this need and can also meet some of the other basic requirements such as support of both voice and
data. In order to meet some of the other basic requirements, such as local switching of calls within the
enterprise, the deployed network within the enterprise needs to be voice over IP (VoIP)based; signaling
will still travel to the service provider server, but real-time transport media (RTP) will be switched
locally. Further, to meet the need of innovative converged services, session initiated protocol (SIP) shall
be assumed as standard for VoIP deployment within an enterprise.
The convergence of an enterprise network with a public network based on these technical solutions
achieves the following:
A Wi-Fibased enterprise network which will be set up by the service provider providing both
voice and data services and requiring minimum equipment or wiring within the enterprise.
83
Voice services will be locally based on VoIP protocols and data services will be directly provided
over a Wi-Fi network.
If the service provider upgrades its network to make it VoIPbased, then integration is
straightforward. Otherwise, it needs to deploy a gateway solution which can translate the
enterprise voice to a network-based solution.
In addition, service providers will also need to provide a solution to handle seamless roaming
between the public network and enterprise networks.
Any new service addition in the network will require interworking over Wi-Fi or VoIP protocols.
The service provider will need to provide security for all data access.
The service provider can use either a fixed line connection or wireless connection to connect the
enterprise to the public network.
The service provider may require a software upgrade to the switching software if it does not
already support enterprise solutions (like a Centrex or hosted PBX solution)
The above described solution will form the foundation of future converged networks where the service
provider will also upgrade its public network to support the real converged network (VoIP or in general
media over IPbased network), which will result in lower interworking requirements and hence lower
equipment cost.
Challenges Involved?
Building a real converged network will be a big challenge for service providers; however it will be
deployed on an incremental basis, with each step adding more convergence into the solution, either at the
network level or at a service level. Some of the basic issues that should be addressed for basic deployment
are as follows:
84
Wi-Fi network deployment within an enterprise should be resistant to security attacks, such as
parking lot attacks. (The IEEE is working on enhanced security standards.)
A Wi-Fi network shall provide quality of service for real time traffic, such as voice. (The IEEE is
working on quality of service [QoS] standards.)
Solution deployment shall support seamless roaming between enterprise and public networks. It
shall have common authentication and security mechanisms. (There are companies working on
roaming between GSM and Wi-Fi networks.)
Availability of dual-mode handsets will be a big challenge. (There are few vendors producing these.)
Interworking between Wi-Fi enterprise networks and cellular networks must be supported.
Sunil Mahajan
For data convergence, service provider networks must support enterprise specific security
policies, including firewalls and network address translations (NATs) for each enterprise solution.
A business model for both enterprises and service providers must be created.
Conclusion
As discussed, convergence is happening at various levels in the telecom domain, within both public
networks and enterprise networks. Enterprise and public wireless network convergence has started to
happen in a small way with the trials related to roaming between GSM and Wi-Fi networks, which is a
move toward real converged networks. It is apparent that public networks are converging to a single
unified networkIMS for 3Gand enterprise networks also converging to either VoIP or VoIP over WiFi with data over the same network, which will eventually lead to convergence of these two networks.
With the right business model from a service provider, converged networks will bring value to all users.
Though the discussions in this white paper primarily focused on convergence or public mobile networks
and enterprise (wireless) networks, the same principals can also be applied to convergence of public fixed
network operators and enterprise networks (wireless or IPbased).
85
87
States, for example, has been triggered by the implementation of the Telecommunications Act of 1996. This
increased competition, along with the threat of market saturation looming, has forced telecommunications
companies to look across the border at the potential for acquiring a share of the global marketplace. However,
as we continue to enter into this realm, marveling at all of the possibilities, it has also become apparent that
standardization has become a critical issue. When telecommunications companies develop products, it is
important that these products can operate in a global economy, thus alleviating the necessity for creating
multiple telecommunications products that use multiple standards and protocols. From a wireless
telecommunications standpoint, the type of standard is characterized by certain factors such as the type of radio
transmission technology that is used, the frequency band that is used, and the modulation scheme that is used.
Though there are other factors that affect standardization, these three factors play a very important role.
Never has the issue of standardization been more apparent than it is in Asia, where there has been a
proliferation of telecommunications services. Just recently, Nihon Keizai Shimbun (2004) reported that
Japan, China, and Korea, in an effort to adopt a unified communications protocol, have decided to jointly
develop 4G technology and communications. This move alone could have a major impact on the
telecommunications industry given that these three countries combined account for 30% of all cell phone
users worldwide (Nihon Keizai Shimbun, 2004, P.1).
To better understand the impact of standardization and operability, two charts are used to highlight the
differences between existing wireless technologies. Table 1 provides a comparison between two mobile
access technologies, 3G mobile and 4G mobile. Table 2 shows a comparison between a number of fixed
access technologies including: 802.11b, 802.11g, 802.11a, 802.11e, 802.11i, 802.11h, 802.11j, 802.11n,
802.15.1 (Bluetooth), and 802.16a (WiMAX).
Characteristics
4G Mobile
Network Architecture
Wide-area cell-based
Speed
Frequency Band
Bandwidth
Switching Scheme
Component Design
IP
L. Calvin Price
Physical
Layer
802.11 b
WiFi
802.11 g
WiFi
802.11 a
WiFi-5
802.11 e
QoS
802.11 i
Security
802.11 h
Europe
802.11j
Japan
802.11 n
(draft)
802.15.1
Blue-tooth
802.16 a
WMAN
11 Mbps
54 / 11
Mbps
54 Mbps
----
----
54 Mbps
54 Mbps
108 Mbps
10-55 Mbps
2-200 Kbps
70 Mbps
DSSS:
CCK/BPS
K/QPSK
OFDM/
CCK/
PBCC
OFDM
----
----
OFDM
OFDM
MIMOOFDM
FSSS/
w/GFSK
OFDM/
OFDMA
2.4 GHz
2.4 GHz
5 GHz
----
----
5 GHz
5 GHz
5 GHz
2.4 GHz
2 - 11 GHz
< 200 ft
< 200 ft
< 50 ft
----
----
< 1.5 km
< 1.5 km
TBD
< 30 ft.
4-6 miles
< 300 ft
40-300 ft
----
----
1.2 - 9 km
1.2 9 km
TBD
30 300 ft
30 miles
Security
WEP /
WPA
WEP-152
bit
----
TKIP (1st)
AES (2nd)
WPA /
AES
WPA / AES
TBD
WEP 128
bit
Triple-DES
RSA-1024 bit
HDTV/
MPEG 2
QoS
w/ HCF
TKIP (1st)
AES (2nd)
TPC &
DCS/DFS
TPC /&
DCS/DFS
MIMO
cheap
Last Mile,
Q o S & TPC
Data rate
Modulation
Scheme
Freq Band
AP - AP
Distance
Special
Features
WEP64/128 bit
CD audio/
MPEG 2
Channels 11/(3) @ 22
Access
Method
11/(3) @
22 MHz
12/(8) @ 20
MHz
----
----
MHz
19 @ 20
MHz
4 @ 20
MHz
@20 MHz
or 40MHz
79 (3) @ 1
MHz
8/16 UL @
1.5-20 MHz
CSMA/
CA
CSMA/
CA
CSMA/
CA
HCF &
EDCF
----
CSMA/
CA
CSMA/CA
TBD
CSMA/CA
Dynamic
TDM/TDMA
89
CDMA2000 operating mode, interoperability between operating modes (CDMA2000 and W-CDMA)
does not exist and remains an issue for 3G.
Second-Generation Issues
2G, though not referred to in Table 1 experiences a slightly different but related problem. There are
actually multiple standards for 2G. 2G on its own is not formally considered a standard by the ITU. The
three most common 2G standards are CDMAOne (IS-95), Global System for Mobile Communications
(GSM), and Time Division Multiple Access (TDMA) (IS-136). Each of these digital standards can
operate in the same personal communications service (PCS) frequency range of 1.9 GHz. However, each
utilizes a different radio technology and modulation scheme, which are characteristic of the standard.
Therefore, the lack of interoperability in 2G occurs because of lack of standardization.
802.11 Issues
One problem surrounding the lack of standardization in 802.11 has to do with the use of different
frequency bands. In Table 2, each 802.11 physical layer (PHY) represents a different standard. A
comparison can be made between 802.11g wireless fidelity (Wi-Fi) (802.11b upgraded) and 802.11a (WiFi5, i.e., 5 GHz) to highlight the issue of lack of standardization. An assumption is sometimes made that
the two standards are interoperable because they can theoretically achieve the same data rate54
megabits per second (Mbps). However, although both standards can operate at the same data rate
(theoretically), they do not share the same frequency band. Thus, interoperability between 802.11b/g and
802.11a is not possible. 802.11a operates in the less congested 5.5 GHz band while 802.11b and 802.11g
operate in the unlicensed and very congested 2.4 GHz band. On the other hand, 802.11g is backward
compatible with 802.11b since both standards can operate in the same frequency band (i.e., 2.4 GHz), and
share a common modulation scheme (complementary code keying).
In addition to the standardization issue associated with incompatible frequency bands, 802.11 standards
do not all use the same modulation schemes. For example, although 802.11b and 802.11g both use
multiple modulation schemes (i.e., adaptive modulation), namely orthogonal frequency division
multiplexing (OFDM), complementary code keying (CCK), and packet binary convolution coding
(PBCC), they both share the ability to use CCK modulation. This allows interoperability to occur between
the two standards. However, 802.11a does not use multiple modulations schemesit is limited to OFDM.
Although the use of OFDM in and of itself is not a bad thing, it does limit 802.11as ability to
communicate with 802.11b. Note that Intel has recently developed a chip that can be used to work with
both 802.11a and 802.11b. Thus, some efforts are being made to address some interoperability issues.
However, it is very clear that lack of standardization and interoperability are significant issues for fixed
access technologies such as 802.11.
Quality of Service
Convergence
An interesting trend has been witnessed recently. It would be difficult to deny the fact that technology and
demand are moving towards convergence, that is, the integration of voice, video, and data all traveling
over a converged mobile access and fixed access network. In order to support convergence, the necessity
for implementing QoS has become a priority. This is a crucial issue, especially when one considers that
voice and video are time-critical applications. In other words, unlike data transmission, voice and video
must be delivered in real time.
Black (1997) highlights several challenges associated with the integration of voice, video, and data
applications. First, voice and low quality video transmissions show a significantly high tolerance for
errors. However, data protocol data units (PDUs) show a low tolerance for errors. Second, there is the
issue of network delay. The ability to translate digitized voice back to an analog signal in real time
90
L. Calvin Price
requires a constant and relatively low network delay tolerance, typically less than 200 milliseconds (ms).
Therefore, timing between sender and receiver is crucial. Packet data, however, does not fall under the
same constraints since it can be transmitted asynchronously without regard to timing. Voice transmission
falls under similar constraints, but voice fidelity will not be significantly affected as long as packet losses
do not exceed more than one percent of the total packet transmission. Finally, in an effort to help reduce
delay and make it predictable, voice and video transmissions require short queue lengths at each network
node, even at the expense of experiencing some packet loss. In contrast, data packets require longer queue
lengths during overflow conditions to prevent packet loss (Black, U., 1997, P. 414).
Video transmission becomes an even greater issue in that it requires more bandwidth than voice and data.
Part of the bandwidth issue is addressed by creating bigger pipelines. It is also addressed by using
compression algorithms such as Moving Pictures Experts Group (MPEG)-4. Thus, there is a need to
allocate bandwidth dynamically. This is known as dynamic bandwidth allocationa key aspect of QoS.
Convergence QoS 4th Generation
Notice that in the above relationship, evolution went straight to 4G and not 3G. This is primarily because
3G cellular networks, like their 2G predecessors, were not originally designed for QoS. QoS was
developed to address real time transmission issuessomething that has not been satisfactorily addressed
in cellular networks, where the primary focus has been to deliver voice and some data. Now that the need
for convergence is imminent, 3G networks that are currently being built do not completely address the
need for convergence. The definition of 3G does not include convergence and when it comes to the issue
of convergence, these networks are only hybrid in nature at best, addressing voice and video transmission
separately. Additional modifications have to be made to 3G networks in order to allow them to address
the need for convergence.
Benefits of QoS
In many networks, QoS is the most important issue. In fact, in moving towards convergence many would
argue that QoS would rank first in comparison to other issues such as security, transmission speed,
interoperability, and spectrum saturation. Following are some of the reasons why this is so.
First, the goal of QoS is to provide a level of guarantee to all traffic within the network. This is especially
true for real-time applications such as voice and video. To be more specific, new applications such as
voice over IP (VoIP) and streaming video are less tolerant to network delay (latency) and as networks
grow larger, bandwidth management becomes even more crucial. Bandwidth management is an important
aspect of QoS.
Second, QoS is also designed to address such issues as denial of service (DoS) attacks. In other words,
QoS products are being tested and designed to handle and manage bandwidth, even in the face of DoS
attacks that can add significant amounts of congestion to a network.
Third, QoS can help address the issue of edge-network bottlenecks. For example, as high speed local-area
network (LAN) traffic is handed off to a lower speed wide-area network (WAN), a potential bottleneck
can occur. Traffic management helps to anticipate and alleviate this bottleneck.
Finally, from a business perspective, QoS will become even more vital as convergence looms nearer. This
is true not only because of applications such as VoIP or streaming video, but also for important and
specialized business applications and processes such as PeopleSoft. This can be a very intensive process
on a network and it must be assured network availability.
91
L. Calvin Price
same performance. In addition, the coverage for a single 802.11a AP is between 40300 ft. However, the
high data rate of 54 Mbps promised by 802.11a drops off dramatically as the distance from the AP
increases, typically around 60 ft. This has a significant impact on 802.11a AP coverage. One of the not so
obvious reasons for this is that 802.11a operates at a higher frequency (5.5 GHz) band than 802.11b/g (2.4
GHz) and wavelengths (lambdas) are proportional to range or coverage.
Alternatively, many current users of 802.11 customer premise equipment are perfectly satisfied with
802.11b throughputs (up to 11 Mbps). They are not necessarily concerned with achieving the 54 Mbps
data rates offered by 802.11a (or 802.11g for that matter). However, as more and more bandwidth
intensity applications are added to fixed wireless networks, and as more and more users are added,
bandwidth and data rates will become a premium.
To look at this in more detail, note that 802.11a offers more bandwidth than 802.11b. 802.11a uses 12
non-overlapping channels. Most 802.11a products typically use only eight channels giving a maximum
throughput of 432 Mbps (eight times 54 Mbps). 802.11b products are designed for 11 channels, but only
three are non-overlapping (i.e., a maximum of three APs per area). This means that the maximum
throughput for 802.11b is around 66 Mbps (three times 11 Mbps).
This paper will now introduce two solutions to address these issues. One solution can be used to address
the mobile wireless access issues found in 2G and 3G, while the other solution can be used to address the
fixed wireless access issues found in 802.11. The first discussion considers the mobile access solution, 4G
mobile IP. This will then be followed by the second solution, 802.16e, WiMAX with limited mobility.
Both solutions are considered 4G.
Mobile Broadband Wireless Access 4G Mobile IP Solution
One of the primary advantages of 4G mobile IP is that it uses the IPv6 Internet protocol. The goal of 4G
is to forge standardization within the telecommunications industry by developing an all IPbased mobile
network. However, an allIP network is not its only advantage; 4G addresses all of the issues discussed in
2G and 3G. Along with adhering to a single and advanced IP protocol (IPv6) standard, 4G mobile IP uses
two advanced modulation schemes: OFDM and multi-carrier CDMA access (MC-CDMA). In addition,
4G mobile IP works in a very wide frequency band between two to eight GHz. In fact, 4G IP mobile is
compatible with and independent of most radio transmission technologies, including those for
CDMA2000, W-CDMA, 802.11, 802.16, 802.15, and Hyper-LAN I and II. Finally, 4G mobile IP will be
capable of achieving data rates as high as 256 Mbps with channel bandwidths of 100 MHz.
In addition, support for 4G is very strong, both nationally and internationally. Those companies that
support 4G have a common goalopen architecture. One such architecture is known as Open Wireless
Architecture. According to the Fourth Generation Mobile Forum (4GMF), the Open Wireless
Architecture, is designed to support the convergence of broadband wireless mobile and wireless
access, as well as convergence with wireline networks (4GMF, 2004, p.1). This architecture modifies the
application (IP), media access control (MAC), and PHY layers. Although 4G has not officially been
designated a standard, a number of organizations, both nationally and internationally, have began to lay
the foundation for 4G. The 4GMF is an international technical body whose primary focus is on broadband
wireless mobile communications. Probably the largest organization by far, which was created with the
goal of helping to accelerate the deployment of Internet-based standards for mobile wireless networks is
the Mobile Wireless Internet Forum (MWIF). Founding membership in the MWIF is a list of whos who
companies in the telecommunications industry. The list in alphabetical order includes the following:
Alcatel, Bell Atlantic Mobile, Celletra, Cisco Systems, Compaq, Comverse Network Systems, DDI
Corporation, Ericsson, Fujitsu, Hewlett-Packard, Hitachi, Hyundai Electronics, IBM, IP Mobile, KPMG
Consulting, LG Information and Communications Ltd., Lucent Technologies, Malibu Networks,
93
Microsoft, Motorola, NEC, Nokia, Nortel Networks, Orange, Portal Software, Qualcomm, Samsung
Electronics, Sharp Corporation, Siemens, SK Telecom, Solect, Sony, Sprint PCS, Sun Microsystems,
Synacom Technology, Tekelec, Telcordia, TELOS Technology, Telstra, Telesystem International
Wireless (TIW), Toshiba, 3Com,Vodafone AirTouch, and WIND.
Sun Microsystems, also an MWIF founding member, has taken on a research lead in an effort to develop
4G mobile IP. SUN is also attempting to create an open system environment, mainly for software and tool
development, based on the Solaris operating system. According to James Kempf (2002) of Sun
Microsystems, 4G is all about an integrated, global network that is based on an open systems approach.
The goal of 4G is to replace the current proliferation of core cellular networks with a single worldwide
cellular core network standard based on IP for control, video, packet data, and VoIP (Kempf, 2002, p.1).
Fixed Broadband Wireless Access WiMAX Solution
There are two modes of WiMAX (802.16a and 802.16e) and each has its own priority, although in general
they will serve similar purposes. WiMAX-802.16a is being primed to carry backbone traffic for WLAN
hotspots. It is also being promoted as a viable solution for one of the biggest bottlenecks in a
telecommunications network, the last mile. Those who are pushing 802.16a look to replace the
dependence on T-carrier one (T1), digital subscriber line (DSL), and cable and also look to deploy it in
areas where there is currently no wireline infrastructure in place. WiMAX-802.16e on the other hand, has
a different purpose. It is being primed as a means to add mobility to the WLAN or WiMAX network.
WiMAX-802.16e allows for the combined operations of both fixed and mobile wireless in the same
frequency bands. WiMaX also has a large following. A forum was created to address interoperability
issues based on the specifications from the 802.16 task group. This is called the WiMaX Forum and
includes companies such as Intel, Agilent, and Nokia, to name a few. Note that Intel plans to integrate
WiMAX into its mobile chipsets with Wi-Fi, so computers and laptops that use Intel chips can be
expected to be able to support both Wi-Fi and WiMAX (See Table 1).
Fixed Broadband Wireless Access WiMAX Benefits
Like 4G mobile IP, the benefits of WiMAX are numerous, giving rise to the opinion held by many who
perceive it as a disruptive technology. A selection of the benefits follows.
94
Designed with QoS. This means that as we move more towards convergence (ie. voice, video, and
data transmission all within the same network), WiMAX will already have the ability to manage
bandwidth between multiple applications such as VoIP, high speed data, and streaming video.
Designed with non line of sight (NLOS) in mind. The original 802.16 standard was designed for
frequencies between ten to 66 GHz (line of sight environments). The 802.16a standard is
designed to operate in the two to 11 GHz frequency range, making it possible for NLOS
environments. This was achieved because of modulation changes. 802.16a uses three modulation
schemes, single carrier (SC), 256-OFDM, and OFDMA modulation. OFDM modulation performs
well in NLOS environments and this is one of the reasons why OFDM is so important. Mesh,
beam forming, and multiple inputs, multiple outputs (MIMO) antenna technology helps to further
improve NLOS in WiMAX.
Designed to operate in both licensed and unlicensed frequency bands. 802.16a can cover a very
wide frequency range, from 2 GHz to 11GHz. 802.16e also covers a wide frequency range,
though slightly less, from 2 GHz to 6 GHz. Currently, it covers three primary bands: 2.5 GHz
(licensed); 3.5 GHz (licensed); and 5.8 GHz (unlicensed).
L. Calvin Price
Has a very high data rate. Typical channel transmission rates can reach 100 Mbps (theoretically)
and 70 Mbps (15 Mbps for 802.16e). Thus, a single base station with 4 access units could
conceivably deliver 280 Mbps.
Works with different packet delivery mechanisms including IPv4, IPv6, Ethernet, and virtual
local-area network (VLAN).
Highly secure. 802.16a uses triple data encryption standard (DES) (128 bit) security and RSA
(1024-bit)
Uses OFDMa very spectrally efficient (about five bits per second per hertz bps/Hz) modulation
scheme. OFDM was also chosen over spread spectrum technology (CDMA) because it is even
more spectrally efficient than CDMA (about 1.6 bps/Hz). CDMA modulation techniques,
especially direct sequence, use processing gain (PG) to overcome co-channel interference.
Therefore, the available bandwidth has to be larger than the data throughput.
Uses a dynamic access method for MAC. WiMAN-802.16a/e uses time division multiplex on the
downlink and time division multiple access on the uplink (TDM/TDMA) for efficient bandwidth
usage. This is a more versatile access method than CSMA/CAa contention-based access
method. TDM/TDMA is good for delay-sensitive applications such as voice and video and allows
for collision-free access to the channel. CSMA/CA does not offer guarantees on delay.
802.16a equipment has a range of up to 30 miles (three miles for 802.16e), outdistancing all other
WLAN equipment. With this kind of range, services can be delivered to numerous homes and
businesses without having to use multiple APs. Note that advanced antennae technology is also used.
Designed to use transmit power control (TPC). TPC was also included in standard specifications
802.11h (Europe) and 802.11j (Japan). TPC allows users that are close to an AP to reduce
transmission power in order to reduce interference with other users.
How does the mobile network determine the optimal access technology to use during handoff? Is
the signal quality better on the WiMAX network or on the 4G mobile IP network?
QoS variations between networks or adaptation of multimedia transmission across networks. (i.e.,
what happens when the QoS requirements that manage bandwidth for multimedia are different
between two networks during a handoff? If this is not properly addressed, multimedia quality is
impacted, defeating the advantages of having QoS).
95
4G Mobility Issue
As noted, both 4G mobile IP and 802.16e are designed around the IPv6 protocol. However, it is worth
pointing out that from a mobility standpoint, there are some inherent problems with IPv6. One problem is
that when it comes to fast handoffs, IPv6 is only suited for global mobility, that is, inter-domain mobility.
Unfortunately, studies show that almost 70% of user mobility is intra-domain or site, or stated another
way, mobility between access points within a network. IPv6 falls short in its ability to address both
macro-mobility (intra-domain mobility) and micro-mobility (rapid mobile node movement between base
stations within a subnet/intra-subnet).
IPv6 also has problems with location mobility. Basically, it uses simple location updates that fall short of
addressing problems with signaling scalability (network performance as nodes increase) and hand-off
latency (delays associated with MN binding updates to the home address and correspondence hosts).
These are two problems that can have a serious impact on multimedia transmission.
IPv6 will have to be supplemented with protocols that can address macro- and micro-mobility handoff
issues. Some of these are cellular IP, tele-MIP, hierarchical MIP (H-MIP), HAWAII, TR45.6, and DMA.
Tele-MIP can be used for micro-mobility management and can also be used to reduce location updates.
However, it does not provide for macro-mobility. HMIP, HAWAII, DMA and TR45.6 can be used for
macro-mobility and global mobility, but not micro-mobility. On the other hand, cellular IP compares
favorably to the others. It addresses all three hand-off issues: intra-subnet, intra-domain, and interdomain. In addition, a new protocol, intra-domain mobility management protocol (IDMP), is being tested.
It provides for a multi-casting and duration-limited solution to the problem.
4G Paging Issue
Another problem regarding IPv6 is that it does not support paging, unlike what has been accomplished in
other cellular infrastructures. In IPv6, a location update is accomplished by the mobile or MN, not the
network (i.e., a mobile switch). This has a significant impact on battery power. However, by
supplementing IPv6 with cellular IP, this problem can be addressed.
Additionally, the use of mobility anchor points (MAPs) can be implemented into IPv6 to act as additional
nodes or middle-men to help reduce handoffs. IPv6 is not scaleable; as the number mobile node hosts
increase, so also do the binding updates that they send to the home address (HA) and correspondence
hosts. Unfortunately, this increases the overall signaling within a network, which increases network
congestion and delay. Since IP networks can be divided into domains and subnets, MAPS can be assigned
to each domain. A MAP can hide local mobility from the HA and correspondent agents. MNs need only
send one binding update message to the local MAP instead of sending multiple messages to the HA and
correspondence agents that are further away. This cuts down on congestion. Also, when the MN is outside
the domain, the correspondent agent is not notified.
4G Fixed and Mobile Wireless Access Technology WiMAX and 4G Mobile IP
Fixed and Mobile Wireless Convergence
A very important aspect to the development of 4G mobile IP is its parallel migration with 802.16
WiMAX. As noted previously in the introduction, 4G has been defined as a standard for the transmission
of integrated voice, video, and data over a converged mobile broadband wireless access (MBWA) and
fixed broadband wireless access (FBWA) allIP network. The diagram depicts the migration from
802.16e to 4G mobile IP. The ultimate goal is for WiMAX and 4G mobile to converge.
96
L. Calvin Price
2G
14 Kbps 144 Kbps
1.9 GHz
802.11 e
802.11 g
(QoS Enhancement)
802.11 a
54 Mbps @ 5GHz
OFDM
802.11 n (Draft)
100 Mbps @ 5GHz
802.16 a (3G)
802.16 e (4G)
4G Mobile IP
802.11 i
HIPERLAN2
QoS & Security
54 Mbps @ 5 GHz
OFDM
3G
144 Kbps 2 Mbps
< 2.7 GHz
802.11 h & j
(Security Enhance)
WPA. TKIP & AES
802.20 (4G)
1 - 4 Mbps @ < 3.5 GHz
OFDM
54 Mbps @ 5GHz
OFDM
DFS, TPC, DCS
Figure 1
Summary
This paper has discussed four issues: lack of standardization and interoperability, QoS limitations, high
infrastructure development costs, and data rate limitations. The analysis described issues from a mobile
wirelessaccess perspectivenamely 2G and 3Gand also from a fixed wirelessaccess perspective
namely 802.11. After analyzing these issues, this paper proposed two solutions. 4G mobile IP was
introduced to show its potential for resolving the issues associated with previous mobile-access
technologies. 802.16e (WiMAX) was introduced (with 802.16a) to show its potential for resolving the
issues associated with previous fixed-access technologies. Also analyzed were some of the potential
problems that both solutions could face due to their reliance on IPv6, which was not designed for true
mobility. Finally, this paper shows how both solutions are connected in a way that they will converge,
creating a true 4G IPbased network.
References
Black, U. (1997). Emerging Communications Technologies, 2nd Edition, Prentice Hall Series in
Advanced Communications Technologies.
Cyber-Media (2004), Sting Brings WiMax in a 'Box' to India. Retrieved August 8, 2004,
http://www.ciol.com/content/news/2004/104071608.asp
Das, S., McAuley A., and Dutta, A, (2000). IDMP: An Intra-Domain Mobility Management Protocol for
Next Generation Wireless Networks, Columbia University, Research Web Site:
http://www1.cs.columbia.edu/~dutta/research/idmp.pdf
Haraldsvik, R. (2004) Two Worlds Come Together over IP, Wireless Week, 1. Retrieved August 12,
2004, http://www.wirelessweek.com/article/CA426379?spacedesc=Departmentsandstt=001
97
Kupetz, A. and Brown, K. (2004). A Look into the Future of Wireless Communications, Rollins Business
Journal, 1. Retrieved August 10, 2004, http://www.crummer.rollins.edu/journal/articles/
2004_1_4G.pdf
Marek, S. (2003) U.S. Spotlight Shines on EV-DO, Wireless Week, 1. Retrieved August 12, 2004,
http://www.wirelessweek.com/article/CA292170?spacedesc=Wireless+Internetandstt=001
NEASIA (2004), Japan, China, Korea to Jointly Develop 4G Cell Phone Technology, Nikkei Business
Publications,
Inc.
Retrieved
August
10,
2004,
http://neasia.nikkeibp.com/wcs/leaf?CID=onair/asabt/news/299929
Mobile Info.com (2001), 4G Beyond 2.5G and 3G Wireless Networks. Retrieved August 10, 2004,
http://www.mobileinfo.com/3G/4GVisionandTechnologies.htm
Mobile Tech News (2000), Mobile Wireless Internet Forum Formed in Cannes, 1. Retrieved August 10,
2004, http://www.mobiletechnews.com/info/2000/02/03/141822.html
Sun Microsystems (2002), All IP Wireless, All the Time. Building the 4th Generation Wireless Network
with Open Systems Solutions. Retrieved August 10, 2004, http://research.sun.com/
features/4g_wireless/
TD-SCDMA Forum (2004), China Sees TD-SCDMA Ready by Mid-2005, TD-SCDMA News, 1.
Retrieved August 13, 2004, http://www.tdscdma-forum.org/en/news/see.asp?id=254
3G Today, (2004), 3G Is Here Today 3G Operators. Retrieved August 10, 2004, http://www.3gtoday.com/
4Gcoulk Limited.com (2004), 4G Wireless Trial. Retrieved August 10, 2004, http://www.4g.co.uk/
PR2004/Jan2004/2009.htm
4Gcoulk Limited.com (2004), Fast Track to 4G for India. Retrieved August 10, 2004, http://www.4g.co.uk/
PR2004/May2004/2024.htm
98
99
3G Technologies
What Is 3G?
According to the International Telecommunication Union (ITU), the International Mobile
Telecommunications-2000 (IMT-2000) is the global standard for 3G wireless communications, defined
by a set of interdependent ITU Recommendations.
The IMT-2000 provides a framework for worldwide wireless access by linking the diverse systems of
terrestrial- or satellite-based networks, exploiting the potential synergy between mobile communications
networks technologies and networks for fixed and mobile wireless access systems.
The ITU activities on IMT-2000 reflect international standardization, including frequency spectrum and
technical specifications for radio and network components, tariffs and billing, technical assistance, and
studies on regulatory and policy aspects.
In the ITUs vision, the 3G services should provide the following:
144 kilobits per second (kbps) high mobility (vehicular) data transmission
384 kbps low mobility (pedestrian) data transmission
Two Mbps stationary (untethered) wireless data transmission
These data rates should allow a new variety of services offered over the wireless mobile network, such as
on-line gaming, broadcast video, multimedia messaging, video on demand, fixed broadband, push-to-talk,
camera phone, notebook personal computer (PC) access, Web content, and of course, telephony.
The ITU also identified the following bands for IMT-2000 3G services:
There are various incompatible air interfaces in 3G, leading to several implications. The most obvious is
that users traveling worldwide will need more than one phone until multimode phones are available in the
market. Moreover, the radio interface determines the capacity of the network and dictates how the system
deals with such issues as multipath effects, interference, and handing off active calls from one base station
to another as users move from cell to cell. Figure 1 shows the IMT-2000 Terrestrial Radio Interfaces
defined in ITU-R Rec. M.1457.
100
CDMA
CDMA
CDMA
FDMA
TDMA
TDMA
TDMA
IMT-2000
CDMA
MULTICARRIER
IMT-2000
CDMA
DIRECT SPREAD
IMT-2000
CDMA
TDD
IMT-2000
TDMA
SINGLE CARRIER
IMT-2000
FDMA/
TDMA
(CDMA2000
1X AND 3X)
(UWC-136: EDGE)
(DECT)
101
DECT
Up to 500 kbps
Circuits
RF Channel
1728 MHz
2G: GSM
Data 9.6 kbps
Circuits
RF Channel
200 KHz
DECT
Up to 2 Mbps
Circuits
RF backward
Compatible
2.5G:GSM/GPRS
Data 40 Kbps
Packets
RF backward
compatible
3G: UMTS
384Kbps-2Mbps
Packets
New RF 5MHz
Bandwidth
2G: IS-136
Data 9.6 kbps
Circuits
2.5G: IS-136+
2.5-3G: EDGE
384 kbps
Packets
RF
compatible
2G: IS-95-A
Data 14.4 kbps
Circuits
RF Channel
1.25 MHz
2.5G:IS-95-B
Data 64 kbps
Packets
RF backward
Compatible
CDMA200
1X MC
153 kbps
Packet
RF backward
compatible
PDC
Data 9.6 kbps
Circuits
PDC
28.8 kbps
Circuits
RF backward
Compatible
CDMA200
3X MC
384 kbps
Packet
RF backward
compatible
1xEV-DO (HDR)
PHASE 1
2.4 Mbps
Packet
RF backward
compatible
1xEV-DV
PHASE 2
2.4 Mbps
Packet
RF backward
compatible
W-CDMA
(JAPAN)
2003-2005
102
Standard
Nickname
Data Rate
Data Rate
Forward Channel Reverse Channel
1x
153 kbps
(also called 1xRTT)
CDMA2000 Rev. A 1xEV-DO
2.4 Mbps
CDMA2000 Rev. B
CDMA2000 Rev. C 1xEV-DV
3.1 Mbps
CDMA2000 Rev. D 1xEV-DV
3.1 Mbps
CDMA2000 Rev. 0
153 kbps
307 kbps
307 kbps
1 Mbps
103
Internet data model. Enhancements to the 1xEV-DO standard, referred to as Revision A, increase data
speeds, reduce latency, and provide quality of service (QoS) mechanisms making the standard viable for
real time applications such as voice over IP (VoIP).
1xEV-DV in CDMA2000
The CDMA2000 the 1x evolution to data and voice (1x EV-DV) provides support for voice, mixed
voice and data, dedicated data, and real-time two-way services on a single carrier. A single 1.25 MHz
bandwidth is shared between voice and data users. It provides 3.1 Mbps peak data rate, and voice
users are usually scheduled first. The ability to support simultaneous (circuit) voice and packet data
sessions to the same terminal is extremely important in establishing a smooth migration path to
3GPP2s all-IP environment.
Standard
Radio
Channel
200 KHz
200 KHz
GPRS
EDGE
CDMA2000 1X
1.25 MHZ 153 kbps
(voice and data)
CDMA 1x-EV-D0
1.25 MHz 2.4 Mbps
(data only)
CDMA 1x-EV-DV
1.25 MHz 2.4 Mbps
(voice & data optimized)
2 Mbps (fixed)
UMTS
5 MHz
144 kbps (mobile)
Table 2: Summary of Data Rates and Bandwidths in the Different 2.5G and 3G Systems
GSM and GPRS
Same spectrum
Same cell sites
Voice (GSM)
Packet data (GPRS)
Low speed data
200 KHz radio channel
UMTS
New spectrum allocation
New cell sites
Higher capacity voice
Higher speed data
5 MHz channel space
1xEV-DO
Same spectrum
Same Cell Sites
Packet data only
Higher speed data
Multimedia services
1.25 MHz radio channel
104
Public access
802.11a
54 Mbps
5 GHz
OFDM
25-75 feet
Incompatible with
802.11a & 802.11g
Limited
Interference issues
Cordless telephones
Standard
Maximum data rate
RF operation band
modulation
Range
compatibility
802.11b
11 Mbps
2.4 GHz
Direct sequence
150 feet
802.11g
54 Mbps
2.4 GHz
DS/OFDM
100-150 feet
802.11g
8021.11b
Maximum
Microwave ovens,
cordless telephones,
Bluetooth
Normal
Microwave ovens,
cordless telephones
105
802.16a (MAN)
The 802.16 is the IEEE standard for fixed wireless metropolitan area network (MAN), called the
wireless last mile. It provides different data rates depending on the band of operation and the
propagation conditions. The 802.16 operates in the band ten GHz to 66 GHz. In fixed position and line of
sight (LOS) propagation conditions it can provide up to 134 Mbps in a 28-MHz-wide channel, suitable for
business-to-business communications.
The 802.11a, approved in January 2003, operates in licensed and unlicensed frequencies between two and
11 GHz that support non LOS (NLOS) communications suitable for residential and small business users,
can have a data rate of up to 70 Mbps per sector, a range of up to 30 kilometers (km), and can support
thousands of users. The ETSI standard HiperMAN is based on a subset of 802.16a).
WiMAX can provide backhaul for Wi-Fi Spots. It also can be use to create hot areas bigger than hotspots,
for example, areas of service not limited to designated lounges, but anywhere in an airport.
The WiMAX Forum is an industry-led, nonprofit corporation formed to promote and certify compatibility
and interoperability of broadband wireless products. Their member companies support the industry-wide
acceptance of the IEEE 802.16 and ETSI HiperMAN wireless MAN standards.
802.16e (MAN)
The IEEE 802.16e mobile wireless MAN standard is a revision of the 802.16 that supports roaming for
low-speed mobile users at pedestrian speed. This standard explicitly does not target high mobiles users.
The applications will include support for both low-latency data and real-time voice services. 802.16e
mobile devices will be backward compatible with 802.16a fixed stations. It has a target data rate of 70
Mbps; it will operate in the two to six GHz licensed bands, with typical channel bandwidths ranging from
1.5 to 20 MHz.
802.16
Air Interface
10-to-66 GHz
LOS
2001
802.16c
Amendments
System
Profiles
10-to-66 GHz
LOS
2002
802.16a
Amendments
MAC & PHY
2-to-11 Ghz
NLOS
2003
802.16d
Revision
PAR for
802.16 &
802.16a
2004-2005
802.16e
Amendments
for pedestrian
mobility
speeds
2-to-6 GHz
NLOS
2006-2007
106
Standard
Maximum data rate
RF operation band
Channel
bandwidths
Range
Mobility
Advantages
802.16a/Rev. d
6070 Mbps per sector
2-to-11 GHz
Flexible between
1.25 and 20 MHz
4-8 km typical,
up to 30 km NLOS
Fixed
(indoor and outdoor)
Long range, high data
throughput, thousands
of users per site
802.16e
Up to 15 Mbps
2-to-6 GHz
Flexible between
1.25 and 20 MHz
802.20
1 Mbps
500-to-3500 MHz
Open
25 km NLOS
Up to 15 km NLOS
Portable/Mobile
Mobile
3G
802.20
UMTS
1xEVDO
EDGE
OPEN
WCDMA
2 Mbps
(fixed)
144 kbps
(mobile)
TDMA
215 km
NLOS
1.5-to-20
MHz
1.5-to-20
MHz
OPEN
5 MHz
2-11 GHz
Licensed &
Unlicensed
4-to-8 km
typical,
30 km Max
NLOS
2-6 GHz
Licensed &
Unlicensed
500-3500
MHz
Licensed
Contained
in 1.7-to-22
GHz
CDMA
2 Mbps
Max
600 kbps
typical
1.25, 2,5
and 3.75
MHz
Contained
in 1.7-to-22
GHz
2-5 km
NLOS
Up to 15
km
NLOS
215 km
NLOS
220 km
NLOS
1 Mbps
384 kbps
200 KHz
800, 900,
1800, 1900
MHz
107
range
30 m
3 Km
15 Km
20 Km
30 Km
Data
rate
50 Mbps
802.11x 802.16a
15 mbps
802.16e
802.20
3G/CDMA2000
1 Mbps
Fixed/nomadic
Low mobility
high mobility
mobility
108
Henry Mitchel
Systems Architect, Modular Communications Platform Division,
Communications Infrastructure Group
Intel
Sanjay Bakshi
Broadband Wireless Network Architect, Mobile Networking Lab,
Corporate Technology Group
Intel
Prakash Iyer
Senior Staff Architect, Mobile Networking Lab, Corporate Technology Group
Intel
Abstract
IEEE* 802.16 is an emerging global broadband wireless access standard capable of delivering multiple
megabits of shared data throughput supporting fixed, portable, and mobile operation. The standard offers
a great deal of design flexibility including support for licensed and license-exempt frequency bands,
channel widths ranging from 1.25 to 20 MHz, quality of service (QoS) establishment on a per-connection
basis, strong security primitives, multicast support, and low-latency/lowpacket loss handovers1 such as
soft handoff and fast cell search features. Predominant deployment of subscriber stations (SS) and access
points2 (AP) for portable and mobile services are expected to be based on scalable orthogonal frequency
* All other brands are the property of their respective owners.
1
Optimization of PHY and MAC handover primitives is ongoing in the 802.16e Task Group and is expected to be
completed by the end of 2004.
2
In this paper the term access point is synonymous with base station.
109
division multiplexing with multiple access (OFDMA). Initial deployments are likely in licensed
frequencies below 11 GHz. A key to broad industry acceptance of 802.16 is a significant reduction in
overall operational expenditure (OpEx) and capital expenditure (CapEx) at comparable or better
performance relative to cellular data networks. This paper describes an open, extensible Internet protocol
(IP)based end-to-end network architecture framework that can be built using IP common off the shelf
(COTS) technologies and flexibly support a variety of radio access network topologies while
accommodating a number of new and incumbent operator IP backend networks. Topics core to an end-toend framework such as QoS, security, mobility management, interworking, and access to managed IP
services are discussed in more detail.
Introduction to MBWA and 802.16
IEEE 802.16 is an emerging suite of air interface standards for combined fixed, portable, and mobile
broadband wireless access (MBWA). Initially conceived as a radio standard to enable cost-effective last
mile broadband connectivity to those not served by wired broadband such as cable multiservice operators
(MSO) or digital subscriber line (DSL), the specifications have evolved to target a broader market
opportunity for mobile, high-speed broadband applications. The 802.16-2004 [1] standard supercedes all
previous versions as the base standard. The 802.16e [2] amendment and the soon-to-be-approved 802.16f
and 802.16g task groups will amend the base specification to enable fixed, portable, and mobile operation
in frequency bands below 11 GHz. 802.16 is optimized to deliver high, bursty data rates to subscriber
stations but the sophisticated medium access control (MAC) architecture can simultaneously support realtime multimedia and isochronous applications such as voice over IP (VoIP) as well. This means that IEEE
802.16 is uniquely positioned to extend broadband wireless beyond the limits of todays Wi-Fi* systems,
both in distance and in the ability to support applications requiring advanced QoS such as VoIP,
streaming video, and on-line gaming.
The technology is expected to be adopted by different incumbent operator typesfor example, wireless
Internet service providers (WISPs), cellular operators (3GPP and 3GPP2) and wireline broadband
providers. Each of these operators will approach the market with different business models, based on their
current markets and perceived opportunities for broadband wireless as well as different requirements for
integration with existing (legacy) networks. As a result, 802.16 access network deployments face the
challenging task of needing to adapt and integrate or interwork with different backend network
architectures and provide low-latency and optimized mobility management, while still supporting
standardized components and interfaces for multivendor interoperability.
To address the challenge and enable multi-vendor interoperability, this paper proposes an end-to-end
architecture framework built around open, IP standards. An allIP end-to-end architecture approach can
deliver significant CapEx and OpEx gains, examples of which are mentioned in the following:
110
CapEx: CapEx benefits over 3Glike deployments will come from lower equipment acquisition
and deployment costs, due to the use of COTS network elements. Dependence on widely tested
and deployed IP standards leads to Internet economies of scale. Furthermore, a common radio
access network (RAN) and consistent end-to-end architecture that is incumbent operator type
agnostic avoids unnecessary proliferation of incompatible networks, thereby simplifying
multioperator, multinetwork interworking.
OpEx: OpEx benefits are likely to come from lower cost to rollout services to users and network
maintenance, cheaper and faster network capacity augmentation, optimal backhaul dimensioning,
and use of lower cost, potentially unlicensed spectrum backhaul wireless links. The use of COTS
software on open mobile client platforms such as laptops and personal digital assistants (PDAs),
use of Internet-scale interoperable solutions for network management, network redundancy, and
overlay of traffic engineering, QoS and fast switching capabilities over less-reliable backhaul
links can all contribute to the desired savings.
Broadband Wireless Usage and Deployment Scenarios
Initial deployments of IEEE 802.16 standards-based networks will likely target fixed-access connectivity
to unserved and underserved markets where wireline broadband services are insufficient to fulfill the
market need for high bandwidth Internet connectivity. Prestandard implementations exist today which are
beginning to address this fixed-access service environment. Standardization will help accelerate the ramp
for these fixed-access solutions by providing interoperability between equipment and economies of scale
resulting from high volume standards based components.
As IEEE 802.16 solutions evolve to address portable and mobile applications, the required features and
performance of the system will increase. Beyond fixed-access service, even larger market opportunities
exist for providing cost-effective broadband data services to users on the go. Initially this includes
portable connectivity for customers who are not within reach of their existing fixed broadband or wireless
local-area network (WLAN) service options. This type of service is characterized by access that is
unwired but stationary in most cases, albeit with some limited provisions for user mobility during the
connection. In this manner, 802.16 can be seen as augmenting coverage of 802.11 for private and public
service networks and cost effectively extending hotspot availability to wider ranges of coverage. Based on
this described capability, this phase of deployment is referred to as portability with simple mobility.
The next phase of functionality, known as full mobility, provides incremental support for low-latency,
lowpacket loss real-time handovers between APs at speeds of 120 km/hr or higher, both within a
network and between networksdelivering a rich end-user experience for high quality multimedia
applications. Figure 1 summarizes our deployment evolution vision of the 802.16 standards.
111
To support the incremental functionality beyond fixed-access deployment, there are required
enhancements to both the air interface and network infrastructure. Both of these enhancements must also
be standardized before the realization of interoperable services meeting end user demands can be
achieved. To understand these requirements, we need to examine usage models and service models for
each stage of 802.16 deployment. From these usage expectations, we can then draw conclusions about
required system capabilities that must be driven into the end-to-end architecture, interfaces, and network
features. The usage evolution is depicted in Figure 2.
112
This usage enhancement over fixed access requires enhancements to security such as strong mutual
authentication between the user/client device and the network (AP) supporting a flexible choice of
credential types. Portable and mobile devices need a means for authenticating trusted APs and detecting
rogue APs. Such mutual authentication is not present in the fixed access standard. A common
centralized mechanism for user authentication is also needed as users may move between different APs
within an IP prefix or subnet, or across APs in different subnets, or even roam to other service
providers in different locales.
The next stage, portability with simple mobility, describes a more automated management of IP
connections with session persistence or automatic reestablishment following handovers between APs.
This incremental enhancement allows for more user-transparent mobility and is suitable for latencytolerant applications that use transfer control protocol (TCP) [13] while not providing adequate handover
performance for delay and packet loss sensitive real-time applications such as VoIP.
In the fully mobile scenario, user expectations for connectivity are comparable to those experienced in 3G
voice/data systems. Users may be moving while simultaneously engaging in a broadband data access or
multimedia streaming session. The need to support low-latency and lowpacket loss handovers of data streams
as users transition from one AP to another is clearly a challenging task. For mobile data services, users will not
easily adapt their service expectations based on environmental limitations that while technically challenging
are not directly relevant to the user (such as being stationary or moving). For these reasons, the network and air
interface must be designed up front to anticipate these user expectations and deliver accordingly.
The 802.16 Radio Scaling to Full Mobility
The 802.16 standard provides an excellent framework upon which systems can be built to satisfy the
broad spectrum of usage models described above. Of the three PHY supported in the standard, scalable
OFDMA is the most versatile and the one preferred for operation across channel widths ranging from
1.25 MHz to 20 MHz. Single carrier access (SCa) will likely be considered for backhaul links while
OFDM with 256-point fast fourier transform (FFT) is best suited for fixed access in up to 10 MHz
channel widths. Scalable OFDMA supports features (enhanced over OFDM) that are especially suited for
high-speed mobile operation such as the following: downlink and uplink subchannelization, fixed
subcarrier spacing (by maintaining constant ratio of FFT size to channel width), and reduced overhead for
cyclic prefix (CP) by keeping its duration constant at 1/8th the OFDMA symbol duration.
The 802.16 MAC is designed primarily for point-to-multipoint (PMP) applications and is based on collision
sense multiple access with collision avoidance (CSMA/CA). The 802.16 access point MAC manages uplink
(UL) and downlink (DL) resources including transmit and receive scheduling. The MAC incorporates several
features suitable for a broad range of applications at different mobility rates, such as the following:
Four service classesunsolicited grant service (UGS), real-time polling service (rtPS), nonrealtime polling service (nrtPS), and best effort (BE)
Privacy key management (PKM) for MAC layer security. PKM version 2 incorporates support for
extensible authentication protocol (EAP)
Manageability primitives
113
Three power management levelsnormal operation, sleep, and idle (with paging support)
These features combined with the inherent benefits of scalable OFDMA make 802.16 suitable for highspeed data and bursty or isochronous IP multimedia applications.
End-to-End Architecture Evolution
Figure 3 conceptually depicts the architecture evolution for 802.16. A basic 802.16-2004based fixedaccess (indoor3 and outdoor) deployment is typically accomplished via a static provisioning relationship
between an SS and an 802.16 AP. The collection of APs and interconnecting routers or switches comprising
the 802.16 RAN can be logically viewed as a contiguous cloud with no interAP mobility requirements
from an SS perspective. The 802.16 RAN(s) interconnect via a logically centralized operator IP core
network to one or more external networks as shown. The operator IP core may host services such as IP
address management, domain name service (DNS) [12], media switching between IP packet-switched data
and public switched telephone network (PSTN) circuit-switched data, 2.5G/3G/Wi-Fi* harmonization and
interworking, and virtual private network (VPN) services (provider hosted or transit).
Indoor operation may require use of beam forming or multiple input multiple output (MIMO) advanced antenna
systems (AAS) which are supported in the 802.16 standard.
114
Going from fixed access to portability with simple mobility involving use of mobile SS (MSS) such as
laptops and PDAs introduces network infrastructure changes such as the following: the need to support
break-before-make micro- and macromobility4 handovers across APs with relaxed handover packet loss
and latency5 (less than 2 seconds), cross-operator roaming, and the need to support reuse of user and
MSS credentials across logically partitioned 802.16 RAN clouds.
Going from Portability to Full Mobility requires support in the 802.16 RAN for low (~zero) packet loss
and low latency (< 100 msec) make-before-break handovers and mechanisms such as idle mode with
paging for extended low power operation.
Besides mobility another important design consideration for the design of 802.16 RAN is QoS. The
802.16 RAN will need to deliver bandwidth and/or QoS on demand as needed to support diverse realtime and nonreal-time services over the 802.16 RAN. Besides the traditional best-effort forwarding,
the 802.16 RAN will need to handle latency intolerant traffic generated by applications such as VoIP
and interactive games. Fixed access and portable usage models need only support acceptable QoS
guarantees for stationary usage scenarios. Portability with simple mobility introduces the requirement
to transfer the service-level agreement (SLA) across APs involved in a handover, although QoS may
be relaxed during handovers. Full mobility requires consistent QoS in all operating modes, including
handovers.
The decoupling of the 802.16 RAN from an operator IP core network permits incremental migration to
fully mobile operation. An operator must however give due consideration to the 802.16 RAN topology
(such as coverage overlap, user capacity, and range) to ensure that the physical network is future-proof for
such an evolution.
End-to-End Reference Architecture
Figure 4 depicts an end-to-end reference architecture for 802.16. Various functional entities and
interoperability interfaces are identified. The network essentially decomposes into three major functional
aggregations: the 802.16 SS/MSS, the 802.16 RAN and the interconnect to various incumbent operator
core and application provider IP networks. The IP core network a) manages the resources of the 802.16
RAN and b) provides core network services such as address management, authentication, service
authorization, and provisioning for 802.16 SS/MSSs.
Micromobility refers to handovers between APs within the same IP prefix or subnet domain. Macromobility refers
to handovers across APs in different IP prefix or subnet domains.
5
Latency may be unacceptable for real-time IP services such as VoIP during handovers but acceptable for TCP and
VPN services as well as store-and-forward multimedia services.
115
subsystem (IMS), remote enterprise intranets, the PSTN, and the Internet. Each RNSN instance manages
a cloud of APs across a hybrid wireline/wireless backhaul network and is responsible for radio resource
management (RRM), data forwarding, and interconnects to backend networks. Functions such as QoS,
mobility, and security are cooperatively managed as a network of managed APs. An RNSN may also host
802.16 RANspecific centralized functions such as paging groups and macro mobility agents, an example
of which could be a mobile IP (MIP) foreign agent (FA) and so on. When interworking with existing core
networks, an RNSN may be rendered via enhancements on a convenient existing network element such as
a packet data gateway (PDG) [5] in 3GPP networks or an equivalent gateway in a 3GPP2 network or a
broadband remote access server (BRAS) [16] in xDSL networks or on a standalone router platform.
AP
An 802.16 AP (referred to in the 802.16 standard as a base station) is a physical entity that implements
802.16 compliant functions such as SS/MSS admission control, uplink and downlink scheduling, over the
air traffic forwarding, handovers and enforcement of over the air QoS, and encryption. We envision a
number of 802.16 RAN topologies as depicted in Figure 5.
117
A combination of an APC with one or more APT instances may render a multi-sector cell. Where
multiple APTs are managed by an APC, the APC may host a common MAC instance across all APTs or a
dedicated MAC instance for each APT. An APC would typically localize all micromobility functions
across its managed APTsand as such would support all relevant 802.16 PHY, MAC, and convergence
sublayer (CS) service access point (SAP) primitives. An APC may also host optional wireless link
services such as header suppression, payload compression, and MSS paging.
An AP hosting more than one logical APC instance can optimize control and management plane functions
across all hosted instances. Factors such as projected scalability requirements (coverage, user density),
degree of mobility, and need for incremental network growth would drive an operators choices between
the different AP configurations. The architecture framework, however, is agnostic to specific 802.16
RAN topologies and can support a mix of all possible variants simultaneously.
Mobile/Fixed Subscriber Station (MSS/SS)
MSS/SSs form the third most important functional aggregation in the end-to-end framework. We envision
that most operator networks would over time have to support a mix of SSs with varying degrees of
mobility support.
Interoperability Interfaces
Figure 4 identified several key interoperability interfaces within the end-to-end framework. The
functionality and purpose of each of these interfaces is discussed below. All interfaces are bidirectional
unless noted otherwise.
SS/MSS and AP Interface
This is the control, data, management and service plane interface between fixed-only or mobile subscriber
stations and 802.16 APs. The functions supported over this interface include but are not limited to the
following:
As noted earlier, the 802.16 standard presents a rich selection of optional features, which in turn presents
significant interoperability challenges to the Industry. We expect the WiMAX* Forum to define profiles
targeting operation in specific frequency bands, channel widths, PHY modes and duplexing modes to
drive multi-vendor interoperability. All such applicable profiles will be incorporated in this interface
between the SS/MSS and AP.
Access Network to Core Network (CN) Interface
This interface has two variants. One represents the control, data and management planes between 802.16
ANs and an operators core network (with interfaces in turn to other remote networks). The other one
represents control, management and service planes to ASP networks. Both of these interfaces are exposed
by the RNSN and enable a consistent allIP interface to diverse core networks. The functions modeled
over this interface may be provided by a cluster of servers, for example, dynamic host configuration
118
protocol (DHCP), DNS, IMS Core Network components such as proxycall session control function (P
CSCF), interrogatingCSCF (ICSCF), servingCSCF (SCSCF), media gateway (MGW), and so on.
These interfaces may also host IP tunnels to carry data between provider networks.
The functions supported over this interface include but are not limited to the following:
Assignment of traffic engineering parameters for provisioned QoS for both control and data
plane traffic
119
during intersubnet handovers. Full mobility requires zero/lowpacket loss and low-latency handovers that
are acceptable to real-time applications such as VoIP.
The end-to-end reference architecture classifies mobility management into macromobility and
micromobility, as illustrated in Figure 6. Micromobility management refers to the set of procedures that
detect the movement (handover) of an MSS between two APs within a single 802.16 RAN administrative
domain, and switch all data flows from the previous (serving) AP to the new (target) AP while attempting
to ensure minimal delay and data loss during the transition. Micromobility may or may not involve a
change in the IP address used to route data to the MSS. In other words, micromobility implies MSS
movement when the IP address via which it was reachable does not change. It utilizes address registration
and associated traffic redirection procedures that are transparent to protocols and procedures operating at
layer 3, i.e., IP (and above) in the protocol stack of the MSS (i.e., micromobility procedures and protocols
operate at layer 2 [and below] in the protocol stack of the MSS.) For the transport mechanism within the
802.16 RAN, this paper recommends the use of multiprotocol label switching (MPLS) [11] or IPinIP
tunneling with DiffServ [10] provisioned QoS to switch data paths across traffic engineered backhaul
links during handovers for micromobility. With MPLS, we recommend fast-preprovisioned label
switched paths (LSP) switching between the RNSN and AP/APC, which perform the role of label edge
routers (LER). Efficient MAC layer handover triggers and limited micromobility signaling would be used
to initiate traffic forwarding/multiple unicasting and switching to minimize handover latency and data
loss between RNSN and AP/APC.
(and above) in the protocol stack of the MSS. For macromobility this paper recommends the use of
session initiation protocol (SIP) mobility for real-time low-latency interactive applications like VoIP and
mobile IP for all other generic applications.
The 802.16 RAN can leverage the IP differentiated services QoS model or MPLSbased traffic
engineering technologies to provide appropriate forwarding treatment to end-user traffic flows as they
traverse between an RNSN and APs.
Harmonization and Interworking with Public Wi-Fi, 2.5/3G, and xDSL Networks
As noted earlier, different incumbent operators are likely to deploy 802.16 networkseither as a data
overlay network or as a standalone broadband access network. Integration with an existing operator
network would involve either harmonization or interworking as subsequently defined.
Interworking
Interworking refers to the technical and business relationship between two operators that have deployed
access networks using the same or different radio technologies. This relationship enables subscribers to
authenticate/authorize to their home operator network via the visited network and utilize system functions
and IP services offered by both networks.
Figure 7 depicts two operators in an interworking relationship, each with an independent core network,
having different types of radio access technologies. In such a relationship, roaming subscribers may be
offered different tiers of seamless service experience in the visited network.
Figure 7: Interworking
Harmonization
Harmonization is defined as the deployment scenario wherein two or more access networks using same or
different radio technologies and operated by a single operator are offered as an integrated network to
subscribers. Figure 8 depicts a simple harmonization scenario wherein a 3GPP operator offers an 802.16
RAN and 3GPP RAN as an integrated network to its subscribers. In this scenario the 802.16 RAN does
not have an independent core but is connected directly into the operators 3GPP core network via an
appropriate gateway. Roaming subscribers may be offered different tiers of seamless service experience
in the visited network.
121
Figure 8: Harmonization
For interoperator 802.16 interworking, the interworking framework should support the following goals:
An operator typeagnostic one-bill roaming (via common, extensible remote authentication dialin user service (RADIUS) [6] and DIAMETER [7] accounting primitives) framework across
802.16 networkseventually leading to seamless IP services mobility across these networks.
Support reuse of credentials and cryptographically strong bilateral authentication and session key
management across these networks.
A provisioning and access framework for advanced IP services that is compatible with the
architecture for Wi-Fi hotspots.
Enable offering of multiple IP services with attributes such as provisioned bandwidths, SLAs,
QoS, and variable tariff profiles.
122
123
125
126
In general, a gateway between the managed services network and 802.16 RANs is configured and
preprovisioned to associate with one or more service session management entities in the service network.
Case Study: Accessing VoIP Service in IMS
As SS/MSS acquires its IP connectivity during network entry followed by acquisition of its corresponding
PCSCF. Then, the SS/MSS registers with the SCSCF. The registration request is forwarded via the P
CSCF. Assuming that the end user of the MSS has subscription with this IMS provider, SCSCF
authenticates the MSS and retrieves the user profile. After successful authentication, the MSS is
registered and ready to initiate or receive VoIP (SIPbased) calls. To setup a VoIP call, the SS/MSS sends
an INVITE to the SIP infrastructure and addressed to the correspondent callee. The PCSCF forwards the
request to the callees SCSCF based on the setup service route information. The SS/MSS always
exchanges SIP signaling such as call setup for exchanging audio codec, bit rate, and quality of service
parameters with its correspondent callee via the PCSCF. When the SS/MSS receives an ACK back from
the corresponding callee, the VoIP call session is established. Note that the MSS may change its subnet
address, as it moves around in the 802.16 access network. When the IP address of the MSS changes, the
MSS needs to register the new IP address with its SCSCF and a REINVITE SIP message is sent to the
MSSs corresponding callee.
Conclusion
Although wireless networks and radio coverage in general has proliferated over the years, data service
offerings continue to be either limited in range (as in 802.11) or deficient in data speed and cost as in
wireless wide-area networks (WWANs). Wireless data rates for WWANs are limited and high-cost partly
due to the inherently granular physical and network layer specifications that burden the WWAN RAN and
core switching fabric, and partly due to the limited available bandwidth for operation. As extended battery
life and reduced size of laptops affords increased portability, so does the need for ubiquitous connectivity
with rich data content at affordable prices. By delivering a combination of higher modulation schemes
within greater channel bandwidths and link budget margins that are comparable to wide area wireless
systems, IEEE 802.16 is uniquely positioned to extend broadband wireless beyond the small islands of
service afforded by Wi-Fi systems today. But one of key requirement for widespread deployment and
adoption of WiMAX is a RAN that is not only improves on CapEx and OpEx requirements over existing
cellular network but can also seamlessly interwork with existing networks. This paper presented a flexible
end-to-end network architecture that can be built using off-the-shelf IPbased building blocks and thus not
only brings Internet scale economies to WiMAX RAN but can also interwork with existing network easily.
References
[1]
Part16: Air Interface for Fixed Broadband Wireless Access Systems, IEEE P802.16-REVd/D52004
[2]
Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems, IEEE
P802.16e/D3-2004
[3]
[4]
[5]
127
[6]
[7]
[8]
[9]
Dial
In
User
Service.
Related
RFCs
at
128
IP Data Communication
over the Wireless Network
A WiMAX Challenge
Shilpa Bhatnagar
Technical Leader
Hughes Software Systems
Meena Belwal
Engineer Trainee
Hughes Software Systems
Madhav Kumar
Engineer Trainee
Hughes Software Systems
Objective
The objective of this paper is to suggest a solution for internet protocol version four (IPv4) data
communication over the 802.16 media access control (MAC). It shall be noted that the technical
challenges, solutions as well as assumptions that are mentioned in this paper might be applicable to other
packet data services in some form or the other. An attempt has been made to discuss at length two basic
implementations IPv4 over 802.16 and IPv4 over Ethernet (802.3) over 802.16. This paper also gives a
complete overview of the Worldwide Interoperability for Microwave Access (WiMAX) technology and
all the enhancements made in this domain to address issues like mobility.
Evolution of Wireless
The wireless revolution for packet data services started in the 1970s at the University of Hawaii with a
simple bi-directional wireless computer network in the star topology which led to further work in the
scientific and industrial sectors into the mid 1980s. In 1985 the band between 902 megahertz (MHz)
and 5.85 gigahertz (GHz) was thrown open to the public. This sparked the development of wireless
adapters by a host of companies and the need for a standard was apparent. The Institute of Electrical
and Electronics Engineers (IEEE) 802 group that had built the 802.3 and 802.5 standards initiated the
802.11 standard for wireless networks. This standard is limited by the fact that the range of operation is
short and the number of users supported is fewer. With a growing number of consumers, this standard
became insufficient.
129
Recently broadband technology has become a product required by a high percentage of the worlds
population. In the past two years alone, the demand has risen rapidly, with a worldwide installed base of
57 million lines in 2002 rising to an estimated 80 million lines by the end of 2003. A healthy growth rate
is expected to rise steadily over the next few years and reach the 200 million mark by 2006. Digital
subscriber line (DSL) operators, who initially focused their deployments in densely populated urban and
metropolitan areas, are now challenged to provide broadband services in suburban and rural areas where
new markets are quickly being established.
The broadband wireless access industry, which provides high-rate network connections to stationary sites,
has matured to the point at which it now has a standard for second-generation wireless metropolitan area
networks. IEEE Standard 802.16, with its air interface, sets the stage for a worldwide revolution in
network access.
What Is WiMAX and 802.16?
The WiMAX organization is a nonprofit association formed in 2003 by equipment manufacturers to
propagate wireless equipment compliant with the IEEE 802.16 Wireless Metropolitan Area Networking
(WMAN) standard. The main focus of this forum is to ensure the compatibility and interoperability of
broadband wireless access equipment for 802.16 networks. As a result 802.16 technology has come to be
known as WiMAX.
WiMAX will primarily be aimed at making broadband network access widely available without the
expense of wires (as in cable-access broadband) or the distance limitations of a digital subscriber line.
This will lie to rest the last mile problem or the expensive and time-consuming process of wiring
subscribers to their immediate network.
In this context, the purpose of 802.16 is to standardize broadband wide area wireless networking for both
fixed and mobile connections, offering extremely high bandwidth connections without requiring line-ofsight communications between the device and the broadcast antenna.
The standard 802.16 specifies the air interface of a fixed point-to-multipoint broadband wireless access
system providing multiple services in a wireless metropolitan area network. The MAC layer defined in
this standard is capable of supporting multiple physical layer specifications optimized for frequency
bands of applications. This standard specifies the physical layer operational on 1066 GHz frequency
ranges and hence line of sight. The MAC is comprised by three sub-layers: the service specific
convergence layer, the common part layer, and the privacy layer.
Service Specific Convergence Layer
The service specific convergence layer is responsible for converting the service specific protocol data
units (PDUs) to the data units which the underlying layer understands and initiating the appropriate
signaling. There are two basic types of convergence sub-layer specifications depending on the overlying
services that are being used, e.g. asynchronous transfer mode (ATM) and packet data. Over these two
client signal (CS) layers various applications can reside. In the case of ATM, it can be ATM adaptation
layer two (AAL2) and, in the case of packet data, it can IPv4 Stack or internet protocol over Ethernet
(IPoE ) amongst others.
Common Part Sub Layer
The second layer, which forms the backbone of the layered stack, is known as the common part layer.
This layer is responsible for the following: systems access, bandwidth allocation, connection
management, connection maintenance, packing, concatenation, and fragmentation.
130
Privacy Layer
The third layer, which deals with data security, is the privacy layer, which is responsible for
authentication and data encryption and decryption.
In this version of the 802.16 standard, a backbone of base stations is connected to a public network, and
each base station supports hundreds of fixed subscriber stations, which can be public Wi-Fi "hot spots" or
fire-walled enterprise networks. The base stations will use the Media Access Control layer as defined in
the standarda common interface that makes networks interoperableand would allocate up-link and
down-link bandwidth to subscribers according to their needs, on a real-time basis. There are two basic
transmission modes: the continuous mode and the burst mode. This is provided by the time division
duplex (TDD) and frequency division duplex (FDD) support. Time division/demand multiple access
(TDMA) is used in the up-link and time division multiplexer (TDM) in the downlink. On the downlink
path, the subscriber station is associated with a specific burst while on the up-link path the subscriber has
to contend for bandwidth and the time at which it can transmit information. The subscriber station does
not transmit and receive simultaneously.
The network is organized in point-to-multipoint (PMP) architecture. The subsequent amendments in the
802.16 systems have standardized the support for meshed network and hence in some way they are
multipoint-to-multipoint. The MAC layer in these systems is connect-oriented and supports different user
environments. There is a protocol independent core (ATM, IP) and the MAC supports multiple 802.16
physical layers.
Subsequent Amendments
With the limited feature set provided by 802.16 and its support of only fixed wireless systems and non
line of support, various amendments are being made over and above this standard to support features like
high mobility, non line of sight, and efficient packet data services. The sections given below capture the
work done and the amendments made in this regard.
802.16a
802.16a was approved as IEEE standard on January 29, 2003 and published on April 1, 2003. It operates
in the 211 GHz frequency band over a theoretical maximum range of 31 miles with a theoretical
maximum data transfer rate of 70Mbps and can support thousands of users.
Since the standard 802.16a specifies that the physical layer is operational in 211 GHz frequency range, it
has the non line of sight support due to larger wavelengths. Hence, 802.16a greatly improves non-line-ofsight performance, and it is the most appropriate technology available when obstacles such as trees and
buildings are present. Stations can be mounted on homes or buildings rather than towers on mountains.
The aspects of 802.16a that are instrumental in powering robust performance include the following:
131
802.16a technology also provides low latency for delay-sensitive services such as circuit-switched
voice traffic or voice over IP, optimized transport for video, and prioritization of data traffic. This is
especially important for businesses that require voice, in addition to data services, from their
broadband service provider.
802.16b
802.16b operates in the 2.4 GHz frequency band over a theoretical maximum range of 100 meters with a
theoretical maximum data transfer rates of 11Mbps and can support dozens of users. An amendment over
this was further made which captured the interoperability issues among various vendors as well as welldefined test suite structures. The standardization of these test suite structures was imperative to solve the
interoperability issues.
802.16e
A breakthrough in WiMAX occurred with the formation of 802.16e working group to standardize the
broadband wireless access for both fixed and mobile access systems. With 802.16e, WiMAX is expected
to support mobile wireless technology in terms of wireless transmissions directly to mobile end users.
This will be similar in function to the general packet radio service and the "one times" radio transmission
technology (1xRTT) offered by phone companies.
The network model as per this specification is such that each base station (BS) unit serves a multiple
services switching system (MSS) over a predefined area. For the mobility across these predefined areas
each BS unit is connected over a backbone network. This backbone network can be provided by different
vendors and could be of different natures. The subscriber stations (SS) in 802.16e are known as mobile
subscriber stations (MSS) as they are no longer fixed. The backbone networks can have there own
authentication and accounting procedures. Due to the existence of this backbone network the BS stack
contains additional support in the MAC known as the mobility agent (MA). This MA is responsible for
de-capsulation of the data arriving from MSS home network, as well as encapsulation when it is destined
to its home network. It is also responsible for connection management for the arrival of new MSS and
departure of the old ones.
802.20: An Introduction
The enhancement of the 802.16 standard to meet mobility requirements was good idea in that
standardization was easier over the existing architecture and backward compatibility was maintained with
systems under development. However, with its enhancements, 802.16 retained its existing disadvantage,
inefficient adaptive antenna support and handoff procedures. Hence a decision for writing a completely
new standard for the mobile wireless access systems was made and 802.20, the IEEE standard for
wireless wide-area networking (WWAN), came into being. The aim of IEEE 802.20 is to develop the
specification for an efficient packet based air interface that is optimized for the transport of IPbased
services. The goal is to enable worldwide propagation of cheap, omnipresent, and interoperable
multivendor, mobile, broadband wireless access networks that meet the needs of business and residential
end user markets. The standard will enable a single base station to support both fixed and mobile
broadband wireless access (BWA). It hopes to bridge the divide between high-data-rate wireless localarea networks (WLAN) and high mobility cellular wide-area networks (WAN). The basic advantage of
802.20 is that it supports mobile devices, a capability not supported by 802.16 until the 802.16e standard
is ready.
802.20 is currently under development and no products supporting 802.20 are expected before 2006.
However, the IEEE Standards Board has approved a mobile broadband wireless access (MBWA) working
group for the establishment of IEEE 802.20.
132
The focus of MBWA is specification of physical and medium access control layers of an air interface for
interoperable mobile broadband wireless access systems, operating in licensed bands below 3.5 GHz,
optimized for IP data transport, with peak data rates per user in excess of one Mbps. It supports various
vehicular mobility classes up to 250 Km/h in a metropolitan are network (MAN) environment and targets
spectral efficiencies, sustained user data rates, and numbers of active users that are all significantly higher
than achieved by existing mobile systems.
Plans have been proposed to support 802.20 MBWA using orthogonal frequency division multiplexing
(OFDM). This is an inverse multiplexing technology that divides a single high-speed channel into
multiple parallel low-speed channels that do not overlap. It is similar to dense wavelength division
multiplexing. In theory, OFDM can offer excellent throughput and reliability.
Comparative Study: 802.16e and 802.20
Both emerging 802.16e and 802.20 standards will specify new mobile air interfaces for wireless
broadband. The basic difference between these two standards is that the 802.16e specification will be
based on an existing standard (802.16a), while 802.20 is starting from scratch. This means that products
based on 802.16e will likely hit the market well before 802.20 solutions. The table below captures the
basic differences between these two mobile technologies
Table 1
The 802.20 interface seeks to boost real-time data transmission rates in wireless metropolitan area
networks to speeds that rival DSL and cable connections (one Mbps or more). Based on cell ranges of up
to 15 kilometers or more, it plans to deliver those rates to mobile users even when they are traveling at
speeds up to 250 kilometers per hour (155 miles per hour). This would make 802.20 an option for
deployment in high-speed trains. However, the 802.16e project authorization request specifies that it will
only support subscriber stations moving at vehicular speeds and hence not at very high speeds.
Although these two technologies differ in their mode of operation they both work on packet data
architecture and ensure low latency.
Data Communication over Wireless
After the advent of wireless for voice communication the most intuitive thing was to communicate data
traffic over the wireless air interface. Although conceptually the communication of voice and data was not
a very different task, the major challenge was to deliver high quality and efficient data services. In case of
133
data traffic, latency is not an issue; it really doesnt matter while browsing a page that one receives the
text first and then the graphics. However, when using some state of the art packet data services like
video on demand, latency is a critical issue. Consumers would not accept technology that delivers
pictures first and then voice. These challenges were even bigger since all this transmission had to happen
over the wireless network.
There are various features that should be part of the MAC layer over the wireless physical layer to
provide reliable data communication service. These are header processing, signaling, payload processing,
and interworking with a packet data layer two functionality. One such example is address resolution
protocol (ARP) resolution when the IP data traffic has to be communicated over the Ethernet packet data
link layer, which exists over a wireless MAC. The core competency of any such implementation is
mapping of the signaling that exists in the wireless network to the signaling that exists in the packet data,
as well as mapping of the parameters across the two technologies.
Two such packet data technologies, IPv4 over 802.16 MAC and IPv4 over Ethernet (802.3) over 802.16
MAC, have been taken as case studies and their implementation in the real time systems is discussed in
the following section.
Technical Considerations of IPv4 over 802.16
As discussed in the sections above, one of the most important functions of any implementation used for
communicating IPv4 data over the 802.16 MAC should be the signaling management and the parameter
mapping. A management entity is thus required for keeping such mappings. The signaling, however, shall
be initiated by the applications existing on the IP stack based on when they generate the start and stop
requests for data transmission.
In the case of IPv4, a key factor for IP data forwarding and routing is the resolution of destination IP
address into its appropriate machine address. This machine address is then used for packet forwarding by
the underlying MAC. The ARP resolution does not make any sense in the case of 802.16. This is because
it is a point to multipoint system. In the uplink direction each SS knows beforehand that its data can be
destined to go anywhere in this world, but it will have to go through the BS. Therefore, it always has to
deliver data to its designated BS. In the downlink direction BS sends the data to SS by sending it on the
appropriate caller identification (CID), rather than the machine address which is a part of the payload. So
ARP resolution can either be straightaway bypassed or can be statically configured during startup. At
startup when SS enters the network, its IP address is either configured statically or is fetched from the
dynamic host configuration protocol (DHCP) server. The mapping of this IP address and the CIDs
corresponding to it is maintained at BS. Based on the destination IP address of the data and the
corresponding CID stored in the database at BS, the BS knows to which SS the data is destined. On the
other hand, 802.16a considers the fully meshed network and interSS communication. In this context,
ARP resolution becomes extremely important.
In the 802.16 networks the data is carried over time domain channels multiplexed over the uplink and the
downlink frequency. These channels are identified by a channel identification also termed as CID. There
are some predefined CIDs that are used for management message communication. For the data traffic this
CID is dynamically allocated by the BS. Allocation of these CIDs, along with connection setup and
management, comprise the signaling for a particular data connection. Based on the IPv4 parameters like
source IP address, destination IP address etc., a mapping is maintained between these parameters as well
as the CID. This mapping database also contains the entries based on their priority and quality of service
(QoS) parameters. The parameters mentioned here along with the ones specified in the IEEE 802.16
specification are collectively known as classifiers. The classifier/CID management is a part of the
management entity
134
In Figure 1, sample data illustrates that IP traffic originating from SS will terminate on BS.
Figure 1
When the DSARSP comes to central procession system (CPS) with the CID, a confirmation is sent to
the MAC CS. The MAC CS then places the CID in the database corresponding to that service classifier in
the management entity database. The management entity will then communicate this to MAC CPS. On
getting this confirmation, the MAC CPS sends back a DSAACK. This ensures that the CID is now
available in the SS database for further data communication.
For details on all the message types and the layered architecture of 802.16a refer to references, number one.
Technical Consideration of IPv4 over 802.3 over 802.16
For IPv4 over 802.3 i.e. Ethernet over 802.16 the main difference is that for the IP stack the MAC layer is
Ethernet, however the actual MAC layer is 802.16. Hence, the interworking should be such that the IP
stack is independent of the existence of 802.16. In other words, it is the Ethernet MAC layer that is
modified to exist on top of the 802.16 MAC. Since IP stack is independent of the underlying 802.16
MAC, the link layer functionality of ARP resolution becomes of prime importance. In the case of pointto-multipoint systems like 802.16 this can be statically configured during startup or during initiation of
the first connection. However, in the case of meshed networks as mentioned in 802.16a, this has to be
dynamically populated. This is so because the topology might keep on changing with multiple SSs
entering and leaving into the network.
135
In Figure 2, sample data illustrates that IP traffic over 802.3 Ethernet, originating from SS, will terminate
on BS.
Figure 2
This data flow scenario makes the assumption that matching classifiers and service flow exists in the
convergence sub-layer, either through provisioning or dynamic creation. At the IP level it means the ARP
resolution has already taken place. Incoming packet PDU from the user layer, in this case IP over
Ethernet/802.3, gets matched according to particular criteria, as per the classification process at the CS
layer. The packet is then delivered to the CPS for forward delivery using the connection defined by the
CID to the peer entity, in this case, BS.
For details on all the message types and the layered architecture of 802.16a refer to references, number one.
Challenges
There are various challenges involved in the architecture and design of such packet data systems
mentioned above. These challenges include the following:
136
Design of the interworking or mapping database and the points of its access
Synchronization of this database with the events in the real time
o How will the database on the BS side be flushed when an SS is removed from the
network?
How will the ARP entries be flushed in a meshed network when an SS is removed from
the network?
IP stack implementations across various OS and its compatibility with the existing 802.16
o
Another very important area, one that should be targeted at the time of design, is how these systems will
be tested.
A sample testing approach is explained in detail in the following.
Sample Testing Approach
For the purpose of IP traffic generation any existing IP stack implementation can be used. IP applications
like ping can used to verify the data flows over the existing MAC. A suggested testing strategy is
described in the sections below. This approach can be used over any OS platform with an IP stack
Two personal computers (PCs) can be connected over Ethernet back to back or via LAN. One PC can act as
the IP traffic generator and will have a normal IP over Ethernet stack and the 802.16 packet encapsulation
stub. The second PC can act as the host machine into which the 802.16 MAC implementation will be
plugged and tested.
For generating traffic IP applications like ping, telnet can be used. The technique of plugging the 802.16
in the OS kernel for testing has been suggested, as this is conceptually an elegant way of generating
Ethernet frames with IP payload.
Setup for Receive Packet Processing
When an Ethernet frame arrives on the prototype host, i.e. PC with stack under test, it is delivered to the
kernel through the device drivers, which have been attached to the kernel by opening up device files.
These frames are typically received by the Ethernet entry point function of the kernel. The packet is
processed, its Ethernet header is detached and the packet is sent to 802.16 stack entry point. After
processing, the stack passes frame to the upper layers. In the case of IPv4 over Ethernet over 802.16
scenario, the frame is again sent to the Ethernet layer which will eventually send it to IP stack.
Setup for Send Packet Processing
On receiving an IP based Ethernet frame from upper layers, 802.16 stack on the prototype host (i.e. PC
with stack under test), delivers the frame to Ethernet layer again so that it can be sent to the peer machine.
The peer machine has 802.16 receive stub, which analyzes the packet and establishes the connection by
forwarding it to upper layers on that machine.
Conclusion
Based on the above discussion it can be concluded that providing IP data services over the wireless air
interface should not be the main objective of any wireless access system. The objective should be to
provide these services in such a way that they are highly efficient, high speed, and minimally latent. This
discussion can be carried further to apply to other packet data services like IP over point to point protocol
(PPP) over Ethernet over 802.16 and IPv6.
The broadband wireless access systems based on 802.16 face competition from wide area cellular second
and a half and third generation (2.5G/3G) technologies. In addition, the IEEE 802.20 standard, defined for
wireless WAN access, could potentially compete with WiMAX since both standards could be seen as
separate solutions to a similar problem.
137
One major factor by which these wireless technologies can win over each other is their time to market.
However, their co-existence with each other still remains a debatable question.
References
IEEE Std. 802.162001 Part 16: Air interface for fixed broadband wireless access system,
http://standards.ieee.org/getieee802/download/802.16-2001.pdf
IEEE Std. 802.16a2003 Part 16: Air interface for fixed broadband wireless access systems Amendment
2, http://standards.ieee.org/getieee802/download/802.16a-2003.pdf
IEEE Std. 802.16e03/07r1 2 Part 16: Air Interface for Broadband Wireless Access Systems
Amendment 4: Mobility Enhancementshttp://ieee802.org/16/tge/docs/80216e-03_07r1.pdf
IEEE 802.20 Mobile Broadband Wireless Accesshttp://grouper.ieee.org/groups/802/20/
138
WiMAX
Overcoming the Limitations of Todays
Fixed Wired Access Technologies
Kathy Burrows
Siemens AG, ICN
Lutz Fielbrandt
Alvarion GmbH
Executive Summary
Governments globally are starting to prioritize broadband as a key political objective for all citizens to
overcome the broadband gap also known as the digital divide. In last mile markets where traditional
cable or copper/fiber infrastructures are either saturated, outdated or simply out of reach, broadband
wireless access (BWA) technology fills the void admirably, providing highly efficient and cost effective
access services for a large number of subscribers who would otherwise be left out of the loop in
developed markets.
With the advent of WiMAX, BWA is undergoing a dramatic change. What differentiates WiMAX from
earlier BWA developments is standardization. Current broadband wireless deployments are based on
proprietary solutions in which each BWA vendor custom-builds their solution, which adds time and cost
to the process. Similar to what has happened recently in the wireless local area network (WLAN) arena
with wireless fidelity (Wi-Fi), WiMAX plans to enforce standards compliance among vendor members.
This compliance will result in interoperability and ultimately plug-and-play products, the cost of which
will benefit from economies of scale and hence bring dramatic improvement to the business case for the
operator. First WiMAX products will be available by the beginning of 2005 and is set to become the
mainstream broadband wireless platform with more than 50% market share used by the predicted 3.8
million broadband wireless subscribers in 2008. While the overall number of subscriber lines is quite
small relative to DSL or cable, the dollar value is growing to the point where even major carriers are
beginning to pay close attention.
It is not only the developed markets that can benefit from WiMAX. For emerging markets, operators
are interested in using WiMAX for low cost voice transport and delivery, which has been very
difficult with proprietary solutions. Overall, the markets without any fixed infrastructure pose the
greatest opportunities.
2006 will see the start of the second stage in the WiMAX evolution with WiMAX chipsets embedded
in laptops and other mobile devices. This step will lead to broadband portability and to CPEless
business model, which makes the case even more compelling for an operator, because the user is
subsidizing the model.
139
WiMAX
All of these reasons together lead to the fact that the fixed broadband access business case was not
compelling. This is changing.
1.2 The Wireless (R)Evolution and WiMAX
Figure 1 illustrates the improvement and advancement of broadband wireless technologies from 2000 to
2005 as well as the evolution from proprietary to standard-based solutions.
2000
2001 2002
2003 2004
Proprietary
Solutions
2005
Standard-based
Solutions
140
Internationally, WiMAX has agreed to cooperate with European Telecommunications Standards Institute
(ETSI) to support certification of products employing the ETSI high performance radio metropolitan area
network (HIPERMAN) standard for European broadband wireless metropolitan area access.
1.2.2 Volume Production
Major chip vendors, such as Intel, are developing standards-based WiMAX chipsets, (which can also later
be integrated into laptop computers), and are leading this revolution towards volume production which
will in turn help to reduce equipment prices ensuring that wireless products can be competitively
positioned against their wired counterparts. WiMAX promises to drive down the costs of broadband
wireless equipment, allowing operators to expand networks and provide lower access fees.
1.2.3 Overcoming the Technical Limitations of Todays BWA
The WiMAX frequency ranges can be licensed or license-exempt and enable non-line-of-sight (NLOS)
performance, making the IEEE 802.16a standard the appropriate technology for carrier-grade last-mile
applications where obstacles like trees and buildings are often present and where base stations may need
to be unobtrusively mounted on the roofs of homes or buildings rather than towers on mountains.
1.2.4 Portability and CPE Elimination
A further enhancement (the second stage) of the standard is the IEEE 802.16e extension, which enables
nomadic capabilities for laptops and other mobile devices allowing users to benefit from metro area
portability of an xDSLlike service. This extension will boost development of built-in chipsets thus
eliminating the external modem altogether and allowing transmission directly to the laptop. This built-in
CPE could lead to a CPEless business model, which makes the case even more compelling for an
operator, because the user is subsidizing the model.
Subscribers (000s)
5000
4000
3000
2000
1000
0
2002
2003
2004
2005
2006
2007
2008
141
WiMAX
The 802.16 standard is set to revolutionize the broadband wireless market with research showing that by
2008 up to 50% of all broadband wireless equipment could support this standard.
Large BWA vendors are concentrating on developing WiMAX equipment because they also see
many benefits from migrating to WiMAX, such as exiting the CPE business and concentrating on
their core business: base stations.
Operators can improve their business model by deploying interoperable, lower-cost equipment,
which is particularly important in emerging markets.
Siemens believes that the BWA market will have a substantial share of the broadband access market and
that WiMAX is going to change the broadband map globally.
2. WiMAX Equipment and Networks
2.1 Application Scenarios
Typical point to multipoint BWA systems are composed of two key elements: base station and subscriber
equipment. The base station connects to the network backbone and uses an outdoor antenna to send and
receive high-speed data and voice to subscriber equipment, thereby eliminating the need for extensive and
expensive wireline infrastructure and providing highly flexible and cost-effective last-mile solutions.
142
The ability to overcome the physical limitations with ranges of traditional wired infrastructure
and still provide residential and business users with comparable throughputs at up to 40km
Broadband service provision in areas where existing plant is not allowing copper based xDSL
based services
Cost-efficient service supply in areas where traditional xDSL is not suitable due to small number
of customers per DSLAM
The avoidance of steep installation costsno outside plant costs necessary for copper/fiber
The ability to quickly provision service, even in areas that are hard for wired infrastructure to
reach, helping operators to overcome the digital divide
Broad bandwidthup to 134 megabytes per second (Mbps) in 28 megahertz (MHz) channel
Multiple services supported simultaneously with full quality of service (QoS) to efficiently
transport multifaceted protocols like internet protocol version four (IPv4), IPv6, asynchronous
transfer mode (ATM), Ethernet, etc.
Bandwidth on demand (frame by frame)
Media access control (MAC) designed for efficient used of spectrum
Comprehensive, modern, and extensible security
Multiple frequency allocation support from 211 GHzorthogonal frequency division
multiplexing (OFDM) and OFDMA for non-line-of-sight applications (licensed and licenseexempt spectrum)
Time division duplex (TDD) and frequency division duplex (FDD)
Link adaptation: Adaptive modulation and codingsubscriber by subscriber, burst by burst,
uplink and downlink
Point-to-multipoint topology, with mesh extensions
Adaptive antennas support and space-time coding
Conformance test specifications as defined in IEEE 802.16d
Mobility extensions (IEEE 802.16)
143
WiMAX
Integrated antenna since as further integration into silicon by major chip suppliers takes hold;
CPEs can be integrated into laptops, phones, and other devices
144
Kelson L. Colbo
Division of Administration
InfoMine of the Rockies, Inc.
Abstract
Rural America, with its large open spaces between its population centers, creates an environment not well
suited for the kinds of telecommunications investment seen in other, more densely populated areas, where
broadband access arises in almost all forms: digital subscriber line (DSL), cable, and wireless, to name
just a few. Yet the very population that can benefit the most from broadband wireless technology
investment is, in fact, rural America. These sparsely populated areas are struggling with a severe
transformation from their humble origins as mostly an agrarian lifestyle to that of a knowledge-based
economy. There is no shortage of individuals wanting to relocate to rural America, to live and work in
settings resplendent with 360 degree mountain views, open rolling vistas, and lakes as clear and deep as
anyone can imagine. But how can this economic development take place? How can this transformation
happen when there is seemingly no business case for making the technological investment in these areas?
Ultimately then, a business case must emerge; one that brings together the right technology investment to
meet the unique needs and struggles of rural America. Such companies are unlikely to be incumbent
players, as the change in business model and the risk of such an investment would never see the light of
their boardrooms. Instead, smaller emerging telecommunications companies tightly tied and already
invested in the rural communities they serve, will likely adopt such a business model and take the risk to
find the capital to implement the model and report to the wider telecommunications community their
findings. Many such projects are well underway; however, only just now are the wireless technologies
that will be central to the adoption and success of such business models becoming visible. The
technologies built around the WiMAX standards will play a pivotal role in these emerging business
models, and this paper provides a case study of such a model that is underway in rural Montana.
Introduction
A business that operates just outside of Butte, Montana in Galen (a small, historically farming
community) is currently provided voice, data and video services by three different providersall of
145
which, of course, are overlay networks riding on the aging, over-burdened public switched infrastructure
in the subscribers rural area which hasnt seen any public switched telephone network (PSTN)
investment in decades. They are totally exasperated with their providers. They can no longer deal with the
poor quality, frequent outages, long response times, lack of any real service level agreement and worst of
all, high price for the poor services they are receiving. They beg and plead for help from a new provider
to offer services for less cost, with a real service level agreement, and at speeds that provide the quality
they desire because they hope that a new network can enforce an end-to-end quality of service model.
How to tell this extremely frustrated customer that a new provider can meet every requirement and
expectation they have, and can do it for less than they have historically been paying? How to get away
from having to make use of the same poorly maintained public network? It is possible for the company
who has created an end-to-end wireless network that needs only connect to the PSTN for the purposes of
exchange with other carriers.
This network is inexpensive to operate as there are no ongoing costs to lease overlay networks from the
incumbent carriers nor is there any cost to operate the infrastructure itself, other than maintenance and
provisioning, once the capital costs of the infrastructure have been paid. The nature of a single network
owned and operated by this company makes the provisioning, servicing, and expanding of the network
easier and less expensive as there is no need to interact with other providers or carriers to accomplish
these tasks. Generally speaking, this means that provisioning procedures, and other operations support
system (OSS) duty cycles, can be streamlined, allowing response to customers to be exceptional as
compared to other providers. Further, the end-to-end network being operated solely by this company
means the technology required to implement service level agreements (SLA) can be put in place;
ultimately providing customers with the levels of service they desire to accomplish their work.
If all of this sounds like a dream, it rather is at this point. However, there are examples of this kind of
provider popping up on the radar, such as the much lauded TowerStream and Airspan. What they all seem
to have in common is their infrastructure is either entirely wireless (end-to-end) or is mostly wireless,
with backhaul services being provided by a technology partner that is not an incumbent carrier of any
kind. In addition, these new business models are not being brought to bear in the market place by the
incumbent carriers, rather, they are relatively new start-up companies1 that have developed a business
model and are in the process of implementing the model to varying degrees of success. More to the point,
they seem to be finding a niche market that has historically been overlooked; the T1 to 100-Mbps
broadband space.
However, what is missing from the radar screen are any blips of providers operating these new kinds of
wireless networks in rural areas of America such as Montana, as suggested in the above example. This is
extremely unfortunate as the population in rural America stands to benefit the most from these newer
wireless broadband technologies, such as those built on the emerging WiMAX standards2.
What Is WiMAX?
WiMAX is actually a set of Institute for Electrical and Electronics Engineers (IEEE) standards (802.16(a
e)) that, unlike the 802.11 set of IEEE standards associated with the wireless fidelity (Wi-Fi) movement,
have been engineered to be compatible with their equivalent European standards. This North American
and European compatibility is central in motivating the manufacturers to adopt the 802.16 set of WiMAX
standards and will likely play a pivotal market driver in forcing lower the cost of chip sets and resulting
customer premises equipment (CPE) hardware, adding to quick adoption. However, unlike the adoption
rates associated with the Wi-Fi 802.11 set of IEEE standards, the WiMAX standards are more likely to
1
146
have adoption rates similar to broadband DSL and cable3. This is primarily due to the incremental
adoption of the full set of 802.16 standards. Table 1 shows the differences between the different IEEE
802.16 addenda.
IEEE
Standard
802.16
802.16(a)
802.16(bc)
802.16(d)
802.16(e)
Description
Fixed Wireless; 1066 GHz; Line-of-Sight Requirement;
Fixed Wireless; 211 GHz; Non-Line-of-Sight Requirement; 70
Mbps; 31-Mile Range
Fixed Wireless; 1066 GHz; Interoperability with Protocols and
Test-Suite Structures
Fixed Wireless; 211 GHz; NLOS; 70 Mbps; 31-Mile Range;
Ratifies 802.16(c) and Forms the Base Standard for Developing
Access Points
Mobile Wireless; < 6 GHz; NLOS; 15 Mbps; 3-Mile Range
Standard
Adoption
April-2002
April-2003
January-2003
July-2004 (est.)
3rd-QTR-2005
(est.)
*All data extracted from the IEEE Working Group Web site; adoption dates are IEEE publication dates.
WiMax starting to make its move by Stephen Lawson, Network World, 06/07/2004
147
access and suggests that access to and use of ICT should make it possible for people to use technology
effectively, as a means to improve their lives. As such, they have produced the following qualifying criteria:
With telecom companies unable to build a business case for investing in rural America and providing the
same broadband access to telecommunications services that can be found in more densely populated areas
of the country, much of rural America is simply on the wrong side of the digital divide, unable to have the
kind of access to ICT envisioned by bridge.org. The consequence to this is not unlike any other disparity
in what has become a vital socio-economic element of these communities economic structure. Small
rural communities are seen struggling to adopt new knowledge-based industries and markets simply
because the access to the required ICT is missing, being instead invested in other communities in larger
service areas where the return on investment can be more readily computed and returned more quickly.
This is true of communities in Montana, for example, that were once thriving city centers of commerce
within the state and the country; such as Butte, Montana, home of the Copper Kings a vibrant economy
of the past. Recently, Butte, Montana was the corporate seat of Touch America, a high-technology
company that sought to be the carrier to the carriers, like so many during the emergence of the tech
bubble. Unfortunately, even when Touch America was operating in the community of Butte, very little
ICT investment was made in Butte or the state of Montana. Most of Touch Americas investments were in
large backhaul fiber networks designed to link carriers in the eastern third of the country to those in the
western third of the country, leaving Montana communities with aging infrastructure unable to support
the broadband capacities required by todays ICT; streaming video, voice over IP, real-time processing,
large data-base operations, and the like.
Economic development companies (EDCs) within each Montana community struggle to attract hightechnology, or other clean industries to the area due primarily to the lack of viable infrastructure and
the cost associated with accessing the services required to do business in the new global economy.
Access to high-speed (broadband) telecommunications that leave the state and connect to other carriers
is still exceedingly expensive as are telecommunications links between cities within the state. Montana
is such a large geographic state, that it contains two local access and transport areas (LATA) within its
borders. Therefore, to get a single T1 communications link from Bozeman, Montana to Helena,
148
Montanaa mere 95 milesrequires an IXC to bridge the LATA; driving costs as high as $1,250
$1,780 per month depending on the interexchange carrier (IXC) used to bridge the LATA. Montana,
however, is not alone. The same limitations of investment in ICT can be seen throughout almost all
rural American communities.
EDCs have a formidable job to be sure and are actively seeking strategies to build the necessary ICT
infrastructure that will lead to a much easier development of technology industry within their
communities. However, EDCs alone are usually not in a position to rally enough of the political will and
action to make investments in the current PSTN elements possible; access, under the bridge.org
definition, remains denied. EDCs will ultimately have to nurture not the incumbents, rather the new
technology innovators that are starting to emerge; who are assuming the risk of adopting new
technologies in hopes the adoption cycle is measured in decades instead of years.
A Difficult Recipe
The way toward getting a viable infrastructure that begins to bridge the gap of the digital divide is to
assemble a wide array of community players who are aligned to the same goal of ICT access for all in
their community. These individuals frequently include as many of the following entities as possible.
Entity
New ICT Companies
Incumbent Carriers
Local and State Government
Economic Development Companies
Chambers of Commerce
University Systems and Colleges
School Districts and Community Institutions
(Libraries)
Hospitals and Health Care Providers
Financial and Banking Institutions
Large Enterprise Businesses (Regional/National
Chains)
High-Technology Small Businesses
Role
Will assume the risk and provide the innovation
Necessary for linking to other carriers and
networks
Provide tax and other incentives and help with
regulation
Establishes links to out-of-state business
opportunities
Represents small businesses within the community
Provides the necessary training opportunities for
use of ICT services and a resume pool for
businesses
A historically large consumer of ICT services and
lays a foundation for training opportunities for
ICT companies
A historically large consumer of ICT services
A historically large consumer of ICT services and
a potential source of seed funding (with other
guarantees)
A historically large consumer of ICT services and
a potential investor in the ICT infrastructure
Consumers and innovators of ICT services
149
A Case Study
The goal of this project is to create an infrastructure that provides broadband access to a wide array of
application domains while providing extremely high reliability (optional to the end user) and end-to-end
quality of service capability; allowing for service level agreements. The infrastructure must connect
subscribers in three extremely rural cities in southwestern Montana using wireless broadband
technologies based on the WiMAX standards and also interconnect these cities together using WiMAX
backhaul equipment.
150
for point-to-point applications would be used in our case study to maximize the distance covered for
our city to city backhaul applications.
Each city in our case study will consist of at least two base stations, allowing for the possibility of two
fault tolerant links from any subscriber station; one to each base station. Each of the base stations will
have high-speed interconnect links, providing a well connected topology and varied route paths.
Application Domains
Before investigating the economics of the case study, it is important to review the various applications
that the case study network must support in order to be a viable option for community entities to
subscribe. No longer is it just possible to offer a faster, larger pipe as a means or incentive to have a
business switch their existing provider and move to the product being offered. In the broadband markets,
adoption of new service providers is driven primarily by the value added services that are provided (or
available) along with the access being provided. Very few entities will switch merely to have access to a
larger or faster access provider, but many will switch if various ICT services are provided on the network
that are not available through competitors.
Application Domain
Internet Access
Voice-overIP Services
Videoconferencing Services
Video-on-Demand Services
Medical Information
Management Services (ASP
Model)
Virtual Private Network (VPN)
Services
Enterprise Security
Management Services
Homeland Security Services
(Video Surveillance)
Description
Broadband (high-speed) access to the Internet at SLAs of T1 up to
multiple T3 speeds; residential competition at the < T1 market levels
will take place but accounts for < 5% of the financial analysis
performed in our case study.
These services include network-based inter-office PBX connectivity;
aggregated voice trunks for both local service, local-long distance
and long-distance; unified messaging services; toll-bypass services.
Real-time, two-way, high-quality ( 768 kbps) IPbased
videoconferencing from the desktop and boardroom.
These include VoD services to the hospitality markets as well as to
the local cable operators.
This is an ASP model designed to consolidate medical information
management in a single location (City-1) and provide ASP access to
this service to all health care practitioners in the three city case
study.
VPN services to link corporate offices within the three city case
study as well as outside of the case study cities.
Manage network security for the participating subscribers; mostly
large customers, such as Hospitals and the like.
Real-time, high-quality video surveillance services to City,
County and State Governments to protect municipal assets; this
same information can be provided to the Department of
Transportation for highway and weather conditions within the
three city case study.
One of the largest industries in rural America is that of tourism. The
ability to provide visitors to the region with information about the
region and other areas within the state is vital; this information can
be provided and updated daily if need be.
152
Estimates are provided for all application domain capital equipment costs, recurring costs and the
associated revenue streams. Again, the purpose of this case study is to show how a three city deployment
would work in a real world rural area.
Capital Costs
The tables that follow show the capital costs of establishing the WiMAX network in each of the three case
study cities and the backhaul element between the three cities. In addition, the capital costs for
establishing the Internet access and the voice over IP services are shown.
Item
1
2
3
4
5
Capital Expenditure
Wireless Base-Station Equipment
Wireless Backhault Services
Internet Access
Voice over IP Services
Total Capital Expenditure
City-1
Subscriber
Base
75
$
191
$
75
$
22
$
$
Cost
45,116.00
23,875.00
3,400.00
3,850.00
76,241.00
City-2
City-3
Subscriber
Subscriber
Base
Cost
Base
Cost
250
$ 77,758.00
250
$ 75,958.00
191
$ 23,875.00
191
$ 23,875.00
250
$
6,400.00
250
$
6,400.00
75
$ 13,125.00
100
$ 17,500.00
$ 121,158.00
$ 123,733.00
Item
1
2
3
4
5
6
City-1
Costs
900.00
1,083.33
1,666.67
1,000.00
900.00
5,550.00
$
$
$
$
$
$
Expenditure
$ 5,550.00
$ 5,550.00
$ 5,550.00
$ 16,650.00
$
$
$
$
$
$
City-2
Costs
900.00
1,083.33
1,666.67
1,000.00
900.00
5,550.00
$
$
$
$
$
$
City-3
Costs
900.00
1,083.33
1,666.67
1,000.00
900.00
5,550.00
Item
1
2
3
4
Description
Provisioning
Installation
Testing
Back-Office
$
$
$
$
Unit
20.00
150.00
75.00
20.00
Qty
575
575
575
575
Total
Cost
$ 11,500.00
$ 86,250.00
$ 43,125.00
$ 11,500.00
$ 152,375.00
154
Item
Recurring Monthly Revenue
1 Internet Access
2
Market T1
3
Market T1 to 5Mbps
4
Market 5Mbps to 100Mbps
5 Voice over IP Services
6
Total Monthly Revenue
City-1
City-2
Subscriber
Monthly
Subscriber
Monthly
Base
Revenue
Base
Revenue
75
250
5
$
449.75
38
$
3,418.10
67
$ 33,500.00
200
$ 100,000.00
3
$ 15,000.00
12
$ 60,000.00
22
$ 2,948.00
75
$ 10,050.00
$ 51,897.75
$ 173,468.10
City-3
Subscriber
Monthly
Base
Revenue
250
45
$ 4,047.75
192
$ 96,000.00
13
$ 65,000.00
100
$ 13,400.00
$ 178,447.75
Item
1
2
3
4
5
6
Case Analysis
Capital Costs
Per-Subscriber NRC Costs
Total Fixed Costs
Monthly Recurring Costs
Monthly Recurring Revenue
Monthly Recurring Gross Profit
$
$
$
$
$
$
City-1
(76,241.00)
(19,875.00)
(96,116.00)
(5,550.00)
51,897.75
46,347.75
City-2
City-3
$ (121,158.00) $ (123,733.00)
$ (66,250.00) $ (66,250.00)
$ (187,408.00) $ (189,983.00)
$
(5,550.00) $
(5,550.00)
$ 173,468.10 $ 178,447.75
$ 167,918.10 $ 172,897.75
Table 8: Case Study Analysis Showing Fixed Costs and Monthly P/L
Conclusions
This paper has examined the reasons why it is central to rural American communities to have increased
levels of investment in information and communications technologies (ICT), even as such investment is
not forthcoming due to the sparsely populated, geographically challenging, low median income of the
states that tend to contain much of rural America. Not surprisingly, there are few dots on anyones
broadband maps that show such states as Montana, Wyoming, Idaho, North and South Dakota, and the
like. Some states, even with significant rural populations, such as Kansas, have managed to develop
centers of high-technology that bring into the state clean high-tech businesses and add to the economic
development of the region. However, more remote and rural areas of even these states do not have
6
The monthly growth analysis detail can be provided to any interested reader by sending an e-mail to the author at
curtiss@imrockies.net and requesting this analysis.
155
adequate investments in ICT to rise to the occasion of closing the digital divide as defined by bridge.com;
the disparity in the telecommunications haves and have-nots remains formidable.
We have shared substantive excerpts from a case study currently underway here in the southwestern
region of Montana that is designed to analyze the feasibility of and establish a business case for
developing and deploying a three city end-to-end wireless broadband wireless network. Due to the
distance, speed, nonline-of-sight, and quality of services features of the WiMAX standard, this wireless
technology was selected as the basis for the wireless components of the network. Of course, there are
many other factors to consider, many raised in this paper, many not, but the bottom line is that new
business models are required to provide the necessary ICT infrastructure that will help rural America
transform into having a viable knowledge-based economy that will greatly benefit from the migration of
clean high-tech industry to these areas. Without such an investment, it remains unclear how these rural
communities will ever bridge the digital divide and fully participate in the ever growing knowledge-based
sectors of out national and global economy.
156
157
On the other hand, the quality of copper infrastructure in many developing countries may not be qualified
for high-speed digital subscriber line (DSL) access, and the situation is even worse in rural and
underserved areas where copper is either insufficient or households are so far from central offices.
Therefore, having a last-mile wireless access seems to be an ideal solution for those countries primarily
for rural areas and in some cases for municipalities, too. The problem in most of the developing nations is
that broadband services are still growing and due to the economies of such countries, service charges
constitute a major factor in penetration. With the growth in Internet usage, however, the aspiration for
broadband access in parts of the developing world is becoming more demanding and here where
broadband wireless may have a great opportunity. A recent report from Pyramid Research shows that the
annual growth in broadband wireless users in the developing countries will be 54 percent over the next
five years vis--vis 34 percent for the developed nations [1].
Nowadays, because the term broadband has become synonymous with high speed, we may first need to
identify what a broadband access is, or in other words to specify the speed at which a connection is
considered a broadband connection. As a matter of fact, the FCC has defined 200 kpbs as a minimum
speed for a broadband link [2], though this figure seems to be quite humble for some countries where
speeds below 1 Mbps are not seen as broadband. This shows that the term broadband is a bit elastic and
based on the applications and their development from time to time and from place to another, so what is
broadband today might not be broadband tomorrow, and if a 256 kpbs connection is a broadband in
Egypt, it might be a legacy service in South Korea. While talking about trends and with the rapid
development in telecom services and applications, however, it does not really make sense to presume that
speeds below 2 Mbps are broadband even for wireless access, especially as to date there are wireless
technologies that do support speeds at 2 Mbps and higher such as LMDS, MMDS, CDMA2000 1xEVDV, IEEE 802.11, and 802.16.
3. Broadband Wireless Revolution
Traditionally, the wireless last-mile access used to be known as FWA. But with the advent of applications
that are bandwidth thirsty such as audio and video downloads, the term broadband has been introduced
for both fixed and wireless access, and the latter has become BWA. Moreover, the mobile revolution has
also influenced the BWA market in the sense that mobile phones can have a BWA, too. Therefore, there
are today FWA services that are not necessarily broadband and BWA services that are not necessarily
fixed. Anyhow, in this paper the term BWA will be used to point out the wireless high-speed access for
both mobile and fixed services.
The telecom world has recently been undergoing new technological trends that may reshape the whole
industry in the near future. In the mobile arena, there have been some 3G success stories as mobile
operators have finally had their 3G services rolled out. At the same time, broadband access seems to be
the name of the game while applications as well as end users are both eager for high-speed access from
anywhere and at anytime. WLAN or more popularly Wi-Fi is another trend for broadband wireless access,
and there has been a lot of controversy about integrating Wi-Fi hotspots with 3G networks to get the best
of each technology. Now another broadband wireless contender is arriving, the WMAN or WiMAX,
which is seen as the natural evolution for Wi-Fi as higher access speeds are sought over further reach.
Compared to Wi-Fi, WiMAX challenges are much more difficult and the disputes around it are even more
furious because it arrives at the time when the whole world is talking wireless, and the fact that it
generated a lot of hype and promised very much of success before it is actually there.
It is quite evident that the demand for wireless access is growing while at the same time the boundaries
between fixed and mobile users are vanishing. In other words, wireless technologies are evolving today in
a sense that the whole world is becoming mobile, and the need for a ubiquitous high-speed access is
essential for everyone around. This ideology was at the core of 3G since the early days of its
158
Baher Esmat
development, and today following the boom of Wi-Fi hot spots along with the propaganda surrounding
the emerging WiMAX technology it should come at no surprise that a major shift in the wireless
landscape is predicted over the next few years. How is this going to happen and when? What is the role of
WiMAX in this revolution? This is what this paper will try to answer.
4. IEEE 802.16: WiMAX
The IEEE 802.16 Working Group was founded in 1999 to expand the work of the 802.11 and develop a
global standard for BWA. So, while the 802.11 is the universal WLAN standard, the 802.16 is its
extension toward the metropolitan size of networks known as WMAN.
4.1 IEEE 802.16 Standards Overview
The first release of the standard named 802.16 was approved in December 2001, working within high
frequency bands between 10 - 66 GHz, and therefore supporting only Line-of-Sight (LOS) operations.
Significant amendments were made to 802.16s physical (PHY) and medium access control (MAC) layers
in order to support operation at lower frequencies within 211 GHz and consequently allow for nonlineof-sight (NLOS) deployments. This modified release was eventually the 802.16a standard that was
approved in January 2003 [3, 4]. In fact, 802.16a has gained a lot of popularity over the past 18 months
but because it was merely focused on fixed broadband access, it only supports outdoor CPEs, thus there
was an enhanced version called 802.16 RevD, which addressed the indoor CPE and was just ratified in
June 2004.
Although both 802.16a and 802.16 RevD support speeds up to 75 Mbps, this is actually the maximum
data rate shared among the users of one cell, provided that 20 MHz channel is utilized. If less channel
size is available, which is most likely the case, the data rate will decrease accordingly. On the other
hand, as the current WiMAX standards claim to reach distances up to 50 km, this requires optimum
conditions that can realistically be unfeasible. Practically speaking, and according to WiMAX
equipment manufacturers, NLOS operation would afford distances around 5 km, while with clear LOS
this figure may go up to 15 km.
The next coming standard is 802.16e, which will support mobility so that the end user can move with his
laptop, personal digital assistant (PDA), or any mobile device while maintaining high-speed access all the
time. The 802.16e which is not expected to be finalized before the second half of 2005 is in fact
overlapping with the mandate of the IEEE 802.20 Working Group that is concerned with wireless wide
area networks as a migration for the current mobile systems (2.5G and 3G) toward what may be called 4G
mobile [5].
4.2 WiMAX Forum
The WiMAX Forum is a nonprofit organization established in 2003 aiming to promote and certify BWA
products that comply with both the IEEE 802.16 and the ETSI HyperMAN standards in addition to
interoperability between different vendors products. Members of the WiMAX Forum represent more than
100 equipment vendors and chip manufacturers as well as telecom operators. The forum has been
working on defining procedures for interoperability conformance and soon will start conducting the tests
and certify products that should be available by early next year [4]. The role of the WiMAX Forum with
802.16 is equivalent to that of the Wi-Fi Alliance with 802.11. Throughout this document and for
simplicity, we will refer to any of the 802.16 variants as WiMAX, while keeping in mind that the current
standard specifications are within the fixed wireless access boundaries, whereas the future enhancements
are supposedly to provide full mobility.
Because BWA products have never had a standard regarding the frequency bands used, the WiMAX
Forum believes that one of the driving forces for its technology is to harmonize the worldwide spectrum
159
of WiMAX. So, as part of the certification process, the Forum will ensure that the first generation of
certified products to support the unlicensed 5 GHz as well as the licensed 3.5 GHz. Also, the licensed 2.5
GHz is highly considered by the Forum, though may take a little bit of time to materialize. On a longer
term lower-frequency bands such as the 700 MHzused by TV broadcastcould be an attractive
candidate to WiMAX because they have good penetration characteristics. Moreover, the WiMAX Forum
has recently formed a regulatory working group (RWG) to address governments from all over the world
in order to get feedback on their countries spectrum regulations for broadband. Setting up the RWG is an
attempt in the same direction to streamline efforts towards unifying the WiMAX frequency bands
worldwide so that the technology can fly, penetrate and maintain sustainability over time [4].
5. WiMAX Propaganda
There is no doubt that momentum is growing in the broadband wireless arena. The advent of the
Internet and its rapid development over the past decade has let everyone eager to be always on.
Almost over the same period of time, the mobile industry was soaring and demand for mobility has
become as important as being always connected to the Net. Blending both Internet and mobile has
caused a tremendous shift in long-standing telecommunication paradigms and led to the ever-increasing
thrust in broadband wireless that we live through nowadays. Needless to say that the growth in
applications has been going in line with the advancements in communication technologies to the extent
that it is uncertain whether communication protocols and standards are driving applications, or visaversa, or perhaps they are evolving in such a harmonized manner as part of a new converged world in
which the main investments are put in infrastructures (communication platforms) whereas the majority
of revenue streams come from services (applications).
Having said that, and observing what have been said about WiMAX over the past 18 months, there are
many questions that WiMAX has posed and need to be answered in order to realize what this technology
is about. Is WiMAX a new technology? Why is it gaining this popularity? What makes it different from
previous BWA technologies? Will it rule the broadband world? When is this expected to take place? To
answer these questions, we have to go few years back and recall some incidents as we presently may be
more able to analyze them.
5.1 Is WiMAX New?
As mentioned earlier, BWA is not new especially when we talk about fixed wireless, it has been there for
many years, and there have been many technologies providing this kind of services. The problem with
those technologies has always been lack of standardization, and even standard ones have never
guaranteed interoperability. It was evident that as the demand for BWA goes up, the call for a global
standard that allows this market to grow increases too. This fact was realized by the IEEE and thats why
they developed 802.11 and because it had limitations in its reach, the 802.16 Working Group was formed
and WiMAX came out to life. Realistically, WiMAX as a technology was more or less there, all what
IEEE did was the adoption of several BWA technologies in order to standardize the physical as well as the
MAC interfaces, hence offering better and more widespread services. Because the IEEE is only concerned
with the development of the standard rather than the way vendors implement it into their products, the
WiMAX Forum was established to develop the conformance and interoperability tests and certify
products accordingly. So, WiMAX is not new, it is actually the base standard for many proprietary
technologies with the assurance that various vendors equipment do interoperate in order to achieve wider
dissemination and meet the growing demand of the BWA market.
5.2 BWA: Limitations and Shortcomings
Looking few years back, there have been many wireless technologies that promised much more than they
did actually deliver, even those who did, never made it on time. For FWA, although LMDS has gained
some edge in parts of the world, it has never made a global impact. Mobile Internet on the other hand was
160
Baher Esmat
the main pitch that third-generation mobile networks have always promised to deliver, but has 3G come
on time, or has its roadmap ever been clear? Even the road towards Wi-Fi has never been straightforward
since the early development of WLAN systems in late 80s till Wi-Fi was eventually booming couple of
years ago. So, lets look at each of these examples with a bit more of a focus to see how WiMAX has
learned from them.
LMDS has been approached by many operators and service providers in North America and other
scattered spots around the world, but it has never made a real breakthrough in the BWA market
worldwide. The technology by itself seemed to be ideal not only for rural areas where there is not plenty
of wired infrastructure, but also for downtowns where demand for high-speed last-mile access is
ascending while digging through the roads is impossible. Except for mobile operators that exploited
LMDS for backhauling, large service deployments are no more than a handful and accordingly the
shipments of base stations and CPEs have drastically declined over the past couple of years. The main
reason for that was the lack of standardization, which led to high equipment prices, as each vendor used
to build his own proprietary platform, and operators consequently had to put more investments so that
their business cases have appeared to be more unfeasible.
On the mobile side, the amount of argument 3G has made among the industry analysts, and the challenges
it brought to operators were unique. The standard bodies were established in the early 90s and vendors
together with operators have been active since then to come up with the first releases of standards by end
of the last decade. At the same time, the telecom regulators started to put in place the licensing framework
so that many licenses were actually issued between years 1999 and 2001. Nevertheless, and with the
exception of Japan and South Korea, it was not until second half of 2002 when 3G networks have
commercially been kicked-off. Although the reasons why 3G services rollout have been delayed are
complex and may differ from one country to another, the truth is that it was not due to the lack of
standards. Without going into too much of details, we can briefly say that operators burdened by 3G
license fees while not gaining yet any return on general packet radio service (GPRS) (2.5G) services,
doubts surrounding the availability of 3G handsets with reasonable prices, visibility of content and new
applications that would stand for a viable business case and most of all the industry downturn experienced
at that time, were all to be blamed for this delay [6].
Even the recent propaganda of Wi-Fi did not happen overnight, nor was it on the roadmap of the IEEE
802.11 Working Group when it was set up back in 1987. It took IEEE 12 years till 1999 when the
802.11bthe standard that is well known to everyone todaywas released. During the 90s there were
tens of proprietary wireless local area network (WLAN) products that were mostly used on the enterprise
or the campus level to peer between two end-points instead of laying fiber between them [7]. In late 90s
as the Internet started to change and twist a lot of concepts, and with the increase of laptops, it was
evident that WLAN was gaining more weight and endorsement from the whole industry. As a result the
standards have come through: 802.11a/b (1999), 802.11g (2003), and 802.11n is yet on the pipeline and
not expected to be ratified before 2005 [5]. Also, the Wi-Fi Alliance was established in order to set the
procedures and tests for interoperability, and certify products. So obviously, WLAN or Wi-Fi has passed
through a rough road probably with some disappointments along the way, and as things changed and
mistakes were resolved, hot spots are today the name of the game in BWA.
5.3 Learning Lessons
So what has WiMAX learned from those previous examples? Being standardized by IEEE and with the
support of giant telecom providers and manufacturers, WiMAX products will shortly be available offering
solutions for last-mile high-speed access. Although the first generation of equipment will be a bit
expensive, which may sound like another LMDS version, the fact that WiMAX has emerged with a
standard as well as a body that looks after interoperability issues, in addition to the unprecedented support
from the largest microprocessor builder (Intel), all this can eventually give WiMAX good economy of
161
scales and drive the costs down. At the same time, WiMAX is not another next generation network. In
other words, it is not like 3G in the sense that there are new generations of applications and advanced
handsets so that the absence of one would make the other useless and the shortage of both would lead to
delay in service launching. That is exactly what happened with 3G when the standards were ready; the
new generation of handsets and applications were yet not commercially matured. The beauty of WiMAX
is its ability to complement and integrate with other technologies. It is on the last mile to extend the fiber
metro to end users without the need to lay more cabling infrastructure. It can also help mobile operators
and wireless Internet service providers (WISPs) expand their networks coverage leveraging the high
capacity it can provide in backhauling. It is also doubtless that WiMAX has benefited a lot from the WiFis success, and that is why it gained much of its hype. There was a sincere effort to avoid the drawbacks
of Wi-Fi at the beginnings of WiMAX.
This is mainly what makes WiMAX different from other previous technologies and gives it this
popularity. We yet need to see if WiMAX is going to rule the BWA market and when this is possibly to
happen. This is what the next section aims to answer.
6. Who Are the WiMAX Players?
This section intends to look at three various categories of operators and service providers and see the
different possible scenarios of incorporating WiMAX into their networks. The discussion below illustrates
that almost all service providers can benefit from WiMAX as it is more integrating with current
technologies more than competing with them. In addition, the section provides a brief perception about
the future of the mobile world in which several BWA technologies are collaborating to provide
ubiquitous, high-speed services to end users.
6.1 Fixed-Line Operators
Today, WiMAX is a new candidate in the last-mile business, offering superior QoS and much higher
speeds than those provided by other BWA technologies. To date, with its ability to actually reach
distances between 515 km offering shared bandwidth of 75 Mbps, it seems to be a superlative fixed
wireless access technology. WiMAX, however, is yet not the cheapest technology as prices of the first
generation of CPEs, which is due to arrive in Q1 2005 will be around $350, and the second generation of
CPEs expected to be available in Q3 2005 will be priced around $200. But as WiMAX services spread
out, volume of products increases so that prices will continue to decline and should eventually fade away
when Intel incorporates the WiMAX chipset into its processor [8]. More importantly, because WiMAX
supports QoS with variable bit rates [9], it can carry and deliver any real time applications, so it is not
only data and Internet, it can actually deliver VoIP and on-demand video. There will also be commercial
WiMAX CPEs that provide E1/T1 interfaces for integrated services digital network (ISDN) and legacy
voice services.
Another reason why WiMAX should be appealing to fixed operators is the fact that the capital investment
for a WiMAX infrastructure is significantly less than copper and cable infrastructures. Consequently, in
the developed countries where last-mile copper and coax are mostly saturated particularly in urban areas,
fixed operators may think of confining their expansions in building new infrastructures and enlarge their
WiMAX coverage instead. It may even be more compelling for the developing nations to overcome their
lifelong problems of last-mile by capitalizing on a standard-based BWA technology like WiMAX.
Once WiMAX starts to pick up, however, fixed-line operators will have to revisit some of their business
models. In other words, although it is predicted that services like DSL, cable, and fiber-to-the-home are
not going to disappear, the potential generated by WiMAX is raising many questions about pricing
pressure it may bring to those services.
162
Baher Esmat
both Wi-Fi and/or WiMAX together with 3G, so that while the mobile user is indoor he will be on the WiFi mode and once he gets outside he can switch to WiMAX and/or 3G. This is not science fiction,
research and development is underway and as 802.16e is expected to be ratified by 2006, it is inevitable
that commercial products shall take another couple of years to 2008 till they see the light. On the other
hand, previous experience with 2G and 3G has shown that developing standards for each generation takes
around 10 years, and then there is another 10 years for the technology to be commercially matured [10].
Having this concept in mind, the development of the new 4G air interface that will be able to detect and
deal with multitechnologies such as 3G, Wi-Fi, and WiMAX, will take approximately 10 years. So, we are
basically talking about 2015 till 4G becomes a reality.
The above analysis shows that 3G operators should not be anxious about their investments. It is at least
another 10 years till they need to migrate, during which they can benefit from Wi-Fi and WiMAX in order
to boost their data services. Also, the same analysis demonstrates that future mobile networks are going to
be broadband, flexible, scalable, and built around multitechnologies. So there is not a single BWA
technology that is going to rule, and WiMAX among other technologies will shape the world of BWA in
the coming future.
7. Conclusion
Interoperable BWA equipment has always been a dream for operators and service providers, and with the
arrival of WiMAX this dream has finally come true. Thanks to a standard that is extraordinarily backed by
giant industry leaders aiming to draw the new map of the wireless world. Like any other technology,
however, WiMAX is not going to happen overnight and though products will be available next year as
outdoor CPEs, the real WiMAX breakthrough will not happen till the chipset is incorporated in laptops
and, the industry hopes, other portable devices, so that prices are pushed down, and this pragmatically
makes us talk about 2008.
At the same time, the integration between WiMAX and other technologies like Wi-Fi and 3G is legitimate
and should be beneficial to the different parties involved: fixed operators, mobile operators, wireless ISPs
as well as the end customer. Again, this will take some time to materialize because not only many players
will be involved but also regulatory and business aspects should carefully be taken into account.
The general theme that WiMAX will solely rule the broadband market is not to a great extent realistic.
The most pragmatic vision is that 3G, Wi-Fi, and WiMAX will integrate together while providing
seamless roaming so that to build the future mobile world known today as 4G. We must bear in mind that
this development is yet in its infancy stages and is unlikely to see the light of day before 1015 years in
the future, in addition to the fact that everyday there are new evolving technologies that generate
innovative services along with irregular business models. Accordingly, the panorama that anyone would
envision today for this 4G world may dramatically be changing over time.
8. References
[1]
A. Reinhardt, The Next Big Thing for Wireless?, Business Week Online, January, 2004.
[2]
S. Ismail and I. Wu, Broadband Internet Access in OECD Countries: A Comprehensive Analysis,
A Staff Report of the Office of Strategic Planning and Policy Analysis and International Bureau,
October, 2003.
[3]
[4]
164
Baher Esmat
[5]
M. Paolini and S. Fili, Wi-Fi, WiMAX and 802.20 The Disruptive Potential of Wireless
Broadband, BWCS Ltd., UK, 2004.
[6]
K. Wieland, The Long Road to 3G, Telecommunications International, pp. 1620, February, 2003.
[7]
T. S. Rappaport, Wireless Communications Principles and Practice, Pearson Education Inc., India,
2002.
[8]
R. Resnick, The WiMAX Bridge to Broadband, Telephony, pp. 1819, May, 2004.
[9]
WiMAX Forum White Paper, IEEE 802.16a Standard and WiMAX Igniting Broadband Wireless
Access.
[10] J. L. Hurel, C. Lerouge, C. Evci, and L. Gui, Mobile Network Evolution: From 3G Onwards,
Alcatel Telecommunications Review, Q1 2004.
165
167
WiMAX, NLOS, and Broadband Wireless Access (Sub-11Ghz) Worldwide Market Analysis 20042008
Fiber: Theres not enough fiber; it still only reaches a fraction of the users demanding largebandwidth, high-reliability connections; fiber also requires costly additional electronic equipment
on both sides.
xDSL: Efficiency is limited by distance to the central office and the quality of copper.
Cable modems: The bandwidth is shared by a number of subscribers and the service deteriorates
as more residents go online. In addition, the installed base of cables passes through residential
areas and not through business regions.
In rural and underserved markets, these wireline choices are simply not available as an option. In current
commercial deployments, broadband wireless networks can deliver more bandwidth than traditional copper
cables and are cheaper and faster to deploy. Historically, many operators worldwide have used broadband
wireless technologies (namely point-to-point [PTP] radios) as a proven, service-provider class method of
connecting long-haul networks. PTP technologies have also been used for access in isolated cases with
mixed results. With point-to-multipoint (PMP) BWA, service providers will be able to provide broadband
services over large geographic areas with greater flexibility and improved economies of scale.
What Is WiMAX?
Broadband wireless access technology has not proliferated as was expected by proponents. Systems were
based on proprietary technologies, tying service providers to just one equipment vendor. The effect of this
is that no individual system was able to gain share and reduce production costs.
168
Adlane Fellah
In early 2003, the Institute of Electrical and Electronics Engineers (IEEE), responsible for setting global
communications standards, approved the 802.16 air interface for fixed broadband wireless access systems
(wireless MAN) interface communications protocol, which uses the 2 to 11 MHz frequencies.
The practicality of the standard, however, was limited by the fact that there were neither test
specifications nor conformance statements established yet. That is why in order to ensure interoperability
between vendors competing in the same market, the Worldwide Interoperability for Microwave Access
(WiMAX) technical working groups were created by the leaders in IEEE 802.16 technology.
Issuing a WiMAX Certified label will serve as a seal of approval that a particular vendors system or
component fully corresponds to the technological specifications set forth by the new wireless MAN
protocol. That will in turn reduce customers confusion.
The whole concept around standardization is to reduce equipment and component costs through
integration and economies of scale that will, in turn, allow for mass production and hence less expensive
equipment. In particular, current chipsets are custom built for each BWA vendor making equipment
development and manufacturing both costly and time consuming.
With large volumes, chipsets could sell for as little as $25 and other BWA components could benefit from
these mass volumes as well. We expect the cost reduction impact to be mostly on the CPE in the first two
years of WiMAX deployments as base station costs are more complex to deal with despite the promise of
base station cost of under $20,000. Base stations, however, are less of a factor in the operator economic
equation for deployments.
The other notable WiMAX benefit will be to reduce customers confusion and the advent of a WiMAX
compliance label. Service providers are becoming familiar with WiMAX and include it in their review of
equipment suppliers. That trend will become more important when WiMAX compliant equipment is
deployed in real networks and delivers on its promises. WiMAX, however, will not necessarily bring
higher performance systems in the short term as quite a few current proprietary systems are already
delivering on WiMAX coverage, cost, and performance promises.
Beyond lower cost and compliant equipment, service providers need better coverage to make money.
They need the ability to deliver access without truck roll to all their potential customers around the base
station regardless of natural and other obstacles (trees, buildings, etc.). In fact, people in the industry often
state that their customers would like to have a triple 50: equipment providing 50 Mbps up to 50km
radius with $50 CPEs!
WiMAX Roadmap
The availability of 802.16-compliant equipment on the field will be the result of a sequence of events
which includes the following:
1)
2)
3)
4)
5)
6)
169
WiMAX, NLOS, and Broadband Wireless Access (Sub-11Ghz) Worldwide Market Analysis 20042008
Adlane Fellah
and Cisco (which backs Flarion) are aiming at influencing heavily the standard and are clashing in
standard battles with Intel and Nokia in the race for delivering components to mobile broadband devices.
Solution Vendor Trends
There are both WiMAX want to be as well as proprietary solution vendors. There is also a great variety
of products in the market. Indeed some systems only offer data services including voice over IP (VoIP);
others also provide E1 and plain old telephone service (POTS). Some systems are plug-and-play, others
are outdoor only. Some offer 2 Mbps at the CPE while others, 512 kbps, etc.
Alvarion, the market leader with 25 percent market share across frequencies, may not have the highest
performance system in the market, but it beats every competitor according to several important business
metrics such as customer base, installed base, revenues and financial position. It is followed by
SRTelecom with 12 percent and Proxim with 9 percent.
Some notable subtleties exist when the market share position is analyzed by frequency market. Already
12 vendors offer a 3.5 GHz product and 4 more players will offer a 3.5 GHz product in 2004, which will
render that band market even more competitive.
Among plug and play, NLOS, portable systems, IP wireless is the leader in shipments and revenues,
followed by a small group of companies that include Navini, NextNet Wireless, or SRTelecom (Angel). It
is however difficult to subsegment the whole market on system capabilities (see Table 1).
Vendor
Airspan
Alvarion
Aperto
Axxcelera
BeamReach
Cambridge Broadband
Harris Corporation
Motorola
Navini Networks
NextNet Wireless
Proxim
Redline Communications
Solectek
SRTelecom
Trango Broadband
Vyyo
Wi-LAN
Waverider
ZTE
Others
Total
Market Share
7%
25%
6%
1%
1%
3%
3%
5%
3%
2%
9%
3%
1%
12%
1%
1%
3%
3%
2%
9%
100%
Notes: Market shares were estimated from shipments of base stations and CPEs from vendors. Discrepancies can occur
from noninclusion of services revenues, which are not provided by all vendors. We have tried to include only broadband
PMP products as per our definition of broadband specified in section 2 of this report in order to compare apples with
apples. Figures do not include narrowband wireless local-loop products such as Swing or Proximity. Market shares are
based on revenues from manufacturing shipments rather than final sales value through distributors.
Table 1: BWA Market Share per Vendor for All Frequencies < 11 GHz in 2003
171
WiMAX, NLOS, and Broadband Wireless Access (Sub-11Ghz) Worldwide Market Analysis 20042008
There is a continued industry consolidation. Chinese vendors such as ZTE will take more space in the
BWA market. We also see a clear, growing interest by major infrastructure suppliers in the 802.16
technology for use with their mobile and PTP radios platforms. 802.16 compliant vendors will have to
differentiate themselves from one another once the standard is in place and becomes widely adopted. Will
WiMAX be dominated by nonstartups such as Intel and Nokia (for the base station) like it happened with
Wi-Fi? Specialization and the ability to integrate the product into a turnkey solution will be the key to
success. Expect partnerships with infrastructure suppliers to flourish. Vendors of fixed CPEs will need to
adapt to a market where the line between fixed and mobility applications is increasingly blurred.
Chipset Vendor Trends
The whole industry is benefiting from the late entry into the market of Intel, which is behind most of the
publicity around WiMAX. Intel has signed partnerships with Aperto, Alvarion, and Airspan (among
others), which together already hold more than a 40-percent BWA market share. But Intel did not enter
the game simply to address the fixed BWA market. It is betting heavily on the migration of chipsets into
the millions of mobile devices.
Intel is sometimes perceived as arrogant toward certain solution vendors, as the silicon maker seems to
be pushing its own roadmap towards the integration of chipsets into laptops while the integration of
802.16 at the base station does not seem to be its priority. Other chipset makers such as Wavesat have
more experience in developing OFDM chipsets but only the future will determine if know-how will
triumph over reputation, size and clout. Other serious 802.16chipset contenders include Fujitsu in
partnership with Wi-Lan.
Market Forecasts
Currently, single carrier PHY layer dominates deployments with about 50 percent of modems shipped
worldwide. Maravedis, however, forecasts that both single carrier and CDMA systems will lose market
share, as OFDM/802.16d becomes the widely adopted standard for air interface of BWA systems. Various
flavors of CDMA should however remain strong in niche markets for mobility (see Figures 2 and 3).
Wimax Penetration Rate
60%
50%
40%
30%
20%
10%
0%
2004
2005
2006
2007
2008
172
Adlane Fellah
2004
2005
2006
2007
2008
Proprietary Systems
173
175
result, the WiMAX Forum has been working primarily with component and equipment suppliers to
develop base stations and premises equipment that carriers will use to deliver the service.
The market view of WiMAX has also been confused by the range of applications for which it has been
proposed. According to Margaret LeBrecque, marketing manager for the broadband wireless division at Intel
Capital and former president of the WiMAX Forum, three major phases in development are anticipated:
Phase 1Fixed-location private-line services or hot-spot backhaul: The initial application for
WiMAX type technology is a service that provides traditional dedicated lines at transmission
rates up to 100 Mbps using outdoor antennas. These systems typically use radio equipment that
predates the WiMAX standards. Companies like TowerStream offer wireless Internet access at
speeds ranging from fractional T1 to 100 Mbps (see TowerStream Delivers).
Recognizing the proliferation of hot spots, WiMAX is also being positioned as a means of aggregating
that traffic and backhauling it to a central, high-capacity Internet connection. Equipment suppliers
have also found a market for these point-to-point systems internationally, where they are used for
cellular backhaul or to deliver basic telephone service in hard-to-reach areas.
Phase 2Broadband wireless access/wireless DSL: The first mass-market application for
WiMAX would be broadband wireless access or wireless DSL, offering data rates between
512 kbps and 1 Mbps. The key will be to deliver low-cost, indoor, user-installable premises
devices that will not have to be aligned with the base stationthe antenna in the premises
equipment would be integrated with the radio modem. In the late 1990s, Sprint and MCI
pioneered this type of service, deploying point-to-point systems in about a dozen markets. They
subsequently shelved the idea while waiting for a functional nonline-of-sight radio technology
like the one described in the WiMAX standards.
Currently dozens of small-scale BWA services are cropping up around the country using
prestandard WiMAX technology. The Web site BBWExchange.com lists the top 10 wirelessaccess suppliers in the United States, the largest being DTN Speed of Omaha, with 5,100
subscribers as of April 2004.
The WiMAX Forum hopes to see that figure grow exponentially when larger carriers begin
deploying networks using low-cost ( $200), silicon-based products early next year. Currently
Verizon, Bell South, Nextel, and Earthlink are all testing BWA services.
176
Michael F. Finneran
Wi-Fi
Hot Spot
Home User
Nomadic
User
177
to address the requirements of carriers around the world. The original version of the 802.16 standard,
released in December 2001, addressed systems operating in the 1066 GHz frequency band. Those highfrequency systems require line-of-sight (LOS) to the base station, which increases cost and limits the
customer base. Further, in LOS systems, customer antennas must be realigned when a new cell is added
to the network.
We will focus primarily on the 802.16a standard released in January 2003 that describes systems
operating between 2 GHz and 11 GHz. The lower frequency bands support nonline-of-sight (NLOS),
eliminating the need to align the customer unit with the base station (see Table 1).
802.11
Maximum
Bit Rate
2 Mbps
802.11b
11 Mbps
802.11a
54 Mbps
802.11g
54 Mbps
Standard
Fallback
Rates
1 Mbps
5.5 Mbps
2 Mbps
1 Mbps
48 Mbps
36 Mbps
24 Mbps
18 Mbps
12 Mbps
9 Mbps
6 Mbps
Same as 802.11a
Channels
Provided
3
Frequency Band
Radio Technique
2.4 GHz
FHSS or DSSS
2.4 GHz
DSSS
12
5 GHz
OFDM
2.4 GHz
OFDM
Table 1: IEEE 802.11 Wireless Local Area Network (WLAN) Radio-Link Interfaces
WiMAX Frequency Bands
Where all Wi-Fi implementations use unlicensed frequency bands, WiMAX can operate in either licensed
or unlicensed spectrum. Within 802.16as 211 GHz range, four bands are particularly attractive:
Licensed 2.5-GHz MMDS: In the United States, the FCC has allocated 200 MHz of licensed radio
spectrum between 2.52.7 GHz for multichannel multipoint distribution service (MMDS). Sprint
and MCI used this band for their original point-to-point services.
Licensed 3.5-GHz Band: A swath of licensed spectrum roughly equal to MMDS has been
allocated in the 3.4 to 3.7-GHz range throughout most of the rest of the world.
Unlicensed 3.5-GHz Band: In the United States, the FCC has recently moved to open an
additional 50 MHz of unlicensed spectrum in the 3.653.70 GHz band for fixed location wireless
services.
Unlicensed 5 GHz UNII Band: In the United States, 555 MHz of unlicensed frequency has been
allocated in the 5.1505.350 GHz and 5.4705.825 GHz bands. That spectrum is called the
unlicensed national information infrastructure (UNII) band, the same band used for 802.11a
wireless LANs. The allocation was increased from 300 MHz to 555 MHz by an FCC order in
November 2003.
Michael F. Finneran
802.11b ( 11 Mbps): The 802.11b radio link uses a direct sequence spread spectrum technique
called complementary coded keying (CCK). The bit stream is processed with a special coding and
then modulated using quadrature phase shift keying (QPSK).
802.11a and g ( 54 Mbps): The 802.11a and g systems use 64-channel orthogonal frequency
division multiplexing (OFDM). In an OFDM modulation system, the available radio band is
divided into a number of subchannels, and some of the bits are sent on each. The transmitter
encodes the bit streams on the 64 subcarriers using binary phase shift keying (BPSK), QPSK, or
one of two levels of quadrature amplitude modulation (16- or 64-QAM). Some of the transmitted
information is redundant, so the receiver does not have to receive all of the subcarriers to
reconstruct the information.
The original 802.11 specifications also included an option for frequency hopping spread spectrum
(FHSS), but that has largely been abandoned.
802.16 Radio Modulation
The 802.16a standards define three main options for the radio link:
The first wave of products to hit the market will use the 256-subcarrier OFDM option. As a result, the
WiMAX Forum is initially developing test suites and interoperable test plans for that option initially. It
also conforms to the ETSI HIPERMAN standard.
Channel Bandwidth
The Wi-Fi standards define a fixed channel bandwidth of 25 MHz for 802.11b and 20 MHz for either
802.11a or g networks. In WiMAX, the channel bandwidths are adjustable from 1.25 MHz to 20 MHz.
That will be particularly important for carriers operating in licensed spectrum. The transmission rate of
that channel will be determined by the signal modulation that is used.
179
Uplink
Mandatory
Mandatory
Mandatory
Optional
Optional
Downlink
Optional
Mandatory
Mandatory
Mandatory
Optional
FEC Coding
1/2, 3/4
1/2, 2/3, 3/4, 5/6, and 7/8
1/2, 3/4
2/3, 5/6
3/4, 7/8
Bits/Symbol
1/2, 3/4
1, 4/3, 3/2, 5/3, and 7/4
2, 3
4, 5
6, 7
180
Michael F. Finneran
time coding feature allows the use of two transmit antennas at the base station and a single subscriber unit
antenna that can combine the two signal images. Longer term, the working group envisions use of
multiple input-multiple output (MIMO) systems to improve overall range and transmission rates (see
Table 3).
Spectrum
Configuration
Bit Rate
Modulation
Mobility
Channel
Bandwidth
Typical
Radius
Cell
802.16
1066 GHz
LOS
32 to 134 Mbps (28 MHz
Channel)
QPSK, 16 QAM, 64
QAM
Fixed
802.16a
211 GHz
NLOS
13 miles
35 miles
802.16e
< 6 GHz
NLOS
Up to 15
Mbps
Same
as
802.16a
75 MPH
5
MHz
(Planned)
13 miles
181
frequency (FDD). As with a cable-modem system, outbound transmissions are broadcast in addressed
frames, and each station picks off those frames addressed to it.
WiMAX Channel Access
In WiMAX networks, access to the inbound channel will be controlled by the base station. Users wishing
to transmit inbound must first send requests on a contention-based access channel. Exclusive permission
to use the inbound traffic channel is then allocated by the base station using a system of transmission
grants. As only one station is given permission to send at one time, there are no inbound collisions.
Request/Grant QoS Capability
The major benefit of WiMAXs request/grant protocol is that it supports QoS. As inbound access is
controlled by the base station, WiMAXs access mechanism can support four primary types of service.
Those connection types can be set up dynamically:
Unsolicited grant real time: Consistent delay (i.e., isochronous) service for real-time voice and
video, where a station is allocated inbound transmission capacity on a scheduled basis.
Real-time polling: Another real-time service that operates like the 802.11 point control function
(PCF), where the base station polls each user device in turn.
Variable bit rate non-real-time: Variable-delay data service with capacity guarantees akin to
frame relays committed information rate for high-priority commercial users.
Variable bit rate best-effort: An Internet protocol (IP)like best effort data service for residential
Internet users.
The grant mechanism specifies that the entire capacity of the inbound channel can be allocated to one user
for a set time period. There is also a unique inbound allocation mechanism for OFDM channels where
multiple simultaneous user transmissions (up to 16) can be supported by allocating different subchannels
to different users.
Wi-Fi QoS (802.11e)
There are plans to incorporate QoS capabilities in Wi-Fi with the adoption of the IEEE 802.11e standard.
The 802.11e standard will include two operating modes, either of which can be used to improve service
for voice:
WME
The WME option uses a protocol called enhanced distributed control access (EDCA), which is an
enhanced version of the DCF defined in the original 802.11 MAC. The enhanced part is that EDCA will
define eight levels of access priority to the shared wireless channel. Like the original DCF, the EDCA
access is a contention-based protocol that employs a set of waiting intervals and back-off timers designed
to avoid collisions. With DCF, however, all stations use the same values and hence have the same priority
for transmitting on the channel. With EDCA, each of the different access priorities is assigned a different
range of waiting intervals and back-off counters. Transmissions with higher access priority are assigned
shorter intervals. The standard also includes a packet-bursting mode that allows an access point or a
mobile station to reserve the channel and send three- to five-packets in sequence.
182
Michael F. Finneran
Wired equivalent privacy (WEP): An RC4based 40- or 104bit encryption with a static key
Wi-Fi protected access (WPA): A new standard from the Wi-Fi Alliance that uses the 40- or 104
bit WEP key, but changes the key on each packet to thwart key-crackers. That changing key
functionality is called the temporal key integrity protocol (TKIP).
IEEE 802.11i/WPA2: The IEEE is finalizing the 802.11i standard, which will be based on a far
more robust encryption technique called the advanced encryption standard. The Wi-Fi Alliance
will designate products that comply with the 802.11i standard as WPA2. Implementing 802.11i,
however, will typically require a hardware upgrade, so while the standard should be completed in
mid2004, it might be some time before it is widely deployed.
WiMAX Encryption
Given that it was designed for public network applications, virtually all WiMAX transmissions will be
encrypted. The initial specification calls for 168bit digital encryption standard (3DES), the same
encryption used on most secure tunnel virtual private networks (VPNs). There are plans to incorporate the
advanced encryption standard (AES). As a result, we anticipate none of the security concerns that plagued
early Wi-Fi implementations.
Mobile WiMAX
One last option is the 802.16e specification for mobile WiMAX, which is due out later this year. While
the details are still being worked out, that standard will describe a mobile capability with hand-offs for
users moving between cells. The basic requirement is that it be backward compatible with the fixed
location service. One of the imperatives will be to reduce the power requirements for battery-powered
mobile stations. The plan is to support data rates up to 500 kbps, essentially equivalent to the highest
speed cellular offerings (e.g., Verizon Wireless 1xEV-DO service).
Wi-Fi Roaming
The IEEE has begun development of a roaming standard for Wi-Fi, though the specification is not expected
until 2005 or 2006. In the meantime, WLAN switch vendors like Cisco, Aruba, and Airespace have developed
183
their own proprietary hand-off protocols. We have seen similar capabilities in the Wi-Fi mesh products. That
means that providing a hand-off capability, however, requires implementing a vendor proprietary solution.
Table 4 compares the major attributes of the WiMAX and Wi-Fi technologies.
Primary
Application
Frequency
Band
Channel
Bandwidth
Half/Full Duplex
Radio Technology
Bandwidth Efficiency
Modulation
FEC
Encryption
Access Protocol
Best Effort
Data Priority
Consistent Delay
Mobility
Mesh
WiMAX
(802.16a)
Wi-Fi
(802.11b)
Wi-Fi
(802.11a/g)
Wireless LAN
Wireless LAN
25 MHz
20 MHz
Half
Half
OFDM
(64 channels)
2.7 bps/Hz
BPSK, QPSK,
16, 64QAM
Licensed/unlicensed
2 G to 11 GHz
Adjustable
1.25 M to 20 MHz
Full
OFDM
(256 channels)
5 bps/Hz
BPSK, QPSK,
16, 64, 256 QAM
Convolutional code
Reed-Solomon
Mandatory3DES
OptionalAES
Request/Grant
Yes
Yes
Yes
Mobile WiMAX
(802.16e)
Yes
Convolutional Code
OptionalRC4
(AES in 802.11i)
CSMA/CA
Yes
802.11e WME
802.11e WSM
OptionalRC4
(AES in 802.11i)
CSMA/CA
Yes
802.11e WME
802.11e WSM
In development
In development
Vendor proprietary
Vendor proprietary
184
Wireless ISPs (WISPs): One surprising business that grew out of Wi-Fi was the WISP. This is the
idea of selling an Internet access service using wireless LAN technology and a shared Internet
connection in a public location designated a hot spot. T-Mobile and Wayport are currently the
largest operators. While the proliferation of hot spots has been widely reported, no one seems to
be able to make any money at this. There are two fundamental obstacles, one technical and one
business oriented.
Michael F. Finneran
From a technical standpoint, access to the service is limited based on the transmission range of
the WLAN technology. You have to be in the hot spot (i.e., within 100m of the access point) to
use it. From a business standpoint, users either subscribe to a particular carriers service for a
monthly fee or access the service on a demand basis at a fee per hour. While the monthly fee
basis is most cost effective, there are few intercarrier access arrangements so you have to be in a
hot spot operated by your carrier in order to access your service. Some are now predicting that the
real business model will not be fee-based services, but a free service that is offered by the
property owner to attract customers. Thats not a business plan, its a charity!
City-wide mesh networks: To address the limited range, vendors like Mesh Networks and Tropos
Networks have developed mesh network capabilities using Wi-Fis radio technology. The idea of
a radio mesh network is that messages can be relayed through a number of access points to a
central network control station. These networks can typically support mobility as connections are
handed off from access point to access point as the mobile station moves.
Some municipalities are using Wi-Fi mesh networks to support public safety applications (i.e.,
terminals in police cruisers) and to provide Internet access to the community (i.e. the city-wide
hot spot). The mesh technology and hand-off capability, however, are not within the scope of the
Wi-Fi standards, and so it is vendor proprietary; that means you must purchase all of the
equipment from the same manufacturer. In the final analysis, we are cobbling together a set of
wireless LANs to do the job for which WiMAX was designed.
Whither WiMAX?
The market forecast for WiMAX is not clear at this point. Clearly, the major target will be broadband
wireless access or wireless DSL, though carriers must first choose to deploy the service. Their success
will depend on the cost and functionality of their offerings when compared to other broadband access
alternatives like DSL and cable modems. When chip manufacturers like Intel begin delivering WiMAX
compatible chipsets in late 2004, we will have the possibility of consumer devices costing $100 or less.
The carriers, however, will have to invest in the base station equipment, and they must decide if there is
sufficient demand and an adequate business case to justify the investment needed to deliver a broadband
wireless access service.
Three Potential Markets
As we noted at the outset, there are three potential markets for WiMAXprivate line, broadband wireless
access, or mobile service. Lets take a brief look at each of these.
Point-to-Point Systems
Point-to-point systems for delivering basic telephone service, hot spot, or cellular base-station backhaul
should continue to be a viable, carrier-oriented market niche. This is particularly true in lesser-developed
countries that lack a wired infrastructure. In the United States, TowerStream is planning an aggressive
build out of its wireless Internet access service in major markets. They will have to compete with much
higher-capacity fiber access alternatives from the incumbent local exchange carriers (ILECs) and
competitive local exchange carriers (CLECs). Deploying wireless access to a customers building,
however, should be faster and cheaper than providing fiber access. Carriers like WinStar and Teligent
failed in that wireless local loop segment in the late 1990s, but the redundancy built into TowerStreams
service is clearly superior to those first-generation offerings.
Broadband Wireless Access/Wireless DSL
Broadband wireless or wireless DSL offers the greatest near-term potential, but it also faces the greatest
competition. A late arriver in the market, WiMAXbased systems will have to compete with entrenched
cable modem and DSL services that are available to roughly 80 percent of U.S. households.
185
Carlton ONeal, vice president of marketing for base-station maker Alvarion, sees opportunities both in
migrating dial-up subscribers and extending broadband access to unserved communities. ONeal notes
that because only 20 percent of U.S. households currently subscribe to broadband access, the battle is
just beginning.
Further, a wireless solution should have a significant cost advantage in reaching the 20 percent of
households where broadband access is currently unavailable. Extending cable modem and particularly
DSL to those thinly populated areas will increase the cost per subscriber, and with a lower-income
population, the take rates will likely be less than in urban areas. With chip-level WiMAX certified
components, manufacturers will be able to deliver low-cost, user-installable, indoor stations that can
mimic the cable modem/DSL experience.
The advantage of a wireless solution has not been lost on the DSL carriers. Verizon has been testing
BWA on licensed frequencies in Herndon and Centerville, Virginia, using equipment from BeamReach
Networks. To minimize the cost, they are installing the antennas on existing cell towers. BellSouth has
begun trials using equipment from Navini Networks in Palatka, Port Orange, and Holly Hill, Florida, to
assess cost and technical viability. BellSouth has also been testing a first-generation wireless broadband
technology in Houma, Louisiana, since 2000. Nextel has announced its Nextel Broadband wireless access
in the Raleigh, Durham, North Carolina, area using MMDS spectrum it acquired from MCI. In March
2004 wireless pioneer Craig McCaw acquired Texas-based BWA carrier Clearwire Holdings. In June
McCaw announced plans to turn up WiMAXbased BWA service in 20 markets by the end of 2005.
Mobile WiMAX
A mobile WiMAX services could produce a real dust-up, however. Intel has been the primary backer for
WiMAX, and hopes to repeat the success it has had with Wi-Fi. Cisco and Motorola, however, are
backing a competing standard called Mobile-Fi (IEEE 802.20). Mobile-Fi proponents note that their
solution will be optimized for IP in high-speed mobile environments. While technology will be as
important, being first-to-market with an all-encompassing solution (i.e., at home and mobile) can be a
major advantage for WiMAX.
Mobile service can also change the picture for the cable modem and DSL carriers. They currently
dominate the fixed-location market, but they will have to develop service adjuncts to support users
outside of their homes. The free Wi-Fi capability that Verizon now offers its DSL customers is the first
such add-on, however it is only available in Manhattan. In the meantime, the cable companies are
pursuing joint marketing agreements with Wi-Fibased wireless ISPs to round out their offerings. A
combined home/mobile WiMAXbased offering will put the onus on cable modem and DSL suppliers
to provide an on-the-go capability or face the prospect of losing customers to a more flexible wireless
alternative.
The cellular carriers will likely come out on the short end of the data battle. Their 2.5/3G data offerings
have been only moderately successful; Verizon Wireless noted recently that only 3 percent of their
revenues came from data services. Further, those sales have been tied primarily to new consumer-oriented
applications like camera phones, short messages, and downloadable ringers rather than bread-and-butter
network access for commercial users. With higher-speed, data-oriented services coming on the market,
the cellular carriers will have a much tougher time winning over enterprise data buyers. Further, if
wireless voice over IP (VoIP) starts to catch on, the cellular companies might find themselves in a
defensive battle to hold on to their basic voice business.
186
Michael F. Finneran
Conclusion
WiMAX signals the arrival of the next wave of wireless data technologies. Unhampered by the short
range and data orientation of wireless LANs, these technologies hold the promise of taking high speed
wireless out of the coffee shop and out on the road.
The flexibility of the WiMAX technology gives it a significant advantage, addressing both fixed and
nomadic users, operating in licensed or unlicensed bands, providing both consistent- and variable-delay
services while operating in a carrier-scale environment. On paper, WiMAX looks like a strong contender.
Now we will have to see if the proponents can translate that technology into marketable services.
Copyright 2004 dBrn Associates, Inc.
187
A Business Justification:
WiMAX Service Providers
and Security Investments
Ralph P. Martins, Jr.
Senior Consultant
Booz Allen Hamilton
Graduate Student
George Washington University
Abstract
The broadband wireless access (BWA) market is on the verge of substantial growth. More specifically,
after becoming standardized, enjoying assistance from the U.S. federal government and being marketed
by well-known companies, a technology called WiMAX (also known as 802.16) looks almost certain to
become a significant part of the wireless network market. However, there have been instances in the past
where wireless technologiesa good example being 802.11x or wireless fidelity (Wi-Fi)were rushed to
the market without being implemented securely. If WiMAX were to suffer the same fate, rather than
putting individual networks at risk, the entire Internet infrastructure would potentially be put at risk by
increasing the vulnerabilities assumed by BWA service providers. After all, if people cant reach the
Internet in a relatively secure manner, whats the point? The whole reason for having an Internet is to
facilitate communication between people. Participation depends on security. While legislation and
government regulation would be one way to guide the BWA market in a more secure fashion, considering
President Bushs recently expressed wishes to use the U.S. government to encourage the expansion of the
BWA market, another method should be found. Thus, while external pressure may not be the answer,
internal incentives may. BWA service providers are private sector companies who have a bottom line like
any other profit-seeking company. By drawing a correlation between increased profits (or a lessening of
risk) and the implementation of information security, BWA service providers may just find a financial
incentive to keep their services (and our infrastructure) more secure. It is the purpose of this paper to
encourage security managers and executives who are employed by BWA service providers to take a stab
at calculating the return on security investment. It is not an exact science, but it can provide an individual
responsible for budgeting with the proper framework for making decisions as to how and when to spend
money on the critically important goal of securing the Internet.
Introduction
In March of 2004, U.S. president George W. Bush called for universal and affordable broadband Internet
access by 2007. He argued that there was a link between peoples access to information and their ability to
participate in the economy. In order to promote growth and access in the broadband service provider market,
189
he ordered the federal government to improve the process by which broadband service providers gain access
to federal land. Additionally, to help prevent any deterrents that may slow the growth in the broadband
market, he called for a continual ban on Internet taxes by Congress. Although real products have yet to hit
the market, standardization and capital investments by many companies are all but guaranteeing the flood of
new WiMAX products that will be available in the next few years. Predictions are being made before
services are even being offered yet. Some are even claiming1 that WiMAX subscribers will number more
than seven million by 2009. There are organizations, such as the nonprofit WiMAX Forum, that are
dedicated to encouraging and promoting the spread of wireless broadband networks. The WiMAX forum is
involved in such activities as marketing and certifying products as being 802.16x compliant. However, with
all this planning, is it not logical to be sure that adequate steps are taken to be sure new technologies such as
WiMAX can be implemented in a secure manner before they hit the market? It is known that WiMAX is not
as secure as it can be. It is also known that previous wireless technologies such as 802.11 were rushed to the
market before any consideration of security was made. While WiMAX appears to be a bit quicker to address
security issues than 802.11 was, it is safe to say that WiMAX products that have significant security flaws
are going to hit the market before they can be solved. It is also true that, with their air interface that can
broadcast for up to 30 miles, anyone who attaches a WiMAX device is assuming a certain amount of risk in
doing so. It is therefore logical for BWA service providers to take action and ensure that they are properly
investing in the security of their infrastructures before implementing WiMAX. This can be accomplished by
both awareness and incentives. When corporate executives are made aware of the implications of the
security risks of their technologies, they will certainly dedicate more resources into fixing them.
Furthermore, if a correlation between financial benefits and security investing can be made, the decision to
invest in capital can be a simple one.
What Is WiMAX?
WiMAX, short for Worldwide Interoperability for Microwave Access, is defined by a series of Institute
of Electronics and Electronic Engineers (IEEE) 802.16x standards and is the accepted specification for
wireless metropolitan area networks (MANs). In terms of standards, when the term WiMAX is used it
typically refers to 802.16REVd which incorporates and revises 802.16, 802.16a and 802.16c. WiMAX is
also a type of broadband transmission. The quality that distinguishes a broadband signal from a nonbroadband signal (such as wideband or narrowband) is its ability to carry several different transmission
channels at once. In the case of wireless broadband, it is capable of utilizing multiple frequencies and can
be segregated into separate, independent bandwidth channels. It is this quality that will allow a single
WiMAX device to service large numbers of customers simultaneously. WiMAX has a range of up to 31
miles, although in practice each base station is expected to cover an area with a radius of about ten miles.
It can transfer data at speeds of up to 75 megabits per second (Mbps), and is point-to-multipoint as
opposed to point-to-point, which is also referred to as a broadcast. This means each hotspot can support
thousands of users simultaneously. However, in reality a single pipe will be limited to about 500 users per
base station. That does not take into consideration the fact that channels can be combined to form an
aggregate signal, effectively multiplying the throughput several times. This makes a pipe with a
bandwidth of several hundred Mbps possible.
WiMAX may turn out to be more than just a broadband backbone though. Intel in particular believes WiMAX
will complement Wi-Fi and not compete with it. The company plans to create chipsets that can handle both the
802.11 and 802.16 technologies. Laptops with these chipsets would use whichever signal was strongest in a
given area. This does not necessarily make the two technologies competitors, as the scenario that most
envision is WiMAX being used to fill the gap between the wireless local-area network (WLAN) and the
Internet service. As can be seen in Figure 1, WiMAX closes the gap between the broadband wireless access
service providers network, which is connected to the Internet, and the subscriber.
1
http://www.smarthomeforum.com/start/show_news.asp?NID=300
190
Figure 1: How WiMAX Is Incorporates into the Existing Internet Service Provider Market
This gap between the subscriber and the service provider is sometimes called the last mile. This
term is especially important as it highlights why WiMAX has the potential to be so popular. Up until
wireless broadband, users had to have their Internet backbone pipe in via cables, using technologies
such as DSL or cable modems. While this is simply a matter of a small cost in some instances, there
are many cases where this does not make a lot of business sense or is just downright cost-prohibitive.
While most commercial and residential structures currently access the Internet using cables, less than
five per cent of all commercial buildings have broadband-capable fiber-optic connections readily
available. These can be extremely expensive to install, especially if it becomes necessary to dig up
concrete streets in order to drop the fiber. According to Anand Chandrasekher2, vice president and
general manager of the Mobile Platform Group at Intel, even for residential installations it can cost
$400 just to get a truck deployed to the install site, and then another 20 minutes to two hours to
complete the install. It is simply not worth the investment to install new, wired broadband
connections in many cases. Furthermore, potential customers in rural areas have had similar
problems getting broadband Internet access. The cost to run fiber from the urban areas where the
service providers backbone is located to a remote rural area is expensive. The fact that populations
are less dense in rural areas decreases the return on investment for running cable to rural areas.
However, since WiMAX can provide the same service without any new cable installs needed, it has
the potential to be very valuable to the previously mentioned individuals or organizations that would
otherwise not have access to broadband. Figure 2 is a very simple example of how, when natural
geography, distance, and a low concentration of subscribers would otherwise prohibit wired
broadband access, WiMAX would be the cost-effective solution to this problem.
http://www.computerworld.com/mobiletopics/mobile/story/0,10801,86093,00.html
191
http://www.oreillynet.com/pub/a/wireless/2002/04/19/security.html
http://www.metroactive.com/papers/metro/12.05.02/wifi-0249.html
5
http://www.theregister.co.uk/2004/07/02/intel_wimax/
4
192
potential costumers. Therefore, the fact that WiMAX does have the 31-mile range exposes it to more
potential attackers than does a Wi-Fi network. Along with this extension, there also is the fact that an
attacker does not need to be driving the streets in order to attack a WiMAX broadcast. He or she can be in
the safety of their own home, comfortable in the knowledge that if their system does not happen to be
configured with any identifying information (i.e. name, company, address), they can attack with the
knowledge that the chances that they can be identified or located are slim to none.
A typical broadband service provider who offers both wired and wireless WiMAX broadband access
has four distinct sources where attackers can enter their network (see Figure 3). These are from their
Internet pipe, the wired connections to their subscribers, the air interface, and from internal to the
corporate intranet.
http://wirelessman.org/tgd/contrib/C80216d-03_60r1.pdf
http://www.ieee802.org/16/mgt/contrib/S80216mgt-04_02.pdf
193
a security task group that would fall under the 802.16 working group. Based on the history of previous
standards, this will most likely result in the eventual creation of another 802.16 standard designed with
security. However in the meantime, there will be a need to secure a network that includes a WiMAX air
interface on it. Regardless of what the newer standards and enhancement of 802.16 bring, the fact is that
the products that will be on the market in the next few years will all be based on standards that already
exist today, that is, an insecure 802.16. There is sometimes a two or more year time lag between the
proposal of a standard and the emergence of products on the market that implement that standard. The
challenge will be encouraging middle and upper management of wireless broadband service providers to
take the initiative to make the appropriate investments in assuring the security of their networks.
The Business Value of Investing in Security for WiMAX Service Providers
How then does a broadband wireless access service provider justify the expenditure on the
implementation of hardware, software, and services dedicated to preserving the security of infrastructure?
One method is by showing the financial reasons for spending money, and this is through demonstrating a
return on security investment (ROI). It is not a simple task to calculate ROI security. There is no hard and
fast formula for defining a return on investment when it comes to spending capital on the implementation
of information security hardware, software and/or services. This is true throughout the information
technology field, and it is no less true when it comes to wireless broadband service providers. There are
some formulas that have to be considered when attempting to calculate a return on security investment,
one of which is as follows:
ROI = (revenue change + savings) / cost of investment
On the spectrum of private and public organizations, there are those who must invest heavily in security
as it is necessary for their very survival and there are those whose success is not quite as dependent on
securing their information assets. There is an argument to be made that just about every organization in
existence has some need for information security. However, an e-commerce company that relies on the
trust of its customers, such as on-line auction site E-Bay or the online store Amazon.com, certainly has
much more to lose in terms of reputation and customers than does a nonprofit philanthropic organization
that has a homepage. It is not a stretch to say that a wireless broadband access service provider has quite a
bit at stake when it comes to the negative publicity of an attack, not to mention the downtime and angry
customers who demand nothing less than constant, uninterrupted service. Consequently, it is in the best
interest of such a company to not only invest in the security of their company, but to do so only after
considerable thought and planning.
There are a variety of complications and obscurities associated with calculating a return on security
investment. Many of this is due to the lack of tangible information. First, not all costs or benefits can be
stated in the same currency. How does one measure the reputation of a brand name? And how would one
measure damage to such a name after the effects of a security incident have been realized? If a service
provider has a large number of customers that conduct banking transactions online, will they be
comfortable to continue doing so if they hear that their service provider frequently is penetrated by
attackers who wreak havoc on their systems? The difficulty in a situation such as this is that, in this
scenario, the service provider would have no accurate way of measuring how many customers
discontinued service after learning about a series of successful attacks. And even if there was a way to
measure this number, how would one determine which attack caused how many customers to flee to the
competition? And how would one place a monetary value on that? When considering currency, the other
consideration is the cost to prevent an incident. Fortunately, in some cases these costs can be a bit easier
to calculate. It is not impossible to calculate the cost of a firewall, the cost of the staff to provide
operations, maintenance and log analysis, and the cost of vendor support. While such costs may not be
able to be calculated with absolute precision, it is possible to take a very educated guess. While dealing
194
with solutions there can be a quite a bit of information readily available, but when dealing with incidents
there can be a confusing lack of quantifiable data.
This brings us to the second problem: lack of information. For obvious reasons, organizations that are
victims of incidents are reluctant to make such an event public. There are many good reasons not to, and
no good reasons to do so. With the lack of information, it is difficult to make predictions based on risk. In
order to calculate risk, we need to know both the frequency of an event and the severity of its occurrence.
A simple risk formula would look like this:
Risk = frequency of incident * severity of incident
However, how is one to know the frequency of an incident when data is never released? Who knows how
bad a virus can affect a companys infrastructure if each company hides their numbers? How is one to
determine the number of medical records stolen from a hospital database when the cover it up?
Once the risk can be determined, is comes down to a simple comparison. If the risk is greater than the
investment, spend the money. If it is less than the investment, do not spend the money. It is also necessary
to keep in mind the time value of money. The value of money today is less than the value of that same
amount of money used for other purposes until tomorrow.
What this paper will not provide is a cookie-cutter, cure-all solution for any wireless broadband service
provider to calculate their return on security investment down to the exact cent. There is no cut-and-dried
formula to measure the return on investing in security for WiMAX networks or any other security
investments for that matter. Due to the ambiguities associated with security incidents and investments that
would be impossible. It is however, the goal to introduce the reader to the issues involved with WiMAX
and provide him or her with a framework with which to think about how best to begin to arrive at such a
solution in a real-world scenario. After all, any capital expenditure begins with the understanding and
endorsement of the decision makers at the top of an organization. With some sort of understanding of the
financial side of security, a security manager will be armed with the weapons to justify the budget that he
or she needs.
Appendix: 802.16 Standards
The following is a brief summary of some of the IEEE 802.16x WiMAX standards:
802.16 The original standard, published in April 2002, which is titled Air interface for fixed
Broadband Wireless Access Systems is also known as wireless MAN. With this standard, nonline of sight is not possible. This specification was designed to standardize local multipoint
distribution services (LMDS).
802.16a This amendment, among other things, extends the spectrum range for transmissions
that makes non-line of sight communications possible. This is intended to make 802.16a a viable
alternative to DSL and cable modem for last mile connectivity services.
195
802.16e This standard provides mobility in that it allows portable clients to join the network
and transmit data while moving. This may compete with IEEE 802.20. This spec is not yet
finalized as of July 2004.
802.16REVd This standard combines, revises and updates 802.16, 802.16a and 802.16c.
Bibliography
196
http://www.computerworld.com/managementtopics/outsourcing/isptelecom/story/0,10801,91728,00
.html
http://www.computerworld.com/managementtopics/outsourcing/isptelecom/story/0,10801,91728,00
.html
http://www.oreillynet.com/pub/a/wireless/2002/04/19/security.html
http://www.fcc.gov/Daily_Releases/Daily_Business/2004/db0520/DOC-247411A1.pdf
http://www.cnp-wireless.com/wsp.html
http://www.netstumbler.com/
http://www.ieee802.org/16/mgt/contrib/S80216mgt-04_02.pdf
http://www.wi-fiplanet.com/columns/article.php/961181
http://www.mobilepipeline.com/specwatch/80216.jhtml
http://www.dailywireless.org/modules.php?name=News&file=article&sid=2741
http://www.giac.org/practical/GSEC/Julie_Schuller_GSEC.pdf
Richard Harris
Professor and Director
Centre for Advanced Technology in Telecommunications, RMIT University
Abstract
With the growing importance of the Internet to business, commerce, and personal communications,
there has been a substantial increase in demand for access to the Internet and increases in the speeds of
accessing information. Users are expecting services to be available anywhere and at anytime. Users are
not confined to city regions and there is a growing demand to rapidly roll out high-speed service access
to remote and rural areas. Delivery of fixed network resources via fiber cables to remote, difficult-toaccess, and rural subscribers is expensive, and alternative approaches using fixed wireless broadband
networks are now an attractive and feasible option. A new Institute of Electrical and Electronics
Engineers (IEEE) based standard known as IEEE802.16 has been proposed, which promises to provide
a broadband wireless solution to the above problem. This white paper aims at providing a
comprehensive study of this technology.
Introduction
The main advantage of a network based on IEEE standard 802.16, also known as WiMAX (an
acronym for worldwide interoperability for microwave access [1]), is that it can be deployed in areas
where wired digital subscriber line (DSL) service is hard to provide and the cost is high. Not only
does it support high speed data communication, but it also has the ability to maintain dedicated links
and voice over Internet protocol (VoIP) services can be very reliable and of high quality. Currently
for a service provider, deploying a DSL or T1/E1 service for a business customer to a relatively
remote location or outer suburbs can take several months and the cost involved can be significant.
With the help of WiMAX, a service provider will be able to provision that service in a few days and
in a very cost effective manner.
IEEE has established various wireless standards in a hierarchical fashion with help from the expertise of
many communication engineers. Some of the established wireless communication standards include the
following: 802.15 (personal-area network Bluetooth), 802.11 (local-area network Wi-Fi) and 802.16
(metropolitan-area network WiMAX). A new standard, 802.20 for wide-area networks (WANs), has
been proposed and is currently under development.
197
Each of the IEEE standards has been developed to meet certain requirements and they complement each
other. For indoor use, 802.11 has been optimized and it fulfils requirements for home or office
connectivity. 802.16 has been designed to meet outdoor and long-range use for providing last-mile
solutions. To service providers, it also presents itself as a quick and cheap option for backbone service.
WiMAX has been developed by keeping in mind the requirement for cheap and quick deployment of fast
connectivity for both voice and data communication to remote and difficult terrain locations. This white
paper will discuss the development, specification, application, current research work and the future of
WiMAX.
Development
Brief History
IEEE first released the 802.16 standard (IEEE Standard 802.16-2001) on April 8, 2002 and was approved
by the IEEE-SA Standards board on December 6, 2001 [2]. This standard included the physical layer
specification for systems that will operate between ten and 66 gigahertz (GHz). Since then, there has been
two major amendments; the first one (IEEE Standard 802.16c-2002) was released by the IEEE Standard
Association (SA) board on January 15, 2003 [3] and the second one (IEEE Standard 802.16a-2003), on
April 1, 2003 [4].
Although the scope of this release was limited to ten to 66 GHz, the first amendment gave a detailed
system profile that lists sets of features and functions to be used in typical implementation cases [3]. Also,
corrections were made to errors and inconsistencies from the previous release (IEEE 801.16-2001).
The second amendment extended support of WiMAX at frequencies from two to 11 GHz by providing
additional physical (PHY) and medium access control (MAC) layer specifications. It also specified the
standard where the MAC layer is able to support multiple physical layer specifications, point-tomultipoint (PMP), and mesh topologies. The standard also covers specifications for systems operating
between two to 66 GHz (the entire range).
The latest development is the approval of IEEE 802.16-2004 standard being approved by the standards
body on June 24, 2004 [5]. This standard is due to be published by the end of July 2004. It is a revision of
the previous three releases (as discussed above).
Current Developments
In order to make the global adoption of WiMAX successful, steps like interoperability, cost of
deployment, and testing have been taken. The WiMAX forum, which is an industry-led nonprofit
corporation formed for the development and promotion of WiMAX [6], is in the process of helping this
industry by conducting proper tests and later certifying vendor products and systems as WiMAX
CertifiedTM. Currently initial testing of vendor products is taking place and, by the end of 2004,
WiMAXcertified solutions are expected to be available in the market.
198
Description
January 2003
April 2003
Initial System Profiles Selected: 256 OFDM at 2.5, 3.5 and 5.8
GHz
199
Typically, a common WiMAX setup configuration will consist of a base station, as either a small
independent tower or mounted on top a building. 802.16a has a line-of-sight radius of 50 kilometers and a
non line-of-sight of ten kilometers. The non line-of-sight range, also known as the cell radius, can vary
from six to ten kilometers depending on the type of obstacles present (trees, huge buildings, etc.). Within
the cell radius, the throughput and performance is optimal.
802.11 (Wi-Fi)
802.16 (WiMAX)
Technical Differences
Range
Maximum 100
meters. Access
points needs to be
added for greater
coverage
Line-of-sight: Up
to 50 kms.
Nonline-of-sight:
between 6 to 10
kms
Spectrum
Uses unlicensed
spectrum only
Coverage
Designed for
outdoor LOS and
NLOS services.
Scalability
Number of users
range from one to
tens. Intended for
LAN use. Channel
sizes are fixed at 20
MHz.
Number of users
can be thousands.
Channel sizes are
flexible (1.75 ~ 20
MHz)
Data Rate
2.7 bps/Hz,
maximum of 54
Mbps in a 20 MHz
channel
5 bps/Hz,
maximum of 100
Mbps in a 20 MHz
channel
QoS
Supports multiple
QoS and it is built
into MAC.
In order to increase the efficiency of WiMAX, a number of concepts are used in its MAC implementation.
One such concept is the use of a variable length protocol data unit (PDU) [8]. It is possible to combine
multiple PDUs into a single burst. This, in return, saves on PHY overhead. Similarly, to save on the MAC
overhead in a single MAC PDU, multiple service data units (SDUs) for the same service can be
combined. This approach is very helpful in maintaining good QoS.
A trade-off between capacity and robustness in real time is possible because the frame structure of
WiMAX allows dynamic allocation of uplink and downlink burst profiles to its terminals according to
their link conditions. On average, it also gives nearly twice the capacity while maintaining link
availability, in comparison with non-adaptive systems. Using dynamic adaptive modulation, WiMAX is
able to trade throughput for range. When the signal gets weak and the base station is unable to establish a
robust link using the highest order modulation scheme, 64 quadrature amplitude modulation (QAM), it
will reduce the modulation order to 16 QAM or quadrature phase shift keying (QPSK). In this process,
the system is able to increase the range by compromising throughput. It also has a high spectral efficiency
and is tolerant of signal reflections.
In the WiMAX specification, three system profiles have been selected. Each of the threecovering the
5.8 GHz (license exempt), 2.5 GHz, and 3.5 GHz bandsfalls under the 256 orthogonal frequency
division multiplexing (OFDM) PHY mode for the 802.16a standard. Other profiles have been planned
and are in development, including the 2.3 GHz band [7]. To ensure a uniform base for implementing
WiMAX, the PHY will be combined with non-optional MAC.
WiMAX supports flexible channel bandwidth in both licensed and unlicensed spectrum, which simplifies
network planning [10]. It has a tremendous potential to provide solutions for many different service
provider scenarios and meeting their requirements to support voice, data, dedicated links, and best-effort
services, among others. However, providing any of the services tends to compromise the other. For
example, an operator is assigned 20 MHz of spectrum. The operator could divide it into two sectors of ten
MHz each, or four sectors of five MHz each. For a smaller sector size, there is a smaller data-transfer rate,
but it can support a wider range of services (data, voice, dedicated links, best-effort, etc.).
The support for smart antennas gives service providers an option to focus the power into various narrow
sectors. This will increase the coverage distance in that sector direction and increase the number of users. By
creating proper isolation of sectors at the base station, service providers can further increase the number of
users by reusing the same spectrum. To improve non line-of-sight performance, advanced techniques such
as mesh, beam-forming, and multiple inputs, multiple outputs (MIMO) can also be used [7].
The ability to provide both voice and video services is important to WiMAX. These services are latency
(delay) sensitive and to ensure that this sensitivity is properly accounted for, the dynamic time division
multiple access (TDMA) (grant/request) characteristic of the MAC has been implemented. This enables
service providers to provide premium quality level services such as dedicated T1, voice, and video, along
with other high volume services (best-effort) in the same cell area.
802.16a and 802.16e
Both IEEE802.16a and IEEE802.16e are derivatives of IEEE802.16 standard. IEEE802.16a is used for
broadband wireless access (BWA) networks operating in the range two to 11 GHz. IEEE802.16e
standards specifications are currently under development and it is planned to be an extension of the
IEEE801.16a standard [7]. It will add mobility to the existing IEEE802.16a standard that is designed for
fixed operation. Simultaneous development of IEEE802.20 is often considered redundant. However, the
head start that IEEE802.16e has, by being an extension of the established IEEE802.16 standard, over its
rival standard promises to provide a significant advantage. The standards specification for IEEE802.20 is
expected to be released by the end of 2004.
201
One of the major potential vulnerabilities for users of wireless communication is security. The open
nature of its medium gives anyone the ability to listen to or tap into communication packets that have
been transmitted. From the outset, WiMAX has made it mandatory to protect users by employing the data
encryption standard (DES) and triple DES security encryption [10]. In a recent development, the profiles
taken from the advanced encryption standard (AES) encryption mechanism have been taken into account
because of their more secure nature (uses 128, 192 or 256 bit long keys).
Application
Due to a lack of standardization and interoperability the BWA market was fragmented. This caused
problems in adopting a single technology that could be globally implemented. WiMAX addresses these
shortcomings and promises to provide a universal last-mile solution. Today, where DSL services are only
available to high density urban areas, creating a digital divide between urban and rural residents and
businesses, WiMAX comes in as the bridge maker for this division. The ability of quick and costeffective deployment, irrespective of terrain and population density, over a large area does bring the long
time wish of many service providers to reality.
The number of services that WiMAX can provide includes the following:
202
Internet service. Using a satellite upload and downlink to a central location in such towns,
WiMAX can easily provide residential broadband services and also voice communication
services to the local community at no additional cost.
203
Internet. 802.16a offers a large coverage area, up to 50 kilometers with line-of-sight or ten kilometers non
line-of-sight with a shared data rate of up to 75 Mbps. This standard also supports providing latency
(delay) sensitive services such as voice and video.
802.16e is currently under development and it is planned to be an extension of 802.16a [7]. The
purpose of this standard is to add mobility to the currently available standard that provides last-mile
solutions to fixed operations. The standards specification is due to be approved and released by IEEE
by the end of 2004.
Is WiMAX the Same as Wi-Fi (802.11)?
WiMAX is not the same as Wi-Fi. Wi-Fi provides wireless solution for local area networks, suitable for
hotspots and indoor use. WiMAX on the other hand provides wireless solutions for long range outdoor
use, including backbone services for 802.11 hotspots.
What Is the difference between 802.16 and 802.20?
802.16 and 802.20 are different technologies designed to meet different requirements. The purpose of
802.16 is to provide last-mile broadband access to clients at very high speed. The 802.20 standard is being
designed to provide high speed connectivity to mobile devices.
As 802.16e Aims at Providing High-Speed Mobile Connectivity, Does 802.20 Becomes a Standard That
Is Redundant?
802.16e and 802.20 are working at providing the same solutions, although there are some minor
differences. If 802.16e becomes a success then it is feared that 802.20 will lose its purpose. 802.20 is
being developed from scratch, which is a disadvantage as its competitor, 802.16e, already has a head start,
thanks to 802.16a.
When Will WiMAX Products Bbe Available in the Marketplace?
WiMAX products are expected to be available in the market by the end of 2004 and mass production of
WiMAX equipment is expected to begin in 2005.
Conclusion
The difficulty with deployment of wired telecommunication networks with respect to cost, terrain, and time
has created a gap between rural and urban consumers in the developed world. The same reason is responsible
for a much slower growth of telecommunication services in third world countries. Mobile communication did
solve some of the problems with respect to PSTN networks, however as the world gets more dependent on the
Internet, today broadband access to anyone, anywhere, anytime is a high priority.
Development of mobile PSTN networks also created a gap within itself due to the differences in standards
worldwide. WiMAX addresses all the issues and problems faced by the wired network for both voice and
data communication and also joins every effort of its development under a single standard resulting in a
truly global wireless solution. Flexibility, lower cost, high performance, and a global standard will boost
the industry in bridging the gap between rural and urban users.
References
[1]
O'Shea, D., WiMAX Seeks Bigger Impact with Expanded Membership, Telephony, Volume 244,
Issue 8, April 21, 2003, ISSN/ISBN: 00402656, pp. 1415.
[2]
IEEE Std. 802.16-2001, IEEE Standard for Local and Metropolitan Area Networks, Part 16: Air
Interface for Fixed Broadband Wireless Access Systems, April 8, 2002, ISBN: 0-7381-3071-2.
205
[3]
IEEE Std. 802.16c-2002, IEEE Standard for Local and Metropolitan Area Networks, Part 16: Air
Interface for Fixed Broadband Wireless Access Systems Amendment 1: Detailed System Profiles
for 10-66 GHz, January 15, 2003, ISBN: 0-7381-3494-5.
[4]
IEEE Std. 802.16a-2003, IEEE Standard for Local and Metropolitan Area Networks Part 16: Air
Interface for Fixed Broadband Wireless Access Systems Amendment 2: Medium Access Control
Modifications and Additional Physical Layer Specifications for 2-11 GHz, April 1, 2003, ISBN: 07381-3566-6.
[5]
Marks, R. B., IEEE Standards news release, July 24, 2004, http://ieee802.org/16/arc/80216list2/msg01651.html [last accessed: July 31, 2004].
[6]
[7]
[8]
[9]
Vaughan-Nichols, S. J., IEEE Journal, Computer, Volume 37, Issue 6, June 2004, pp. 1013.
[10] Agarwal, A., WiMAX: Moving Broadband A Giant Step Forward, Wireless Design &
Development, Volume 12, Issue 5, May 2004, pp. 1821.
[11] Antonello, G., Yaghoobi, H., Agarwal, A., General Session, Technical Working Group, January 20,
2004,
http://www.wimaxforum.org/news/events/wimax_day_agenda/WiMAX_-_General_
Session_TWG.pdf [last accessed: July 31, 2004].
[12] Wongthavarawat, K., and Ganz, A., Packet scheduling for QoS support in IEEE 802.16 broadband
wireless access systems, International Journal of Communication Systems, Volume 16, Issue 1,
2003, pp. 8196.
[13] GuoSong, C., Deng, W., and Shunliang, M., A QoS architecture for the MAC protocol of IEEE
802.16 BWA system, 2002 International Conference on Communications, Circuits and Systems
and West Sino Exposition Proceedings (Cat. No.02EX591), Volume 1, 2002, pp. 435439.
[14] 3g.co.uk, Fixed WiMAX Subscribers to Surpass 7 Million in 2009, Press Release, July 12, 2004,
http://www.wimax.co.uk/PR2004/July2004/2063.htm [last accessed: July 30, 2004].
[15] Yunker, J., WiMAX will capture 60% of broadband wireless market, but not without a fight,
Pyramid
Research
Press
Release,
November
18,
2003,
http://www.pyramidresearch.com/info/press/release_031118.asp [last accessed: July 30, 2004].
206
WiMAX Promises a
New Era in Telecom
Athena Platis
Wireless Industry Analyst
National Telecommunications Cooperative Association (NTCA)
WiMAX, short for Worldwide Interoperability for Microwave Access, is a non-line-of-sight, point-tomultipoint broadband wireless access (BWA) technology that is promising to change the face of
telecommunications. Designed for longer distances and broader area coverage than Wi-Fi1, the WiMAX
standard seems ideal for metropolitan area networks (MANs) as a wireless alternative to cable, digital
subscriber line (DSL), and T1, or for last-mile broadband access. Industry forecasts say 802.16a will
revolutionize the MAN market, just as 802.11b (Wi-Fi) did for local area networking, but WiMAX is
more than just bigger and better fixed wireless. If the technology lives up to its promises of 50 mile
ranges and the ability to transfer data, voice, and video at speeds of up to 70 megabits per second (Mbps),
it has the capability of turning the entire telecom industry upside down.
Standards Take Time
The Institute of Electrical and Electronics Engineers (IEEE) first approved the 802.16 standard in April
2002 and originally intended it for fixed, line-of-sight wireless access for the first and last mile. This early
standard, which allowed signals to travel up to 30 miles, had limited appeal because it focused on licensed
spectrum between ten gigahertz (GHZ) and 66 GHz.
The newest version, called 802.16a or WiMAX, was approved in April 2003 and incorporated significant
improvements and modifications that gave the technology more widespread appeal.
For instance, the new standard operates in the unlicensed 2.4GHz and five GHz frequencies and supports
signals that reach up to 50 miles. While the technology was not designed specifically for rural areas, it
could be ideal for areas where low population density has prevented the build out of fixed wireline or
mobile wireless networks in the past.
Because it is optimized for fixed and mobile2 broadband in the wide area network, and spans well beyond
the 1,000 feet of 802.113, WiMAX is kind of like Wi-Fi on steroids. Additionally, it already includes
1
Wi-Fi, short for wireless fidelity, is also known by its technical specifications 802.11b, the first standard, and
802.11a, the upgraded standard. Wi-Fi transmits using unregulated spectrum in the 2.4GHz (802.11b) and 5GHz
(802.11a) radio bands, and can achieve data rates of up to 11 megabits per second (mbps) and 54 mbps, respectively.
The technology can be used for omni-directional hotspot-style deployments at close range, or line-of-sight
broadband wireless-style deployments at long distances.
2
A mobile version of the WiMAX, known as 802.16, is in the works at the Institute of Electrical and Electronics
Engineers. A draft of the standard is predicted for late 2004.
207
numerous advances that are still in the works for the 802.11 standard, such as quality of service, enhanced
security, higher data rates, and mesh and smart antenna technology, allowing better utilization of the
spectrum. While the WiMAX standard took many years to evolve and develop, its high-speed, long
distance, and spectrum efficiency may prove it to be well worth the wait. Other standards in the pipeline
include 802.16b, which addresses quality of service, 802.16c, which addresses interoperability with
protocols, 802.16d, which brings together all elements of its predecessors, and 802.16e, which addresses
support for mobile broadband.
WiMAX May Turn The Tables
Once WiMAX goes mobile with its 802.16e standard, it stands to really give third generation (3G) a run
for its money. The new standard will support mobile wireless technology, enabling direct broadband
transmissions to mobile devices in a manner similar to general packet radio service (GPRS). Because the
technology allows for speeds up to 30 times faster than 3G, mobile WiMAX has the potential to undercut
the use of traditional cellular networks. This makes mobile WiMAX a serious threat to the profitability of
3G networks and services that were built at such great expense by wireless carriers. Currently, it is
primarily landline carriers that are investing in WiMAX. The pendulum of telecom industry power is
expected to swing back in favor of the landline operators who were long victims of wireless displacement.
In this way, WiMAX could be one of the most disruptive technologies the telecom industry has ever seen.
Wimax Support Grows.
Many experts say the WiMAX utopia will be a hard sell to telecom operators, especially to those burned
in the past by fixed wireless technologies that fell short of their promises. Other say that it is all a matter
of timing and that it is still too early yet for WiMAX to have a great following. There are, however, firms
with long-term vision that have set their sights on WiMAX. Among the technologys supporters are hightech and telecom giants such as Intel, Siemens Mobile, and Yahoo!.
The mother of all WiMAXbased partnerships began with the creation of the WiMAX Forum just 18
months ago, led by electronics giant Intel. The group has now grown to 89 member companies. Intels
aggressive WiMAX strategy has led it to expand on this idea of collective problem solving, by forming
strategic alliances with various high-tech industry players.
Most notably, there was the announcement in July 2003 that Intel would cooperate with broadband
wireless equipment maker Alvarion to develop low-cost WiMAXcertified equipment based on Intels
802.16a silicon. Then in late March 2004, Intel revealed an agreement with telecom equipment maker
Alcatel aimed at the definition, standardization, development, integration and marketing of WiMAX
end-to-end solutions. Products resulting from this alliance are expected to hit the market by the second
half of 2005. According to Intels aggressive WiMAX strategy, it plans to start building it into its chip
platforms, which power around 80% of all personal computers, by 2006.
Will WiMAX Connect Rural America?
Founded in 1999, America Connect is a privately held corporation located in Raleigh, N.C. The
companys mission is to bring the same cost-effective wireless broadband service to rural or ex-urban
communities that readily are available in metropolitan areas. With the help of vendors like Navini
Networks and NextNet Wireless, America Connect currently is targeting the southeastern United States
and has deployed WiMAX systems in two rural communities north of Research Triangle, N.C. The
3
208
Athena Platis
company is looking to partner with small rural telecom carriers to bring broadband to rural areas, by
selling WiMAX at the wholesale level to rural telcos. These telcos would then, in turn, offer the retail
product to consumers.
Because the frequencies below 11 GHz enable non-line-of-sight4, the standard could be especially
appropriate for last-mile applications in rural areas whose earlier fixed wireless deployments were often
plagued by trees and terrain issues. America Connect currently is offering high-speed wireless Internet
service to residential and business customers via a non-line-of-sight network deployment.
In early April, wireless was called critical to global development at a University of California, Berkeley
sponsored conference focused on technology and its abilities to solve social and economic problems in
developing nations. Experts agreed that the most important technological step to take in underdeveloped
areas is to build out communications networks with wireless capabilities. This strategy to revitalize
depressed areas by enabling communications and commerce via wireless technology can be applied to
rural areas in America as well.
Only Time Will Tell
With its unique communications capabilities, WiMAX has the potential to be one of the most disruptive
technologies the telecom industry has seen. Yet there are many challenges ahead for WiMAX, namely
interoperability and standardization of equipment and chipsets. Though little more than a fairytale now,
WiMAX has many in the industry believing it will stand the test of time. Among them is the Scottsdale,
Arizonabased research firm In-Stat/MDR, which predicts the broadband wireless access market will
increase to more than $1.2 billion by the end of 2007, up from $558.7 million last year, as a direct result
of WiMAX. In the end, only time will tell if WiMAX can live up to its industry hype and make a
difference in rural telecommunications markets.
Additional Resources
America Connect, http://www.broadband2go.net
Alvarion, http://www.alvarion-usa.com
IEEE 802.16 Working Group, http://www.ieee802.org/16
Intel, http://www.intel.com
Navini Networks, http://www.navini.com
NextNet Wireless, http://www.nextnetwireless.com/
WiMAX Forum, http://www.wimaxforum.org
WiMAX: Outlining
Business Strategies
Kotni Mohana Rao
Senior Software Engineer
Wipro Technologies, Bangalore, India
Abstract
Worldwide Interoperability for Microwave Access (WiMAX), an emerging broadband wireless access
(BWA) technology, is getting popularized as a wireless alternative to digital subscriber line (DSL), cable
modem, leased lines, or any other broadband network access technologies. WiMAX supporters claim that
the technology will replace everything from third generation (3G) cellular networks to DSL and wireless
local area networks (WLANs). With the market value expected to reach five billion U.S. dollars by the
year 2007, and with the endorsement of many telecommunication giants, WiMAX is poised to change the
broadband wireless market scenario.
The first thing to consider is whether WiMAXs touted potential is merely hype. Furthermore, now that
several pre-WiMAX networks have been deployed, what are the operators saying about quality of service
(QoS) and return on investment (ROI)? How and when will device manufacturers integrate WiMAX into
their products? Finally, what is the business case for using WiMAX over any other established broadband
wireless alternative?
As 802.16 products will not be widely available for at least another year or so, the standard itself should
play an important role in the future network plans. 802.16 has the potential, but there is a need to derive
proper strategy for commercial success of the technology. With that in mind, it is necessary to get up to
speed with the development of various 802.16 standards.
The purpose of this white paper is to study the operating environment, technologies, market dynamics,
and projected evolution associated with WiMAX. This study can be useful to understand what one must
consider to achieve a business strategy for WiMAX technological markets.
Introduction
The WiMAX Forum is a nonprofit corporation formed by equipment and component suppliers to promote
the adoption of Institute of Electrical and Electronics Engineers (IEEE) 802.16 compliant equipment by
operators of broadband wireless access systems.
The IEEE 802.16 wireless network protocols are the next evolution of the 802.x standards that currently
contain both the wireless fidelity (Wi-Fi) and Bluetooth protocols. The IEEE 802.16 standard is one of
the complementary wireless standards created by the IEEE to help ensure interoperability and reduce the
211
risk of wireless technology deployment. 802.16 is designed for metropolitan area networks (MANs) and
defines a wireless protocol with a range measured in tens of kilometers rather than tens or hundreds of
meters. Products based on the 802.16 protocol will enable transmission of broadband connectivity from a
city to outlying villages.
With the expertise of hundreds of engineers from the communications industry, the IEEE has established
a hierarchy of complementary wireless standards. These include IEEE 802.15 for the personal-area
network (PAN), IEEE 802.11 for the local-area network (LAN), 802.16 for the MAN, and the proposed
IEEE 802.20 for the wide-area network (WAN). Each standard represents the optimized technology for a
distinct market and is designed to complement the other standards.
The WiMAX Forum was established to define standards that meet specific service provider needs and
will provide interoperability testing and certification. This ensures that customer premises equipment
(CPE) and network equipment from various vendors work together right out of the box. Technically,
WiMAX/802.16 gear can provide true broadband access with end-to-end QoS support for both
asynchronous transfer mode (ATM) and Internet protocol (IP) QoS systems, and can support applications
like Internet access, voice over IP (VoIP), and mobility.
Brief Evolution of WiMAX
Service providers have been trying to supplement their networks with wireless last mile technologies for
years. Start-ups and established carriers alike spent fortunes on high-end multichannel multipoint
distribution system (MMDS) solutions in the late '90s, for example, finding a way to bypass the localexchange carrier (LEC) and reach customers directly. Wireless Internet service providers (WISPs) have
invested in various technologies to reach customers in rural areas and small towns that are not well served
by wireline technologies. Even the LECs themselves have been conducting trials with wireless vendors,
looking for a way to reach their customers who cannot be accessed economically by DSL. So far, none of
these efforts has been of much use.
Most of these deployments were plagued by high costs and immature QoS. Some smaller WISPs are
having success, but generally they are focused on very small markets. Service providers who want to go
wireless need a solution that mixes low capital costs, which can compare with DSL or other wireline
solutions, with low operational expenses and provisioning costs. Added to the mix is the fact that carriers
want a reliable, standards-based solution from established vendors; betting on unknown start-up
equipment vendors is no longer part of the game, especially for larger carriers.
There is WiMAX, the industry consortium promoting wireless products falling under the 802.16 group of
IEEE standards. 802.16 defines many variants of the standard, in different frequency ranges with various
characteristics, and provides the baseline media access control (MAC) and physical layer (PHY)
definitions that are necessary for any WiMAX gear.
Work on 802.16 started in July 1999. In all these years, the IEEE 802.16 Working Group on Broadband
Wireless Access has delivered a base and three follow-on standards. The timeline for IEEE 802.16 family
of standards follows.
212
IEEE 802.16: Approved in December 2001, it is used for wireless MANs operating at frequencies
between 10 and 66 gigahertz (GHz). (Air interface for fixed broadband wireless access systems)
IEEE 802.16.2: Published in 2001, it specifies a recommended practice to address the operation
of various broadband systems in the ten to 66 GHz frequency range
IEEE 802.16a: This was an amendment to 802.16 made by IEEE in January 2003, which adds to
the original standard operation in licensed and unlicensed frequency bands from two to 11 GHz
Initial System Profile: Introduced in April 2003 with 256 orthogonal frequency division
multiplexing (OFDM) at 2.5, 3.5 and 5.8 GHz
IEEE 802.16 REVd: Expected in the third quarter of 2004 revision PAR for 802.16 & 802.16a
to add WiMAX system profiles and errata for two to 11 GHz in support of 802.16e requirements
IEEE 802.16e: Expected in the first quarter of 2005 for pedestrian mobility and regional roaming.
Amendment for mobile wireless broadband up to vehicular speeds in licensed bands from two to
six GHz. This enables roaming for portable clients (laptops) within and between service areas
Component Makers
o Creates a volume opportunity for silicon suppliers
Equipment Makers
o Can innovate more rapidly because there exists a standards-based, stable platform upon
which new capabilities can be rapidly added
o No longer need to develop every piece of the end-to-end solution
Operators
o A common platform drives down the cost of equipment and accelerates price and
performance improvements unachievable with proprietary methods
o Generates revenue by filling broadband access gaps
o Quickly provides T1 / E1 level and on demand high-margin broadband services
o Reduces the cost associated with deployment due to lesser equipment
o Need no longer be locked into a single vendor since base stations will interoperate with
multiple vendors' CPEs
Consumers
o More broadband access choices; especially in areas where gaps exist - worldwide urban
centers where building access is difficult, in suburban areas where the subscriber is too
far from the central office, and in rural and low populated areas with poor infrastructure
o More choices for broadband access create competition resulting in lower monthly
subscription prices
213
Another customer, an interexchange carrier (IXC) and mobile carrier in Latin America, is deploying
Alvarions gear to create their own last mile link to customers for voice and data services.
Keeping an eye on these deployments might provide a business case for WiMAX. The momentum is
building and WiMAX may very well turn out to be a technology that service providers of all types need
and can afford as well.
2. Identifying Target Markets
To provide a profitable market growth to the BWA network provider, one should identify the potential
markets for the WiMAX. Prior to 2003, BWA was all about under-served markets.
WiMAX offers broadband access where DSL cannot reach customers for technical or economical
reasons. It provides a last-mile alternative for competitive operators with combined voice and broadband
access. It also offers nomadic access complementary to Global System for Mobile Communications
(GSM) and enhanced data rates for GSM evolution (EDGE), wireless fidelity (Wi-Fi), and third
generation (3G). WiMAX can be used to maximize Wi-Fi network efficiency.
To get a significant business advantage from WiMAX, one of the potential steps would be to break into
the residential broadband wireless market.
The strategic approaches from a business perspective for WiMAX might be as follows:
Cellular backhaul
Broadband on demand
Residential broadband, filling the gaps in cable and DSL coverage
Under-served areas
Best-connected wireless service
215
Why Certification?
Quality guaranteed by third party.
Service provider can choose from multiple vendors with interoperable equipment.
Multiple sources of broadband wireless hardware minimize the risk of product availability for
deployments.
216
217
Reference
http://www.wimaxforum.org
218
WiMAX could be the key to breaking through the last-mile barriers that have slowed broadband adoption,
especially in rural areas where the cost of deploying broadband connections has not been economical.
WiMAX can be taken as a wireless alternative to DSL and cable (see Figure 1). The principal advantages
of the systems based on 802.16 are as follows:
The ability to provision services quickly, even in areas that are inaccessible for wired infrastructure
219
220
Amit Rawal
221
222
Amit Rawal
WiMAX is designed to accommodate either frequency division duplexing (FDD), which is more
suited to enterprise traffic, or time division duplexing (TDD), which is more adapted to asymmetrical
traffic. Cohabitation of FDD and TDD techniques is possible within the same bands, provided guard
bands are implemented.
WiMAX: Throughput, Flexibility, Scalability, and Security
IEEE 802.16 uses a robust modulation scheme, delivering high throughput at long ranges with a high
level of spectral efficiency that is also tolerant of signal reflections. Tradeoff of throughput for range can
be achieved by lowering the order of the modulation scheme at the base station.
WiMAX or IEEE 802.16 standard provides flexibility as wireless broadband access can be quickly and
easily set up at new and temporary sites, saving the time needed to get a T1 or DSL line connection. It
also provides flexibility in terms of channelization, carrier frequency, and duplex mode (TDD and FDD)
to meet a variety of requirements for available spectrum resources and targeted services.
The 802.16 standard is scalable, as with wireless broadband access it is easy to ramp up service at a
location for a short period of timesomething that wired broadband access service providers currently do
not do. For example, if an operator is assigned 20 MHz of spectrum, that operator could divide it into two
sectors of 10 MHz each or 4 sectors of 5 MHz each. By focusing power on increasingly narrow sectors,
the operator can increase the number of users while maintaining good range and throughput.
WiMAX provides a MAC layer that uses a grant-request mechanism to authorize the exchange of data.
This feature allows better exploitation of the radio resources, in particular with smart antennas, and
independent management of the traffic of every user.
Market for WiMAX
WiMAX will succeed in every marketbut for different reasons. In emerging markets, operators are
interested in using WiMAX for low-cost voice transport and delivery. In developed markets, WiMAX is
all about broadband Internet access. Overall, the markets without any fixed infrastructure pose the
greatest opportunities.
WiMAX success in the BWA market would be due to standardization, interoperability, state-of-the-art
radio efficiency with NLOS capability, and strong support from the radio equipment manufacturers and
chipset industries. WiMAX will open up three main markets:
It will bridge the digital divide in low-density areas. The prime markets are in Western Europe,
North America, and some Asia-Pacific countries including China and India, which have huge
potential market for broadband users.
It will offer high-speed Internet and voice access in urban and suburban areas. It will also support
nomadic usage.
It will allow portable Internet application by providing broadband access on the move.
WiMAX is a serious threat to third generation (3G) because of its broadband capabilities, distance
capabilities, and ability to support voice effectively with full QoS. WiMAX can slash the single biggest
cost of deployment: access charges for linking a hotspot to a local phone or cable network. WiMAX
223
integrates perfectly into existing fixed and mobile networks. A high frequency version of 802.16 would
allow entrepreneurs to blast a narrow, data-rich beam between antennas miles apart.
Performance: So far, the specifications for WiMAX are still on paper (source WiMAX Forum, March
22, 2004), as there is no commercial product certified by the Forum and the final version of 802.16a
has not yet been approved. It is possible that the certification process will prove more arduous than
expected. There is also a possibility that real-life performance does not meet the expectations,
especially with regard to coverage range and CPE form factor, cost and ease of installation.
Split of WiMAX into multiple semiproprietary solutions: 802.16a and the expected 802.16e standards
could complement each other, providing subscribers with a mix of fixed and mobile access.
Case Studies
There are no known current implementations of 806.16x wireless networking technology. In India, Bharat
Sanchar Nigam Limited (BSNL) is ready to launch wireless fidelity (Wi-Fi), 802.11 standard, at Pune
airport in the month of August 2004. By the end of 2004 there will be public hotspot in Delhi, India. The
state-owned Mahanagar Telephone Nigam Ltd (MTNL) would soon offer wireless Internet through
802.11b and 802.11g (Wi-Fi standards). MTNL would spend about Rs 1.5 crore initially to install
hotspots in 40 locations with 100 access points. Wi-Fi was installed in parts of Bangalore, India, in
October 2003 by Sify. It built 120 WiZones in the city at about Rs 50,000 each.
802.20 Mobile Broadband Wireless Access (MBWA) and 802.16e
The IEEE 802.20 Working Group is a new wireless networking standard for Mobile Broadband Wireless
Access (MBWA). The 802.20 or Mobile-Fi standard defines the physical and MAC layers for a highbandwidth, IPbased, fully mobile wireless network. The groups intention is to fill the gap between
224
Amit Rawal
existing 802 standards with high data rates and low mobility and existing cellular standards with low data
rates and high mobility.
The IEEE 802.16e standard is also addressing the need for high-bandwidth mobile wireless Internet
access within a metropolitan area. This will be similar in function to the general packet radio service
(GPRS) and the radio transmission technology (1xRTT). The 802.16e standard combines fixed and
mobile operation in licensed bands (26 GHz), approved in December 2002. There are some technical
differences between both the standards viz. 802.20 and 802.16e. For one, 802.16e will add mobility in the
2 to 6 GHz licensed bands, while 802.20 aims for operation in licensed bands below 3.5GHz. More
importantly, the 802.16e specification will be based on an existing standard (802.16a), while 802.20 is
starting from scratch.
The 802.20 interface seeks to boost real-time data transmission rates in wireless metropolitan area
networks to speeds those on which rival DSL and cable connections (1 Mbps or more) are based, cell
ranges of up to 15 kilometers or more, and it plans to deliver those rates to mobile users even when they
are traveling at speeds up to 250 kilometers per hour (155 miles per hour). This would make 802.20 an
option for deployment in high-speed trains. The 802.16e project authorization request specifies that it will
only support subscriber stations moving at vehicular speeds of 120 to 150 kilometers per hour (75 to 93
miles per hour). Essentially, 802.16e is looking at the mobile user walking around with a PDA or laptop,
while 802.20 will address high-speed mobility issues. This key difference will define the manner in which
the two standards would be deployed.
Conclusion
WiMAX is the most important of the host of wireless standards emerging from the IEEE and 3G bodies.
We expect WiMAX to be the dominant technology for wireless networking, providing full mobility as
well as low cost fixed-broadband access. Its relationship with other wireless technologies is illustrated in
Figure 4.
225
226
Simulation of 802.16a
Deployment Scenarios and
Their Performance Analysis
Dr. Daniel Rodellar
Project Leader and Telecommunication Engineer
Swisscom Innovations
227
Lately, some network operators have joined the forum, and it is clear that even that the chips and the
equipment could be ready as the WiMAX forum predicts on their roadmaps, there have to be operators
willing to deploy the new networks, and for that the business has to be there. All the market surveys show
a clear push toward wireless broadband connectivity and services (data, voice, and video services,
mainly), but of course, there are already some technologies that could deliver to some extend such
services and other technologies where cellular operators have already invested towards the broadband
mobility solution. To start such networks several requirements have to be met, first there has to be a
business case that makes the financial investment profitable or at least to provide a benefit for the
operators users to stay with it (customer retention). Then there has to be a superiority of this technology
against other technologies, so that a competitor of the operator cannot provide the same services with a
better quality and for a lowest price. Both aspects can be solved by scanning all available technologies
and analyzing the performance of WiMAX to position it accordingly. To solve part of the business case
questions we need to perform studies on the performance in the different scenarios of use of WiMAX.
This paper shows the analysis of the IEEE 802.16a standard within different scenarios, with extreme and
more realistic parameters and taking care of the regulatory constraints. Computations have been
performed to estimate values of throughput versus distance of the subscriber station from the base station.
2. Framework Study and Definition of Scenarios
The study was done under the regulatory framework in force in Switzerland. Frequencies to be used in
Europe for BWA systems are 2.5 to 2.7 GHz, 3.3 to 3.8 GHz and 5.7 to 5.9 GHz (license-free) but for
Europe, and specifically Switzerland, we focused our study on the 3.3 to 3.8 GHz band. The radiated
power of transmission equipments is ruled by the decree on non-ionizing radiations 0: it limits long-time
exposure to an electric field of maximum 6V/m. Bigger installations with transmission power above 6 W
(37.8 dBm) EIRP imply a lengthy procedure for all antenna locations.
Different scenarios have been simulated, leading to different usage cases and business cases. In a first
case the subscriber stations (also known as customer premises equipment [CPE]) located outside the
customers house (see Figure 1).
(a)
(b)
Figure 1: (a) Scenario of Line-of-Sight (LOS) Coverage and Antenna at the Rooftop;
(b) Scenario with Outdoor Antenna and Non Line of Sight (NLOS)
This scenario leads of course to most favorable radio link budget because LOS may often be possible, but
suffers from the need for in-house cabling even though the signal can then be repeated with, for example,
228
Dr. Daniel Rodellar, M. Eng. Ludovic Fournier, and Dr. Christian Fischer
Wi-Fi. The second case still assumes outdoor antennas but not on the roof, assuming NLOS propagation
(see Figure 2). The third scenario now assumes an internal antenna. In this last case the propagation loss
is much bigger than in the previous cases due to high penetration losses. The antenna is still assumed to
be high gain.
(a)
(b)
where d0 is a given close-in distance, is the wavelength of the signal, hB is the transmitter (base station)
height between 10 and 80 m, hS is the receiver (subscriber station [SS]) antenna height between 2 and 10
229
m. The parameters needed to compute are dependent on the terrain type, A, B or C, and are given below
in Table 1. Note that it is possible to include a shadow fading term 0 in the above formula which we have
excluded since fading is accounted for in the link budget.
Model Parameters
a
b
c
Terrain Type A
4.6
0.0075
12.6
Terrain Type B
4
0.0065
17.1
Terrain Type C
3.6
0.005
20
and therefore in dB, we find for the two-ray path loss as a function of distance from
The different propagation models have been programmed and the available throughput at a given distance
was computed by comparing the signal strength at the considered distance and the sensitivity for every
modulation specified in the documentation for a commercially available 256 OFDM 802.16a product.
The different scenarios are then defined through a choice of channel-model and link-budget parameters.
Notably, we approximate predominately benign LOS conditions by the ERCEG C model. The more
difficult and generally NLOS conditions are approximated by the ERCEG A model, with the ERCEG B
model representing some intermediate situation. The two-ray channel model is used as a comparative
generic model. Parameters such as increased penetration loss are taken into account in the link budget
directly. The simulations are based on a commercially available product with the following
characteristics, as shown in Table 2.
Parameters
Rx Noise figure
Bandwidth
Average Tx Power
Tx BS Sectors
SS RX Gain
Tx Antenna Gain
Value
7
1.75 or 3.5
20
6
17
16.5
Unit
dB
MHz
dBm
dBi
dBi
230
Dr. Daniel Rodellar, M. Eng. Ludovic Fournier, and Dr. Christian Fischer
In addition, the manufacturer specifies the following data rates and associated sensitivities, as Table 3 shows.
Data Rates/Sensitivities for 3.5 MHz Channel
94 dBm
2 Mbps
91 dBm
4 Mbps
85 dBm
8 Mbps
79 dBm
12 Mbps
where M is the number of interfering cells or sectors, respectively. While using six sectors with only two
frequencies is economical from a frequency allocation point of view, the sectors will in practice suffer
from side lobes and also radiate some energy backwards, causing cochannel interference from within the
cell. For our purposes, we assume that no energy will be radiated outside of the sector and nothing
backwards, i.e., an infinite front-to-back ratio. Another point worth noting is that the formula given above
is a worst-case assumption insofar as that it gives the C/I at the cell edge. This does not mean that the C/I
cannot be better in the interior of the cell but simply says what the best achievable C/I will be at the edge.
By using the above definition of carrier to interference ratio, it can be shown that the degradation due to
cochannel interference is given by CdB, based on the target signal to interference-plus-noise ratio (SINR).
The noise and cochannel interference term is given by the following:
231
BS antenna height
SS antenna height
Tx bandwidth
Tx power
Tx antenna gain
Rx antenna gain
Rx antenna beam width
Noise psd
Rx noise figure
Total margin
Extreme
80m
10m
3.5 MHz
30 dBm
16.5 dB
17 dB
20 deg.
174 dBm/Hz
7 dB
0 dB
Realistic
30m
6m
3.5 MHz
20 dBm
16.5 dB
17 dB
20 deg.
174 dBm/Hz
7 dB
0 dB
Indoor
30 m
6m
3.5 MHz
20 dBm
16.5 dB
17 dB
20 deg.
174 dBm/Hz
7 dB
26 dB
Erceg A
LOS
Erceg C
NLOS
Figure 3: Bit Rate versus Distance in the Single Cell Scenario with the Parameters
of the Extreme Case for Different Channel Models
232
Dr. Daniel Rodellar, M. Eng. Ludovic Fournier, and Dr. Christian Fischer
There is a factor of more than three for these values. This means that for this case, the planning of where to
place the antennas is critical, and the operation (LOS or NLOS) makes a big difference in terms of coverage.
The second conclusion on these values is that the maximum distance for the NLOS case will be 3 km, and
the only way to compensate the lower values for the other case will be to increase the channel bandwidth
(until 20 MHz). Note that the 10 Mbps are supposed to be a single receiver at the given distance. In a
more realistic scenario, several receivers would be in the coverage area and the transmission parameters
will change from one to another, as well as the modulation parameters.
4.2 Realistic Case with the Single-Cell Scenario
In this second case, the values correspond more to the ones we expect to be in the European deployment,
and the results are much lower than the extreme case:
There is a factor of almost two for these values. But the most impressive difference comes in the
comparison with the previous case, because there is a reduction factor of five for the LOS case and three
for the NLOS (see Figure 4).
Erceg C
NLOS
Erceg A
LOS
Figure 4: Bit Rate versus Distance in the Single-Cell Scenario with the Parameters
of the Realistic Case for Different Channel Models
4.3 Indoor Case with the Single-Cell Scenario
In the indoor case the LOS computation makes no sense and it comes in these results just for been
complete in the simulations with all channel models. Only the Erceg C (NLOS) has to be considered (see
Figure 5).
233
Erceg C
NLOS
Figure 5: Bit Rate versus Distance in the Single Cell Scenario with the Parameters
of the Indoor Case for Different Channel Models
Indoor penetration can vary enormously depending on the construction materials used. In this case the
margin has been increased by 26 dB and the distance is then reduced to 0.3 km to obtain the 10 Mbps. We
also see in Figure 5 that the decrease of the plot is more abrupt than for the two previous cases, which
means that the performance deteriorates faster when distance increases. This case requires a much more
accurate planning in the deployment phase.
4.4 Extreme Case with the Cellular System Scenario
The first conclusion from the extreme case in the cellular system scenario is that we cannot obtain 12
Mbps as we could in the previous section for the single cell case, because we cannot obtain a SINR at the
cell edge greater than the C/I. Of course, in the inside of the cell the 12 Mbps can be reached.
The figure was obtained with a reuse factor of four. To compare the different cases we will use 5 Mbps as
a reference:
There is a factor of three for these values and they are very similar of the ones obtained for the single-cell
scenario but with half the bitrate (here it is 5 Mbps and in the single cell it was 10 Mbps) (see Figure 6).
234
Dr. Daniel Rodellar, M. Eng. Ludovic Fournier, and Dr. Christian Fischer
Erceg A
LOS
Erceg C
NLOS
Figure 6: Bit Rate versus Distance in the Cellular System Scenario with the
Parameters of the Extreme Case for Different Channel Models
4.5 Realistic Case with the Cellular System Scenario
In the realistic case, Figure 7 shows the following:
There is the same factor of two as in the single cell case, but with half the bit rate.
Erceg A
LOS
Erceg C
NLOS
Figure 7: Bit Rate versus Distance in the Cellular System Scenario with the Parameters
of the Realistic Case for Different Channel Models
235
Erceg C
NLOS
Figure 8: Bit Rate versus Distance in the Cellular System Scenario with the
Parameters of the Indoor Case for Different Channel Models
5. Conclusion
WiMAX is the latest generation of fixed wireless technology. What differentiates WiMAX from earlier
BWA iterations is standardization and the NLOS capabilities. The vendors behind WiMAX hope to do
for last-mile wireless broadband what the Wi-Fi Alliance did for Wi-Fi. WiMAX promises to expand the
availability of broadband service to residences and businesses that are currently underserved, including
low-density rural locations in developed countries, as well as in emerging markets. From the perspective
of service providers, the WiMAX standard provides robust support for multiple service levels but the
open question prior to deployment is the coverage calculations.
This paper describes several models for deployment of WiMAX in different scenarios with different
cases (sets of parameters). Taking a generic WMAN 802.16a product with 3.5 MHz channels for a
maximum throughput of 12 Mbps per sector per channel (single cell) or in the cellular system case,
several frequencies reused, in the 3.5 GHz band, the throughput has been derived versus cell radius
(distance) figures.
There is a factor of 3 to 5 from the realistic case to the extreme case, which changes a lot the coverage
and thus the real deployment. There is a factor of 2 to 3.5 from NLOS to LOS, which means to
understand what operation mode is going to be used and their advantages and disadvantages.
There are too many uncertainties to be able to give a concrete and detailed deployment scenario. A
more realistic study needs to be carried out, with real performance values. The hype on this new
technology comes from the claims of the standard that correspond to extreme values of the different
tunable parameters.
236
Dr. Daniel Rodellar, M. Eng. Ludovic Fournier, and Dr. Christian Fischer
The WiMAX standard presents a lot of different parameters, not only in the physical layer but also on the
media access control (MAC) layer, to play with, wisely enough to obtain from each equipment the best
performance and the maximum coverage given the constraints on channel bandwidth, maximum power,
etc., that will allowed by the regulator.
The best advice to see if any deployment is feasible will be to test the equipment in a real scenario and
stress the different parameters to obtain the sensibility on the performance measures.
6. References
[1]
V. Erceg et al., An empirically based path loss model for wireless channels in suburban
environments, IEEE J. Select Areas Commun., Vol. 17, No. 7, July 1999, pp. 12051211.
[2]
V. Erceg et al., Channel Models for Fixed Wireless Applications, 2001, http://ieee802.org/16
[3]
IEEE 802.16a Working Group, Part 16: Air Interface for Fixed Broadband Wireless Access
Systems Amendment 2: Medium Access Control Modifications and Additional Physical Layer
Specifications for 211 GHz, 2003.
[4]
[5]
D. Rodellar, Analysis and Study of a BWA Deployment with IEEE 802.16 in Switzerland,
WiMAX Summit 2004, Paris.
237
Ismail Erturk
Lecturer, Electronics and Computer Education Department
Kocaeli University, Turkey
Cuneyt Bayilmis
Research Assistant
Kocaeli University, Turkey
Summary
Together with the developments in high performance wireless computers and other mobile devices, the
importance of wireless or mobile data communication has been increased. Several wireless networking
solutions have been developed to provide different types of services for various end-user applications.
Currently, wireless networking infrastructures such as wireless local area networks (WLANs) are not
suitable for multimedia applications each requiring a different quality of service (QoS) support with various
traffic parameters. WLANs predominantly support data traffic. Asynchronous transfer mode (ATM)
technology in the wired network is capable of supporting demanding applications such as real-time voice or
video transmission, image browsing, and interactive data exchange in an integrated manner. Due to the
success of ATM, wireless ATM (WATM) concepts and related researches are of importance in the
information technology area. The main objective of WATM, which promises seamless transmission of
different traffic such as voice, data, and video over wireless medium, is to implement high bit rate and QoS
guaranteed data transfer, which is already well achieved by ATM technology over wired medium. To
support QoS guaranteed data transfer over error-prone and low bandwidth wireless medium, an effective
MAC protocol must be designed and utilized. In this paper, computer modeling and simulation of such
wireless networks as IEEE 802.11b WLAN, media access controlguarantee based (MACGB) based
WATM, and packet reservation multiple access with dynamic allocation (PRMA/DA) based WATM
transferring multimedia application traffics are realized using OPNET Modeler. Simulation results of a
MACGB based WATM network are also presented together with comparisons to those of a PRMA/DA
based WATM and to those of an IEEE 802.11 based WLAN. Being the most distinctive feature of the
WATM, apart from the most deployed IEEE 802.11 based WLANs, a MACGB based WATM network
can effectively utilize the scarce wireless medium resources with its constant bit rate (CBR), variable bit rate
(VBR), available bit rate (ABR) and unspecified bit rate (UBR) support for various multimedia and realtime traffic as well as allowing simultaneous multiple connections between wireless terminals.
239
1. Introduction
A wireless LAN is a flexible data communication system implemented as an extension to, or as an
alternative for, a wired LAN. WLANs transmit and receive data over the air using electromagnetic waves
in order to minimize the need for wired connections and to provide all the functionality of wired LANs,
but without the physical constraints of the wire itself. WLAN applications have been increasingly used
for data communications together with the recent advances in digital communications, semiconductor
technologies, and portable computers. This technology, by which users can be mobile and can have
access to a network resources in a given coverage area, has been already deployed in hospitals,
universities, and big stores, among other places [1].
Together with developments in high performance wireless computers and other mobile devices, the
importance of wireless/mobile data communication has been increased. Traditional wireless networks
universally deployed, such as IEEE 802.11 WLAN, cannot provide the necessary QoS guarantees for
bursty traffic such as real-time multimedia applications.
Asynchronous transfer mode (ATM) is a transmission technology considered as the standard for
broadband integrated services digital network (BISDN) with guaranteed QoS to all possible traffic types
in wired medium. The success of ATM technology in the wired network has also initiated much research
on wireless ATM (WATM) concepts [23]. WATM has been proposed to transport different types of
traffic, from voice to data to video in a wireless environment. WATM is principally intended to support
QoS guaranteed data traffic for high bit rate broadband multimedia applications [4].
To avoid the problems resulting from the wireless medium characteristics, for example, low bandwidth is
error-prone, new layers must be added to standard ATM layers. These are namely a medium access
control (MAC) layer providing effective allocation of medium resources shared by many different users,
and a data link control (DLC) layer used for flow and error control [3]. Together they establish the basis
for QoS supported application traffic transfer using WATM.
This paper will compare the performance of different wireless networks in transferring multimedia
applications. The chapters of this paper are organized as follows. Section Two introduces WLAN
technology and describes various types of WLANs. Section Three begins with a brief explanation of
WATM, its network components, and protocol stack. Both PRMA/DA and MACGB protocols are also
described shortly. Section Four focuses on the performance evaluation of a WATM network utilizing the
MACGB protocol, which has been modeled and simulated under different traffic loads using a
commercially available program called OPNET Modeler with Radio Module. The simulation results
obtained are compared with those of a WATM network utilizing the PRMA/DA MAC protocol and with
those of an IEEE 802.11b WLAN network, which were also obtained under the same networking
conditions as the MACGB based WATM network. Finally, Section Five summarizes the performance
results and includes final remarks.
2. Wireless Local Area Networks (WLANs)
A WLAN aims to offer all the features and benefits of traditional LAN technologies (e.g., Ethernet and
Token Ring) but without the limitations of being tethered to a cable. Todays wireless local area
networks are designed to support mobile computing in small areas such as a building, park, airport, or
office complex. WLANs also offer access to legacy LAN applications. WLANs provide cable-free
access to data rates of one Mbps or higher for both indoor and outdoor environments. They can be
broadly classified under radio or infrared LANs [1]. Radio LANs can be based on narrowband or
spread spectrum transmissions.
240
In spread spectrum modulation techniques, the bandwidth of the transmitted signal is much greater than
the bandwidth of the original message, and is determined by both the message to be transmitted and by an
additional signal known as the spreading code.
The most distinctive characteristic of spread spectrum systems is that the signals used for the transmission
of information have associated bandwidths much wider than the underlying information bit rate of the
system. By transmitting the message energy over a bandwidth much wider than the minimum required,
spread spectrum modulation techniques present two major advantages: low power density and
redundancy. The former relates to the fact that the transmitted energy is spread over a wide band, and
therefore, the amount of energy per specific frequency is very low. The latter relates to the fact that the
message is, or may be, present on different frequencies from where it may be recovered in case of errors.
The effect of redundancy is that spread spectrum systems present high resistance to noises and
interference, being able to recover their messages even if noises are present on the medium.
There are basically two types of spread spectrum modulation techniques: frequency hopping (FHSS)
and direct sequence (DSSS). Frequency hopping works very much like its name implies. It takes the
data signal and modulates it with a carrier signal that hops from frequency to frequency as a function of
time over a wide band of frequencies. With frequency hopping spread spectrum, the carrier frequency
changes periodically. The frequency hopping technique reduces interference because an interfering
signal from a narrowband system will only affect the spread spectrum signal if both are transmitting at
the same frequency at the same time. Thus, the aggregate interference will be very low, resulting in
little or no bit errors.
Direct sequence spread spectrum combines a data signal at the source station with a higher data rate bit
sequence, which many refer to as a chipping code (also known as processing gain). A high processing
gain increases the signal resistance to interference.
Infrared LANs use part of the electromagnetic spectrum just below visible light as the transmission
medium. Unlike radio waves, infrared frequencies are too high to be modulated in the same way as radio
frequencies. Infrared LANs operates around the 850 nm wavelength [156].
2.1. WLAN Topologies
IEEE 802.11 supports two different topologies, called independent basic service set (BSS or ad-hoc
network) and infrastructure BSS. The former is the simplest type of IEEE 802.11 LAN. By using a
peripheral component interconnect (PCI) card in computers and a PCMCI card in laptops, a simple adhoc network communicating peer to peer can be easily established. As can be seen in Figure 1, the
smallest WLAN may consist of two stations. This mode of operation is possible when stations are able to
communicate directly and the network does not have an access point (AP). It is principally used to
quickly and easily set up a WLAN. Ad-hoc networks provide the means for wireless terminals to
communicate only with each other.
241
Fixed
Wireless
Mobile
Wireless
Cell 1
Cell 2
Internet
Access
Point
Fixed
Wireless
Server
Mobile
Wireless
Cell 2
242
band using FHSS, DSSS, infrared, and radio frequency (RF). It can deliver a data rate of up to two
megabytes per second (Mbps). The IEEE 802.11 standards family is summarized in Table 1 [7].
IEEE 802.11b is the most widespread and preferred standard of these methods. It operates in 2.4 GHz
ISM band, using DSSS technique. It supports data rates ranging from one Mbps to 11 Mbps. The 2.4 GHz
radio frequency band is also used in other wireless standards, such as Bluetooth, cordless phones, and
microwave oven. Therefore, they can cause interference resulting in decreased throughput.
Interoperability between different suppliers IEEE 802.11b products is tested and certified by the
Wireless Ethernet Company Alliance (WECA), which is now know as the Wi-Fi Alliance [7].
IEEE 802.11a, describing a new license free radio band at five GHz, has constituted an alternative to
different applications using 2.4 GHz radio band. It uses orthogonal frequency domain multiplexing
(OFDM) modulation. It increases the data rate per channel from the 11 Mbps of 802.11b to 54 Mbps.
IEEE 802.11a is much more convenient for multimedia applications with high data rate compared to the
others. Using OFMD instead of DSSS provides better performance and a larger coverage area, although it
also requires much more power supply compared to the IEEE 802.11b [567].
Standard Explanation
802.11
802.11a
802.11b
802.11e
802.11f
802.11g
802.11h
802.11i
802.1X
HiperLAN2 supports different QoS for different connections. QoS support allows the transmission of a
mix of different types of information (e.g. voice, video, or other data). HiperLAN2 also provides unicast,
multicast, and broadcast transmissions. HiperLAN2 MAC protocol is based on a dynamic time
division/demand multiple access / time division duplex (TDMA/TDD) scheme [8].
2.2.3. Bluetooth
Bluetooth technology is designed as a short-range connectivity solution for personal, portable, and
handheld electronic devices so that it provides its users with a low power consumption, high speed, cheap,
reliable and interoperable mobile communication. It can deliver both wireless data and wireless voice
application traffics in a similar manner.
Bluetooth devices communicate using a FHSS technique in the 2.4 GHz ISM band. It operates as a 79
channel frequencyhopping system in the frequency range 2.4022.480 GHz with a channel spacing of 1
MHz. The hopping rate is 1600 hops/s and Bluetooth data rate is one Mbps. The Bluetooth specification
provides mechanisms for Bluetooth devices to discover each other, exchange identities, and establish
communications with each other. This is referred to as an ad hoc network [9].
The smallest network part consisting of Bluetooth devices communicating each other is called a Piconet,
with configuration similar to star topology. A Piconet functions on the master and slave concept, so that a
master may communicate with multiple slavesup to 7 active slaves and up to 255 parked slaves.
Several Piconets may be combined to form a Scatternet [10].
2.2.4. The HomeRF SWAP Standard
The HomeRF shared wireless access protocol (SWAP) is designed to carry voice and data within the
home. It can also interoperate with the public switched telephone network and the internet. The SWAP
system operates in 2.4 GHz ISM using FHSS. It can also be assembled in an ad hoc network or be
controlled by a central connection point like Bluetooth. Currently two versions of it exist: SWAP 1.0
supports up to one to two Mbps and SWAP 2.0 supports up to ten Mbps.
The SWAP technology combines QoS characteristics of existing Digital Enhanced Cordless Telephone
(DECT) and beneficial characteristics of IEEE 802.11 wireless LAN standards to enable a new class
home cordless service. As such, it provides voice support for DECT and TCP/IP support for IEEE 802.11.
SWAP supports both TDMA and carrier sense multiple access/ collision (CSMA/CA) [1].
2.2.5. ZigBee
ZigBee technology is low data rate, low power consumption, low cost, and wireless networking protocol
targeted towards automation, PC peripherals, toys, and remote industrial control applications. ZigBee has
been developed by the IEEE 802.15.4 task group and the ZigBee Alliance to provide the first general
standard for these applications [1112].
It uses two physical layers based on DSSS. The 2.4 GHz PHY supports an over air data rate of 250 Kb/s
and the 868 MHz/915 MHz PHY supports over air data rates 20 Kb/s and 40 Kb/s. This standard uses
avoidance (CSMA/CA) as a channel access method.
ZigBee supports three networking topologies; star, mesh, and cluster tree. A typical ZigBee network
application has 254 client nodes and one master node.
3. Wireless ATM
Recently, there has been a growing interest in the subject of WATM. WATM networks are wireless
extensions to both public and private wired ATM networks. Unlike traditional WLANs, WATM will see
244
a high demand in wireless applications involving multimedia. In 1996, the ATM Forum Wireless
Working Group started activities to develop standards for WATM networks. ETSI has also a
standardization activity ongoing for the physical layer of WATM. Above 300 MHz, unlicensed spectrum
is made available in the US and Europe for wireless access, mainly for the purpose of accessing the
Internet for multimedia applications. Such applications will require some kind of wireless packet
networking, and WATM has the potential to fulfill this recently emerging need [14].
Like traditional WLANs, WATM networks will have to deal with the time-varying nature of the wireless link.
While WATM systems are being implemented, there will be several problems resulting from the physical
characteristics of the transmission medium to be tackled. First of all, ATM technology was originally designed
for fiber-optic transmission links characterized by extremely low bit error rate (about 10-12), although WATM
will have to cope with high bit-error rate (BER) values of wireless environments (about 10-4). Secondly ATM
works with bandwidth rich environments where a wireless channel is an extremely expensive resource [4].
Thus this limited bandwidth must be used efficiently. To avoid these limitations resulting from wireless
medium, available ATM systems must be improved and new modules must be added. Once these problems
can be overcome, WATM will provide several significant advantages including; 1) flexible bandwidth
allocation and service type selection for a range of applications, 2) efficient multiplexing of bursty data and
multimedia sources, 3) a wireless platform with multimedia support remedying any QoS needs in a better way,
4) end-to-end provisioning of broadband services over wireless or wired networks, and 5) simple compliance
with the requirements of wireless access to an ATM network [4].
3.1. WATM Network Components and Protocol Stack
In WATM architecture, wireless terminals (WTs) and base stations (BSs) can be fixed or mobile.
Therefore, some network components constituting a WATM can be likewise different. A WATM system
with mobile users and fixed BSs consists of three major network components (see Figure 3). The first is a
mobile terminal with a WATM network interface card (NIC) and user-to-network interface (UNI)
software granting radio and mobility support. The second component is a BS with radio interface
capabilities and mobility enhanced UNI/network-to-network interface (NNI) software. Finally, the third is
a mobility-enhanced ATM switch also equipped with UNI/NNI software.
WATM Cells
WT1
WATM
WT2
Terminal_2
WATM Base
Station1
WATM Radio
Interface
W UNI
WATM Cells
ATM
Network
Micro Cell I
Mobile ATM
M UNI/NNI
WATM Cells
Mobility Enhanced
ATM Switch
WT1
WATM Base
Station 2
WT2
WATM Radio
Interface
W UNI
Micro Cell II
The system illustrated in Figure 3 involves two new protocol interfaces, the W UNI between
mobile/wireless user terminal and ATM base station and the M UNI/NNI interface between mobilitycapable ATM network devices, including switches and base stations. Both these interfaces are required to
support end-to-end ATM services at a mobile terminal. In particular, the WATM terminal sets up a
connection using standard ATM signaling (UNI) capabilities to communicate with the ATM base station
and network switches. All data transmitted by the WATM terminal is segmented into ATM cells with an
additional radio link level header specified within the W interface. Mobility of the WATM terminal
(i.e., handoff and location management) is handled via switch-to-switch (NNI) signaling protocol
extensions specified in the M interface [3].
User
Plane
Control
Plane
M Mobility
AAL
DLC
MAC
PHY
Wireless Control
ATM
W Wireless
Access
246
frequency division duplexing (FDD) and time division duplexing (TDD), multiplexing, e.g. frequency
division multiple access (FDMA), time division/demand multiple access (TDMA) and code division
multiple access (CDMA), and bandwidth allocation mechanism (fixed assignment, random assignment
and demand assignment) [1314].
WATM technology promises to provide QoS guarantees for multimedia applications together with
traditional services [15]. A demand assignment MAC technique for WATM should be considered to
maintain the bursty traffic natures of such applications. In this technique, a user terminal needs a control
channel in uplink direction to request an access channel from a BS. The BS then assigns bandwidth for
this request if there are enough resources to support the required level of QoS. Well-known MAC
protocols for WATM are; distributed queuing request update multiple access (DQRUMA) [16], EPRMA
[16], MASCARA [16], MDRTDMA [16], DSA++ [16], and packet reservation multiple access with
dynamic allocation (PRMA/DA) [1617].
Most of the previous MAC protocol developments were dedicated to maximize the channel efficiency,
but they were not able to meet all the QoS requirements of real-time applications in wireless networks.
In general, the channel access protocols for mobiles requesting resources have been studied widely. The
scheduling algorithms in the base station, however, still need more investigation, especially for
different QoS guarantees and traffic characteristics. Most of the available scheduling algorithms for
WATM are priority, firstin, first out (FIFO), or round-robin schemes, whereas our proposed MAC,
named MACGB (Guarantee-Based), uses a new guarantee-based scheduling algorithm to support
required level of QoS [1819].
The overall properties of the most popular WATM MAC protocols are given in [1620]. A major
disadvantage of DQRUMA is that it does not make any distinction between VBR and ABR services; it
treats both as bursty traffic. Consequently, it does not offer any priority handling mechanism. Another
disadvantage is that the use of minislots adds physical layer overhead. In MASCARA, use of variable
length frames introduces a vital difficulty in assigning capacity to WTs with CBR services.
Following sub-sections explain a common WATM MAC protocol (named PRMA/DA) and our
proposed MACGB protocol, which will be both utilized for comparative simulation studies presented
in Section Four.
3.3.1. PRMA/DA
PRMA/DA, which is proposed in [17], operates on a frame basis. Time on the uplink channel is divided
into a contiguous sequence of PRMA/DA frames, and each frame consists of available slots, CBR
reservation slots, VBR reservation slots, and ABR reservation slots as illustrated in Figure 5. The number
of available slots depends on the intensity of demand to access the network among the mobile stations. In
contrast, the number of reservation slots assigned to each reserving station is primarily dependent on the
statistical properties of traffic a MS intends to transmit. The BS is responsible for determining the number
of slots allotted to each type, as well as the number of slots assigned to each reserving terminal (MT). The
downlink frame works in the contention-free TDM format, under the total control of the BS in broadcast
mode. The main contribution of the protocol is the dynamic allocation algorithm which helps resolve the
contention situation quickly and avoids waste of bandwidth that may happen when there are several
request slots without the need for them. A drawback of this protocol is that it does not use minislots for
the access request. Instead, the first time a MT needs service, it transmits a request message along with an
information packet. Therefore, if a collision occurs, the effect on the throughput may be greater than if a
small request packet had been used. This may not be important in low traffic situations, where there may
be room for several available slots, but is definitely a problem for high traffic situations [16].
247
Frame (k-1)
Frame (k)
Frame (k+1)
Time
Slot
variable
Na
1 2
Available Slots
variable
Rc
1 2
CBR Reserv.
Slots
variable
Rv
1 2
VBR Reserv.
Slots
Rd
1 2
ABR Reserv.
Slots
Base Station
Downlink
(4 GHz)
CB D
D CB D
D CB D
Uplink
Downlink
Wireless Terminal
Uplink
(3 GHz)
CT D
1
D CT D
99
D CT D
N-100
D
N
Time
D: Data Slot
N: Number of Slot / Frame
transmission errors is also included in all WATM cells. WT creates a control packet called cc_con_end to
terminate the connection and sends it to the BS in the first available empty slot again (see Figure 7c).
TermAdress ServiceClass
a)
(4 bits)
AppNum
SLS1
SLS2
SLS3
SLS4
CRC
(4 bits)
(16 bits)
(16 bits)
(16 bits)
(16 bits)
(16 bits)
(4 bits)
GFC
VPI
VCI
PT
CLP
HEC
CRC
(4 bits)
(8 bits)
(16 bits)
(3 bits)
(1 bit)
(8 bits)
(16 bits)
PAYLOAD
(384 bits)
b)
TermAdress ServiceClass
(4 bits)
c)
(4 bits)
CRC
(16 bits)
-1
N-1 N
5
-1
-1
-1
249
WT9
WT10
WT7
Compressed
Video Transfer
WT6
WT8
WT2
WT1
WT4
WT15
WT3
Voice
Transfer
BI
Critical Data
Transfer WT13
Data
Transfer
WT5
WT14
WT11
WT18
WT17
WT12
WT16
WT20
WT19
250
traffic requiring ATM VBR service, eight of them for critical data application traffic carried with ATM
ABR service support, while the other applications are set to create data transfer traffic requiring ATM
UBR service. For instance, a voice application traffic originating from the WT1 is transferred to the WT5
over a connection providing a CBR service which is sensitive to delay and delay variation (jitter).
Similarly a data application traffic originating from the WT16 is transferred to the WT20 over a
connection providing a UBR service with no QoS guarantees. It should be noted that in a real-life
situation every WT will not usually generate data or video sources at a given time. Another WATM
model analogous to the one above except that PRMA/DA MAC protocol is utilized instead of the MAC
GB is also simulated using OPNET Modeler for consistent performance comparisons. Working
conditions of both models were chosen to be same.
4.1.1. Simulation Results and Performance Evaluation
Simulation results of the both WATM models described above are presented under varying network load
conditions, followed by performance comparisons and evaluation. The simulation utilizes a free space
channel propagation model that predicts received signal strength when the transmitter and receiver have a
clear, unobstructed line-of-sight path between them. To avoid the transient effects the simulation statistics
are flushed after approximately 10 seconds. The simulation parameters used are given in Table 2.
Traffic Sources
Uplink/Downlink Bit
Rate
Frequency Band
Transmitter Power
Modulation
Schema
CBR Parameters
30,000100,000*(Bytes/s)
25 Mbps
Uplink = 3 GHz and Downlink = 4 GHz
BS = 100 mW and WTs = 100 mW
QPSK
Error control schemes used in ATM are designed for the channel BER of 10-10. Therefore, they are not
appropriate for use in time-varying and noisy wireless channels. To provide various QoS guarantees in
WATM environments, FEC and ARQ or combination of the two are frequently used. If the BER value of
the medium is very high (e.g. 10-4), an ARQ scheme is more appropriate to employ. Otherwise, better
performance results can be obtained using FEC scheme. To focus on the performance of the MACGB
and PRMA/DA models, the channel is assumed to be ideal such that there is no distortion, noise, or other
interference for packet transmissions, resulting in low BER. It is also assumed that the CRC bits added to
the packets avoid the bit errors resulting from the physical characteristics of the wireless environment.
For both MAC protocols used in WATM, a slot length of 200 seconds which has been determined
considering 25 Mbps data rate was chosen. With a total number of 1000 slots/frame, each time slot
contained 5 WATM cells.
Varying the message size of all WT application traffics, average and maximum end-to-end delay (EED)
and delay variation results for the voice traffic transfer between WT1 and WT5, for the compressed video
251
traffic transfer between WT6 and WT10, for the critical data traffic transfer between WT11 and WT15,
and for the data traffic transfer between WT16 and WT20 have been collected during the simulation run
time for both WATM models. All of the application traffic was chosen to be equal so that the
performance of both MAC protocols and of all ATM service classes can easily be compared together. In
addition to these performance metrics examined, the network utilization is also worked out analytically
and presented in Figure 10.
In both WATM models, during a voice transfer connection between WT1 and WT5, the delay and delay
variation sensitive traffic utilizes an ATM CBR service support that guarantees data rate determined by
PCR value. During a bursty nature compressed video transfer between WT6 and WT10, ATM VBR
service class is utilized and SCR value is used to determine the amount of guaranteed bandwidth. Critical
data transfer application between WT11 and WT15 is supported by ATM ABR service and the data rate
indicated by MCR value is guaranteed for this connection. Finally, during a data traffic transfer between
WT16 and WT20, any available bandwidth unused or remaining from other service classes is utilized
over an ATM UBR connection with no QoS guarantees. The PCR parameter of the UBR service class
represents the maximum data rate that can be supported in such a given connection.
The bandwidth utilization against the total network load is drawn in Figure 10. As seen from the figure,
network utilization varies from 20% (i.e., when the offered load per application is 30,000 bytes/s) to 65%
(i.e., when the offered load per application is 100,000 bytes/s). It is obvious that the bandwidth utilization
depends mainly on the offered network load.
100
90
Utilization (%)
80
70
60
50
40
30
20
10
0
30
40
50
60
70
80
90
100
Figure 10: Average EED Results for WATM Model with the MACGB
In Figure 11, average EED results of the MACGB based WATM model are presented. Figure 12 also
shows the average EED results for the MACGB based WATM model, which are normalized with those
of the PRMA/DA MAC based WATM counterparts, as a function of the offered load per wireless
application. The voice application traffic (i.e. between WT1 and WT5) with CBR service support
experiences three to five times, the compressed video application traffic (i.e. between WT6 and WT10)
with VBR service support experiences three to five times, and the critical data application traffic (i.e.
between WT11 and WT15) with ABR service support experiences two to three times lower average
message delays in the MACGB based WATM model compared to the same traffic carried with the
PRMA/DA MAC based WATM model. However, the non critical data application traffic in the
PRMA/DA MAC based model (i.e. between WT16 and WT20) supplies three to eight times better results
than those of the WATM model using the MACGB method. This is because, in the MACGB based
WATM model the slots assigned to the UBR connections can be utilized by other services such as CBR,
VBR or ABR whereas in the PRMA/DA reserved slots of any type are used by the applications until the
end of the connection time.
252
10000
1000
100
10
30
40
50
60
70
80
Offered Load per Application (x1000 Bytes/s)
90
100
Figure 11: Average EED Results for WATM Model with the MACGB
10
0,1
30
WT1 WT6 WT11 WT16 -
40
50
60
70
80
Offered Load per Application (x1000 Bytes/s)
WT5 w ith CBR service support (Where WT1 - WT5 w ith PRMA/DA = 1)
WT10 w ith VBR service support (Where WT6 - WT10 w ith PRMA/DA = 1)
WT15 w ith ABR service support (Where WT11 - WT15 w ith PRMA/DA = 1)
WT20 w ith UBR service support (Where WT16 - WT20 w ith PRMA/DA = 1)
90
100
Figure 12: Normalized Average EED Results for the MACGB Based WATM Model
In Figure 13, maximum EED results of the MACGB based WATM model are presented. Figure 14 also
illustrates the maximum EED results for the MACGB based WATM model, which are normalized with
those of the PRMA/DA MAC based WATM counterparts, as a function of the offered load per wireless
application. As seen from the figure, the voice, the compressed video, and the critical data application
traffics in the MACGB based WATM model have better maximum EED performance than those of
PRMA/DA based WATM model. As explained in the previous paragraph, the non critical data transfer
maximum EED results of the MACGB based WATM model are also worse than those of PRMA/DA
based WATM model.
Maximum End-to-End Delay (ms)
10000
1000
100
10
30
40
50
60
70
80
Offered Load per Application (x1000 Bytes/s)
90
100
Figure 13: Maximum EED Results for WATM Model with the MACGB
253
10
0,1
30
40
50
60
70
80
Offered Load per Application (x1000 Bytes/s)
90
100
WT1 - WT5 w ith CBR service support (Where WT1 - WT5 w ith PRMA/DA = 1)
WT6 - WT10 w ith VBR service support (Where WT6 - WT10 w ith PRMA/DA = 1)
WT11 - WT15 w ith ABR service support (Where WT11 - WT15 w ith PRMA/DA = 1)
WT16 - WT20 w ith UBR service support (Where WT16 - WT20 w ith PRMA/DA = 1)
Figure 14: Normalized Maximum EED Results for the MACGB Based WATM Model
In Figure 15, EED variation results of the MACGB based WATM model are presented. Figure 16 also
shows the EED variation results for the MACGB based WATM model, which are normalized with those of
the PRMA/DA MAC based WATM counterparts, as a function of the offered load per wireless application.
Considering the delay variation sensitive voice and compressed video traffics, the MACGB based WATM
model has produced better EED variation results than those of PRMA/DA based WATM model. Lower
message delay and delay variation results are expected as the MACGB protocol, unlike PRMA/DA MAC
protocol, distributes the allocated slots according to required CDV parameter of the connection.
End-to-End Delay Variation (ms)
1000
100
10
0,1
30
40
50
60
70
80
Offered Load per Application (x1000 Bytes/s)
90
100
Figure 15: EED Variation Results for WATM Model with the MACGB
1000
10
0,1
0,01
0,001
30
40
50
60
70
80
Offe r e d Load pe r Application (x1000 Byte s /s )
90
100
WT1 - WT5 w ith CBR service support (Where WT1 - WT5 w ith PRMA /DA = 1)
WT6 - WT10 w ith V BR service support (Where WT6 - WT10 w ith PRMA/DA = 1)
WT11 - WT15 w ith A BR service support (Where WT11 - WT15 w ith PRMA /DA = 1)
WT16 - WT20 w ith UBR service support (Where WT16 - WT20 w ith PRMA /DA = 1)
Figure 16: Normalized EED Variation Results for the MACGB Based WATM Model
254
4.2. A Comparative Simulation Study of MACGB Based WATM Networks and IEEE 802.11b WLANs
The WTs in the example scenario implemented using OPNET Modeler, like the one presented in Figure
9, employ the MACGB protocol to communicate with each other in the same wireless environment. In
this scenario, there are 20 WTs on which two different applications operate to generate and receive data
traffics. There is only one type of application running at the same time on each WT. The data traffic
introduced to the network by any WT is randomly destined to another WT. Two of these applications are
set to create voice transfer traffic requiring ATM CBR service while the other applications are set to
create data transfer traffic requiring ATM UBR service. For instance, a voice application traffic
originating from the WT1 is transferred to the WT11 over a connection providing a CBR service which is
sensitive to delay and delay variation. Similarly a data application traffic originating from the WT2 is
transferred to the WT12 over a connection providing a UBR service with no QoS guarantees. It should be
noted that in a real-life situation every WT will not usually generate data and video traffic at a given time.
Another wireless networking model analogous to the one above, except that an IEEE 802.11b WLAN is
utilized instead of WATM with the MACGB based MAC, is also simulated using OPNET Modeler for
consistent performance comparisons. Working conditions of both IEEE 802.11b WLAN and WATM
models were chosen to be same.
4.2.1. Simulation Results and Performance Evaluation
Simulation results of the WATM and WLAN models described above are presented under varying
network load conditions, followed by performance comparisons and evaluation. To focus on the delay
performance of the wireless networks, the channel is assumed to be ideal such that there is no distortion,
noise, or other interference for packet transmissions. The simulation parameters used are given in Table 3.
WATM Simulation Model
IEEE 802.11b WLAN Simulation Model
20,000100,000*
20,000100,000*
Voice Traffic
(Bytes/s)
Voice Traffic Sources (Bytes/s)
Sources
20,000100,000*
Data Traffic Sources 20,000100,000*
Data Traffic Sources
(Bytes/s)
(Bytes/s)
Uplink/Downlink Bit
Uplink/Downlink Bit
11 Mbps
11 Mbps
Rate
Rate
BPSK
Modulation Schema QPSK
Modulation Schema
TDMA/FDD
CSMA/CA
MAC Protocol
MAC Protocol
PCR = 100 kbps
CBR Parameters
PCR = 100 kbps
UBR Parameters
*Generated using Exponential Distribution Function, Exp(Mean)
Table 3: Simulation Parameters
WATM networks usually support 25 Mbps data rate in micro and pico cells [3]. However, in our first
simulation model, WATM network data rate is considered as 11 Mbps to be able to fairly compare the
results obtained with those of the WLAN whose maximum theoretical data rate is 11 Mbps.
In the scenarios, WLAN and WATM use binary phase shift keying (BPSK) and quadrature phase shift
keying (QPSK) modulation schemes respectively. Carrier sense multiple access/collision avoidance
(CSMA/CA) MAC protocol used in the WLAN model is intrinsically inappropriate for the multimedia
traffic such as voice and data, which are especially sensitive to end-to-end delay and delay variation. On
the other hand, the WATM model employs the MACGB protocol with TDMA/FDD technique to
support QoS guaranteed transfer of the voice and data applications.
255
A slot length of 465 seconds is chosen for the MACGB based WATM model, which has been
determined considering 11 Mbps data rate. With a total number of 1000 slots/frame, each time slot
contains 5 WATM cells.
Varying the message size of all WT voice and data application traffics, average and maximum EED
results of the voice traffic transfer between WT1 and WT11, and EED results of the data traffic transfer
between WT2 and WT12 have been collected during the simulation run time for both WATM and WLAN
models. All of the applications traffic is chosen to be equal so that the justification of the ATM services,
each of which is utilized by a different application of the WATM end users, and WLAN service can
easily be done and compared.
In the WATM during a voice transfer connection between WT1 and WT11, the delay and delay variation
sensitive traffic utilizes an ATM CBR service support that guarantees data rate determined by PCR value
while during a data traffic transfer between WT2 and WT12, any available bandwidth unused or
remaining from other service classes is utilized over an ATM UBR connection with no QoS guarantees.
The PCR parameter of the UBR service class represents the maximum data rate that can be used for a
given connection. On the other hand, the WLAN provides QoS guarantees neither for a voice transfer
connection between WT1 and WT11 nor for data traffic transfer between WT2 and WT12. In addition, its
CSMA/CA MAC mechanism does not allow multiple connections between WTs, resulting in less utilized
bandwidth, higher access delays, and end-to-end delays for multimedia traffic.
Figure 17 shows the normalized average EED as a function of the offered load per wireless application
for the WATM model. Having normalized the average message delay results of the WATM model with
their IEEE 802.11b WLAN model counterparts, it can be justified that even in the worst case (i.e., when
the offered load per application is 100,000 bytes/s), all of the WATM traffic has better average end-to-end
message delays than those of the classical WLAN model. Of especial note, the voice application traffic
(i.e. between WT1 and WT11) with CBR service support experiences 16 to 34 times lower average
message delays in the WATM model compared to the same traffic carried with no service guaranties in
the WLAN. Considering the fact that WATM UBR and WLAN service classes are almost identical, it
would be expected that the average message delay for the WATM data application traffic (i.e. between
WT2 and WT12) with UBR has almost the same characteristics as that of the WLAN model. However,
three to 12 times improved results are obtained in the former using the MACGB protocol. This is due to
the fact that connections with UBR service support can utilize any available bandwidth unused by other
connections with CBR, VBR or ABR service support, and end user applications with this service can be
active together with any other.
0,1
0,01
20
30
40
50
60
70
Offered Load per Application (x1000 Bytes/s)
80
90
100
WT1 - WT11 w ith CBR service support (Where WT1 - WT11 w ith no service support in WLAN = 1)
WT2 - WT12 w ith UBR service support (Where WT2 - WT12 w ith no service support in WLAN = 1)
Figure 17: Normalized Average End-to-End Message Delays for WATM Model
256
Figure 18 illustrates the maximum EED results for WATM model, normalized with those of the IEEE
802.11b WLAN counterparts, as a function of the offered load per wireless application. As seen from the
figure, both the voice application traffic (i.e. between WT1 and WT11) and data application traffic (i.e.
between WT2 and WT12) outperform those of the classical WLAN model. Up to 13 times better
maximum EED results for the voice traffic (i.e., when the offered load per application is 40,000 bytes/s)
can be well achievable with the MACGB based WATM model for both application traffics.
0,1
0,01
20
30
40
50
60
70
80
Offered Load per Application (x1000 Bytes/s)
90
100
WT1 - WT11 w ith CBR service support (Where WT1 - WT11 w ith no service support in WLAN = 1)
WT2 - WT12 w ith UBR service support (Where WT2 - WT12 w ith no service support in WLAN = 1)
Figure 18: Normalized Maximum End-to-End Message Delays for WATM Model
5. Conclusions
WATM differs from its traditional counterparts mainly in that it promises to provide QoS guarantees for
real-time multimedia applications. QoS guaranteed data transfer in a wireless environment entails a
special MAC protocol allowing efficient allocation and utilization of the scarce and error-prone resources
to different end user applications. The fundamental concept developed and presented in this paper
establishes the way in which improved performance can be obtained in wireless environments, bringing
different service classes to desktops using WATM.
In this paper, an example WATM network model equipped with the MACGB protocol has been realized
using OPNET Modeler and simulated under various traffic load conditions. The simulation results are
presented together with comparisons those of a PRMA/DA MAC based WATM counterpart. According
to the simulation results obtained, especially for the connections utilizing CBR, VBR, and ABR services,
average EED, maximum EED, and EED variation results of the proposed MAC based WATM model are
considerably better than those of PRMA/DA based WATM model, as a consequence of new guarantee
based scheduling algorithms introduced in the BS. The simulation results are also compared those of an
IEEE 802.11b WLAN counterpart with no QoS guarantees. Not only does the WATM employing MAC
GB method enable wireless end users with ATM CBR, VBR, ABR and UBR connections, but also
provides over 16 times better average end-to-end message delay results than those of IEEE 802.11b
counterpart for application traffics requiring a CBR service class support.
References
[1]
[2]
B. Bing, High-Speed Wireless ATM and LANs, Artech House Publishers, 2000.
V. B. Houdt, C. Blondia, O. Casals, and J. Garcia, Performance Analysis of a MAC protocol for
Broadband Wireless Networks with QoS Provisioning, Journal of Interconnection Networks,
Vol.2, pp.103130, 2001.
257
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
258
259
260
Chandrakantha T.N.
In the transmit path, a very clean vector modulator would be required, as no narrowband filtering is
carried out. RF filtering is required to minimize out-of-band noise (such as receive band). In the receive
path, direct conversion is used to down-convert to direct current (DC). RF filtering is no longer required
for image rejection, but is still required to reduce the level of interferers to prevent compression in the
local network attachment (LNA) and mixer. Gain and filtering is carried out at DC. Increased dynamic
range of ADCs however means that more of this filtering can be achieved in the digital domain. The main
issue with direct conversion is that DC offsets can occur due to leakage and second order distortion. This
could saturate the following gain stages. Two solutions are to either use a very low IF receiver (although
this also has inherent problems) or to perform some form of DC correction.
As simultaneous multimode operation is required, multiple ADCs and DACs will be required to interface
the transceiver. To keep the power consumption to a minimum, the dynamic range requirement should be
kept as low as possible. Different sample rate and resolution ADCs/DACs will almost certainly be
required for each mode.
Multimode Transceiver
Without even considering coexistence, the transceiver must be able to meet the relevant performance
targets for a particular mode. In terms of transmission, this includes maximum output power levels and
transmitted mask. In terms of the receiver, issues of particular importance are sensitivity and blocking
performance. For a combined WLAN and multimode cellular mobile terminal, leakage of amplified
transmit signals into other part of the RF system will cause system degradation. It will be assumed that
the main interfering signals will be the high power outputs from the PA stages, which vary from 24 dBm
(WLAN) to 33 dBm (GSMEDGE 900).
Consider a typical transceiver front-end (see Figure 4). The only real leakage path to consider is that of
the power amplifier output being injected into the receive path. This is a function of the isolation of the
front-end module.
261
262
Chandrakantha T.N.
mobile phone system (AMPS) signal. Solving this problem relies on using a high isolation front-end
duplexer and careful board layout. This problem does not occur with GSM as transmission and reception
occur in different time slots. Table 1 lists the modes that are required to be able to receive in the presence
of transmit signals.
263
Figure 8
A prerequisite for all these components and paths to be included in a single antenna output module is
additional switching or use of diplexers to combine the antenna outputs, which would in turn lead to
increased loss.
Fabrication
To produce a mobile terminal that is of acceptably low cost, some form of integrated chipset will have to
be developed. It is anticipated that the RF transceiver could be enclosed in a single package. This does
raise the issue of isolation within the integrated circuit however (see Figure 8).
264
Chandrakantha T.N.
265
References
[1]
[2]
[3]
266
ANSI/IEEE Std. 802.11, Wireless LAN Medium Access Control (MAC) and Physical Layer
(PHY) Specifications
Std. 3GPP TS 45.005 V5.2.0, Technical Specification Group GSM/EDGE; Radio Access
Network; Radio Transmission and Reception (Release 5)
CommsDesign Article Direct Conversion: No pain, No gain, http://www.commsdesign.com/
Overview of WLAN
A wireless local area network (WLAN), as the name suggests, is a data communication system that has been
employed as alternative method to implement the wired LAN network within limited geographical areas.
These areas can be as large as shopping complexes, educational establishments, or even a certain section of
a municipality. Recently the WLAN services have extended to airports. As in any other wireless radio
systems, WLAN utilizes electromagnetic signals to convey information. The spectrum of the
electromagnetic signals is well within the range defined for the operation, by the relevant authorities. Mobile
network operators are very much interested in the use of WLAN technologies because the frequency bands
associated with these technologies are licence free radio spectra. The WLAN system can totally eliminate
the usage of cables for its communication link requirements. The WLAN system has, thus, gained its
popularity in various scenarios like business, educational, medical and other engineering applications.
A network where a mobile user can connect to a local area network using wireless communication is
generally referred to as WLAN. 802.11 is a family of specifications for WLAN developed by a working
group of the Institute of Electrical and Electronics Engineers (IEEE). The different specifications in the
family are 802.11, 802.11a, 802.11b, 802.11g, 802.11i and 802.1x. Of these, 802.11, 802.11a, 802.11b are
fully developed and commercial systems based on them are available in the market, while 802.11i and
others are in different stages of development. 802.11, 802.11a, 802.11b and 802.11g use the carrier sense
multiple access with collision avoidance (CSMA/CA) for path sharing across various nodes in the network.
There are mainly two WLAN standards. They are IEEE802.11, the U.S. standard and high performance
local area network (HiperLAN), the European standard.
The first and original 802.11 specifications were put forward in 1997 by the IEEE as a WLAN standard.
This version allowed the one Mbps and two Mbps data rate transmission at 2.4 GHz with the related
supporting signalling protocols. However, the latest versions of standards have come up with transmission
with increased data rates that operate at different center frequencies, up to 54 Mb/s at five GHz band, with
increased reliability of data. HiperLAN also provides similar types of data rates at the given frequency.
However the coverage of WLAN extends to about a few hundredths meters up to 300 meters and hence
can only be used in highly dense areas.
The modulation used was either frequency hopping spread spectrum (FHSS) or direct sequence spread
spectrum (DSSS). FHSS is the repeated switching of frequencies during radio transmissions, often to
minimize the effectiveness of any attempts on interception or jamming of the signal.
Wireless fidelity (Wi-Fi) is a cheap and fast way of connecting computers to each other using high
frequency wireless communication in a local area network. The term Wi-Fi is normally used
interchangeably with IEEE 802.11b specification. 802.11b networks operate in the unlicensed 2.4 GHz
radio bands, with an 11 Mbps data rate that supports fallback to 5.5, two and one Mbps for backward
compatibility with 802.11.
IEEE 802.11b uses direct sequence spread spectrum (DSSS) encoding. DSSS works by taking a stream of
ones and zeroes and modulating it with a second pattern called the chipping sequence. In 802.11 this
sequence is known as the Barker code, an 11 bit sequence (10110111000) that is ideal for modulating
radio waves due to certain mathematical properties. Complementary code keying (CCK), used in 802.11b
employs a series of codes called complementary sequences, which allows higher data speeds and is less
susceptible to interference in multi-path propagation. Details on modulation schemes can be found at [2].
As compared to 802.11b, 802.11a works in the unlicensed five GHz radio band. 802.11a uses a
modulation scheme called the orthogonal frequency division multiplexing (OFDM). This modulation
268
scheme makes it possible to achieve data rates as high as 54 Mbps. Note that since 802.11a works in the
five GHz range, 802.11a is not backward compatible with 802.11b hardware. While 802.11g provides the
same throughput (54 Mbps) as the 802.11a, it also provides backward compatibility to the currently
dominant 802.11b standard as it works in the 2.4 GHz radio band. Because WLANs use a wireless
medium for communication, which is essentially a broadcast, security in such networks becomes critical.
Antennas for WLAN and General Concepts
The primary challenges for WLAN antennas are related to signal propagation. These mainly involve the
indoor environment where most of these antennas must operate. Some of the characteristics of this
transmission medium include the following:
269
Adaptive antenna arrays are an active research topic nowadays due to the improvements and advantages
over omni-directional and directional antenna systems. A smart antenna system is defined by the IEEE as
an antenna system that has circuit elements associated with its radiating elements such that one or more of
the antenna properties are controlled by the received signal. In these systems, each transmitter located at a
certain place has its unique pattern, which is also called spatial signature.
Classification of Adaptive Antenna Arrays
There are several techniques of implementing smart antennas [4]. There are basically two types of smart
antennas: switched beam systems and adaptive antenna array systems. The switched beam system
comprises only basic switching between separate directional antennas or predefined beams of an array
while enabling high directivity and gain. Switched beam systems can be further divided into two groups:
single beam and multi beam directional antennas. In single beam directional antenna systems, only one
beam is active at a given time. No simultaneous transmissions are allowed, because in this system there is
only one transceiver. Multiple beam directional antenna system is an example of spatial division multiple
access (SDMA) system. Here, each directional antenna can be used and transmissions are allowed at the
same time and frequency. The number of beams is equal to the number of transceivers.
In the adaptive antenna arrays, a direction of arrival (DOA) algorithm is employed to determine the
direction of the signal received from the mobile subscriber. In this method, the continuous tracking of
users is done by appropriately following the DOA information. Provisions can be made to the main
algorithm for the detection of the interferers, so that any interference can be cancelled by placing the
radiation pattern nulls in the direction of interferences to increase the signal to interference and noise ratio
(SINR). Operationally, the adaptive antenna array beamforming is more complex than switched beam
systems, due to the complex algorithms involved. Adaptive antenna array beamforming can be divided
into two sections: single user beamforming and multiuser beamforming.
In single user beamforming, the antenna beam is adjusted to track a mobile subscriber and to cancel any
interferers. In this case, a single transceiver is sufficient where only one user is active at a given time. In
multiuser beamforming, there are different beam patterns, and each beam tracks any one mobile
subscriber. Hence, simultaneous transmissions are allowed and SDMA is achieved. There is more than
one transceiver-beam pair in multi user beam forming.
In this paper, emphasis is given to the adaptive antenna array beamforming methods.
In adaptive antenna array signal processing [5], a weight vector is chosen for the array processor in order
to optimize the desired signal according to a given criterion. This can be done through a priori knowledge
of the desired signals direction of arrival or through use of an adaptive algorithm, which uses any known
information about the signals characteristics to maximize the signal to noise ratio (SNR) or signal to
interference and noise ratio (SINR).
The radiation pattern of an array is determined by the radiation pattern of the individual elements, their
orientation and relative positions in space, and the amplitude and phase of the feeding currents. If each
element of the array is an isotropic point source, then the radiation pattern of the array will depend solely
on the geometry and feeding current of the array, and the radiation pattern so obtained is called the array
factor. If each of the elements of the array is similar but non-isotropic, by the principle of pattern
multiplication, the radiation pattern can be computed as a product of the array factor and the individual
element pattern.
271
272
Disadvantages
The main drawback of smart antennas is in their design and their implementation in hardware [6].
Multiple radio frequency (RF) stages can increase the cost and the bulk of the transceiver. Most of the
baseband processing requires coherent signals. This requires that all the signals for various electronic
circuits have to be tapped from the same source.
These difficulties can pose severe challenges in the design of the circuit. The phase characteristics of RF
components can change over time. These changes are relatively static and hence need calibration
procedures to account for phase differences.
Most of the devices used in the RF stages are non-linear. Using adaptive antenna arrays can increase the
number of such components used. This can affect the performance of the array if not checked
periodically. Furthermore, since adaptive antenna arrays use more than one source of signal, the data
bandwidth required for digital processing increases linearly with the number of antenna elements used.
This can limit data rates for different applications.
The use of directional antennas and 1200 sectors has become widespread in cellular systems and can
increase capacity up to three times for WLAN systems, though with an added cost and with the added
complexity, especially in indoor environments, from sectorization. Access points and base stations can
easily host antenna arrays of four or more elements but with existing microstrip or patch antenna
technology, up to three elements can be fitted in a handset. The wrapping of the hand around a handheld
device may diminish the performance of a handheld smart antenna system.
Adaptive Antenna Array Solution for WLANBased Broadband Access
The future of broadband access will face a number of challenges for its smooth deployment and
operation. The networks must provide wireless communication services at a lower cost and
simultaneously support many new types of revenue-generating applications, demanding high bandwidth
and real-time traffic handling.
In adaptive antenna array applications, the objective is to maximize signal-to-noise ratio (SNR) of the
communication in order to reduce error rates and, in wireless communications, better utilize the
capacity of the channel, in a particular direction while steering the nulls to the directions of
interference. In an environment where the noise characteristics change, the system uses adaptive
processing techniques to adjust the complex weighting (amplitude and phase) of each antenna element
to optimize the overall array pattern.
A general view of the implementation of adaptive antenna arrays for WLAN applications is presented
here. This paper has demonstrated the feasibility of using adaptive antenna arrays at base stations for
WLAN networks. Adaptive antenna arrays are superior as the technology for broadband access networks,
due to their efficiency and other inherent advantages.
References
S. Chandran (Ed.), Adaptive Antenna Arrays: Trends and Applications, Springer-Verlag, Berlin
Heidelberg New York, 2004.
802.11b Wireless LAN, http://www.vocal.com/data sheets/full/802.11b.pdf
Mailloux, Robert J., Phased Array Antenna Handbook, Artech House, Boston, 1994.
273
La Porta, Thomas F. (Editor), et al., IEEE Personal Communications Special Edition on Smart
Antennas, Vol. 5, No. 1, Feb. 1998.
Haykin, Simon, Adaptive Filter Theory, 3rd Ed., Prentice-Hall, 1996.
Ramesh Chembil Palat, Dr. Raqibul Mostafa, Dr. Jeffrey H. Reed, and Dr. Seungwon Choi, Smart Smart
Antennas: A System Level Overview for Software Defined Radios for Creating an API, Software
Defined Radio Forum Contribution, SDRF-04-I-0057-V0.00, Dated: 6-1-2004.
274
Prashant Vashisht
Senior Technical Leader
Hughes Software Systems
Masood Ul Amin
Project Manager
Hughes Software Systems
Introduction
There have been very few bright spots on the horizon of the networking industry in the last two years, and
the one that exhibits tremendous potential is the wireless local-area network (WLAN). Just to have an
insight into the growth of WLAN, consider the following:
A study by Gartner group indicates that nearly 50 percent of company laptops around the world
will be having support for WLAN by 2006.
The overall WLAN market grew around 200 percent from 2000 to 2002.
There are over 10 million WLAN users globally, and this figure is expected to double in 2004.
Until recently, most of the business applications for WLAN were limited to industries comprising of
mobile workers (e.g., retail, transportation etc.) but more recently wireless is gaining quite a fanfare
among small office/home office (SOHO) users. This trend could be attributed to the lowering of
equipment costs and better support for security, reliability, and scalability as provided by WLAN. Today,
WLAN are fast becoming a more obvious connectivity alternative for a broad range of business
requirements and markets.
Overview
The first WLAN standard was ratified in 1997, and since then there has been no looking back for
WLAN. The three biggest challenges to growing popularity of the WLAN deployment are security,
reliability and scalability.
275
This paper presents an overview of the WLAN world with respect to the challenges that are being faced
and the developments that are happening to counter those challenges. The paper starts off with a brief
coverage of the security aspects of the WLAN and then gradually dwells upon the reliability and
scalability challenges posed to the WLAN deployment.
WLAN Security Overview
WLAN inherently being a broadcast nature requires user authentication and data encryption to provide
security. Each wireless network has two major components, either stations/clients (STA) or access points
(AP). A client and an access point must establish a relationship prior to exchanging data. Once
established the client-access point relationship could be in any of the following three states:
The exchange of real data is only possible in the third state. Until then the parties communicate using
management frames. Access point transmits beacon management frames at fixed intervals. Client receives
this frame and starts authentication by sending an authentication frame. After successful authentication
the client sends an association frame and the access point responds with an associated response frame.
802.11 Authentication Methods and Weaknesses
The 802.11 specification specifies following mechanisms for authenticating wireless LAN clients:
Open authentication
Shared key authentication
Media access control (MAC) address
Open authentication uses a null authentication algorithm. Open authentication allows any device network
access. If no encryption is enabled on the network, any device that knows the service set identifier (SSID)
of the access point can gain access to the network.
Weakness of Open Authentication
The weakness with the open authentication is that it relies on SSID to grant access The SSID is advertised
in plain text in the access point beacon messages. An eavesdropper can easily determine the SSID with
the use of an 802.11 wireless LAN packet analyzer. The SSID can also be determined by sniffing the
probe response frames from an access point.
Shared Key Authentication
Shared key authentication requires that the client configure a static wired equivalent privacy (WEP) key.
The client sends an authentication request to the access point requesting shared key authentication. The
access point responds with an authentication response containing challenge text. The client uses its locally
configured WEP key to encrypt the challenge text and reply with a subsequent authentication request. If
the access point can decrypt the authentication request and retrieve the original challenge text, then it
responds with an authentication response that grants the client access.
Weakness of Shared Key Authentication
Shared key authentication requires the client use a preshared WEP key to encrypt challenge text sent from
the access point. The access point authenticates the client by decrypting the shared key response and
276
validating that the challenge text is the same. The process of exchanging the challenge text occurs over
the wireless link and is vulnerable to a man-in-the-middle attack. An eavesdropper can capture both the
plain-text challenge text and the cipher-text response. WEP1 encryption is done by performing an
exclusive OR (XOR) function on the plain text with the key stream to produce the cipher-text. Now if the
XOR function is performed on the plain text and cipher-text are XORed, the result is the key stream.
Therefore, an eavesdropper can easily derive the key stream just by sniffing the shared key authentication
process with a protocol analyzer.
MAC Address
MAC address authentication verifies the clients MAC address against a locally configured list of allowed
addresses or against an external authentication server. MAC authentication is used to augment the open
and shared key authentication mechanisms provided by 802.11, thereby reducing the likelihood of
unauthorized devices accessing the network.
Weakness of MAC Address Authentication
MAC addresses are sent in the clear as required by the 802.11 specification. As a result, in wireless LANs
that use MAC authentication, a network attacker might be able to subvert the MAC authentication process
by spoofing a valid MAC address. MAC address spoofing is possible in 802.11 network interface cards
(NIC) that allow the universally administered address (UAA) to be overwritten with a locally
administered address (LAA). A network attacker can use a protocol analyzer to determine a valid MAC
address and use a LAAcompliant NIC with which to spoof the valid MAC address.
WEP
WEP was first designed by the authors of the 802.1 standard. WEP provides security by encrypting data
over radio waves. WEP is used to prevent unauthorized access to the wireless network.
The WEP protocol relies on a secret key that is a preshared key. This key is used to encrypt data packets
before they are transmitted, and an integrity check is run on them. WEP uses the RC4 algorithm, which is
a stream cipher algorithm. A stream cipher expands a short key into an infinite pseudorandom key stream.
To remove the venerability in 802.11 authentication and data privacy issues many vendors have come up
with solutions that augment the 802.11 security by implementing enhancements to 802.11 authentication
and encryption mechanisms. The subsequent sections details some such mechanisms
802.1X Authentication/EAP
Recent enhancements to the 802.11 standard have supplemented the standard to include 802.1Xauthentication framework at the link layer. The IEEE 802.11 Task Group i defines an approach to
provide end-to-end framework using 802.1 and extensible authentication protocol (EAP).
This approach provides for mutual authentication between client and authentication server followed by
mechanism to derive the encryption keys after authentication. It also provides for a centralized policy
control to trigger reauthentication and new encryption key generation on session timeout. 802.1X
authentication mechanism involves following three entities:
The access server creates a logical port per client, based on the clients association ID (AID). To create a
data path between the client and access server requires successful authentication of the client. The client
gets activated on the medium and associates to the access point. The access server detects the client
association and enables the client port. It forces the port into an unauthorized state so that only 802.1X
traffic is forwarded. All other traffic is blocked.
The client sends an EAP start message, although client initiation is not required. The access server replies
with an EAP request identity message back to the client to obtain the clients identity. The clients EAP
response packet containing the clients identity (user ID, password/digital certificate) is forwarded to the
authentication server (remote authentication dial-in user service [RADIUS] server).
The authentication server is configured to authenticate clients with a specific authentication algorithm.
Currently, 802.1X for 802.11 LANs does not stipulate a specific algorithm to use.
The result is a RADIUSACCEPT or RADIUSREJECT packet from the RADIUS server to the access
point. Upon receiving the RADIUS ACCEPT packet, the authenticator transitions the clients port to an
authorized state, and traffic may be forwarded.
This is followed by the RADIUS server and client mutually generating and sending the WEP key called
session key to the access point. The access point encrypts its broadcast key using the session key and
sends it to the client. The client uses the session key to decrypt it. On successful decryption the client and
the access point activates this WEP key and use it for all the communication. The radius server specifies a
session time out for the session key to retrigger its distribution.
Figure 1 shows the steps involved in 802.1x/EAP exchange.
5 client authenticates
ok
1 clent
associates
with access lan
wireless
device
2 access point
blocks all
requests
wireless
medium
4 credentials verified
from RADIUS
server
Radius
server
6 client
port
enabled
Access
device
3 login
8 deliver session
authentication
key to access
credentials
point
9 deliver broadcast
key encrypted using
session
key
7 derive session
key
Access
switch
278
EAPTLS
EAPTLS is an Internet Engineering Task Force (IETF) standard (RFC 2716) that is based on the TLS
protocol (RFC 2246). EAPTLS also involves the mechanism of authentication followed by key
exchanges, as in 802.1x/EAP. The essential difference is that EAPTLS uses digital certificates for both
user and server authentication.
In the first phase of the exchange the RADIUS server sends its digital certificate to the client. The client
verifies the certificate by verifying the contents of the certificate and the issuing authority. In the second
phase the client sends its certificate to the RADIUS server. The RADIUS server validates the clients
certificate by verifying the contents and the issuer.
When the verification succeeds at the RADIUS server then the server sends an EAPsuccess message to
the client. The client and the RADIUS server then derive the dynamic WEP key.
Figure 2 shows the steps involved in the EAPTLS exchange.
6 Client and Access point
activate the WEP key
4 RADIUS server delivers session (WEP key)
to access point
1 clent
associates
with access lan
wireless
device
2 access point
blocks all
requests
wireless
medium
Access
device
Radius
server
Access
switch
systems, however, require a very high degree of reliability in operations of the devices. It is expected that
reliability of WLAN network should be comparable to wired LANs, if WLAN community wants to keep the
same growth rate in future and forays into new area of life critical appliances.
The WLANs have difference of protocol layers at physical and data-link layers (as per the open systems
interconnection [OSI] layer model). WLANs use spread spectrum techniques for communication. The
technology facilitates multiple users to share the spectrum provided the power is limited to 1 watt as per FCC
regulations. IEEE 802.11 has given provisions for two spread spectrum techniques for communication:
In evolution of wireless LAN standards, DSSS was chosen as means of communication for its better
throughput, longer outdoor range and efficiency. All wireless LAN offers an ease of mobility, however,
the reliability of the network heavily depends on the surroundings and data speed degrades as the distance
between wireless capable device increase. The radio frequency (RF) channel interference poses a major
challenge for reliable WLAN network. The 2.4 GHZ industry, science, and medicine (ISM) band is very
crowded and being shared with cordless phones, microwave ovens, and Bluetooth-supported devices. The
other emerging technologies also compete for the already crowded spectrum.
RF Channel Interference and Sources
RF interference involves the presence of unwanted, interfering RF signals in same frequency band;
802.11 operates in listen-before-talk mechanism. Each node transits packets when there is no other node
transmitting. Nodes interested in transmission wait for a free medium. An interfering RF signal of
sufficient amplitude and frequency, however, can appear as a false signal, appearing to be emanating from
an 802.11 compliant node. This causes legitimate nodes to wait for an indefinite time, as long as the
interfering signal exists. The interfering signal not only affect the nodes that are schedule for transmission
but it affects the in process packet transmission. The transmitting packet is dropped at receiving end
because of bit errors. The packet error is recovered by retransmission of the packets, which is overhead
network traffic. If the interfering signal continues in the access medium then protocol switches to lower
data rates. The lowering of data rate and channel reliability can be explained in terms of the well-known
Shannon law of maximum information transmission. In communications theory, Shannons Limit
theoretically predicts that a system can transmit a maximum of 2BW symbols per second, where BW is
the channel bandwidth. The actual capacity depends on the signal to noise ratio of the channel. In
802.11systems, as the S/N gets deteriorated the system falls back to simpler modulation the system falls
back to simpler modulation formats that have more noise tolerance.
There are several sources of interfering signals, including microwave ovens, wireless phones, baby
monitors, Bluetooth-enabled devices, and other wireless LANs. The most damaging of these are 2.4 GHz
wireless phones that people are starting to use in homes and some companies. The proximity of these
phones within the same room as an 802.11b wireless LAN severally degrades the performance and
reliability. A brief description of the RF interference sources can be found in the following paragraphs.
Microwave Ovens
Microwave ovens are the major source of interference as they also share the 2.4-Ghz ISM band.
Microwave ovens operating within 10 feet or so of an access point or radio-equipped user will generally
cause 802.11b performance to drop significantly.
280
281
Lack of security.
Low interoperability
No QoS support for multimedia/Voice traffic support
The ability to manage users or to have a control over the network was minimal and there was quite little
support to avoid bandwidth hogging.
As the popularity of the WLANs grew, the users and developers of the WLANs gradually became aware
of the issues involved in the implementation and usage of WLANs, and this lead to the evolution of the
next generation of WLANs. This generation consists of a series of supplement standards and
enhancements to the initial 802.11 protocol to make it more robust, secure, and manageable with quality
of service (QoS) support.
The following is a brief list of the various standards that have been formulated to achieve the abovementioned aspects:
Having had a brief insight into the evolution of WLANs, we would now delve more into the scalability
aspect of WLANs. The scalability of a WLAN is usually measured on the following benchmarks:
282
Coverage area: This indicates the area of influence of a WLAN in which data transmission is
feasible with the data integrity and correctness being intact. The coverage area could be increased
with the use of multiple successive access points.
User count: This factor indicates the maximum number of users that could be supported by a
WLAN without compromising on the quality of data transmission.
Performance: This aspect of scalability covers mainly the bandwidth utilization, data
transmission rate and the errors and overheads involved in a particular WLAN technology.
One of the key aspects of WLAN that is under continuous focus is scalability. The scalability of initial
technologies in WLAN world left a lot to be desired but the pioneers of the wireless world are leaving no
stone unturned to make WLAN a integral part of communication world.
Among the various WLAN options available to us, we are going to discuss 802.11a, which is one of the
latest advancements in the wireless world and how it scores better than previous WLAN technologies
(especially 802.11b).
802.11a: A Scalable WLAN Solution
The latest offering of IEEE to help design and implement a highly scalable, robust, and reliable WLAN is
802.11a. In this particular section we would mainly focus on the scalability aspects of 802.11a that makes
it such an obvious choice for supporting the ever-growing wireless networks.
The 802.11a standard supports many more channels than its predecessor, i.e., 802.11b. Whereas 802.11b
supports 3 nonoverlapping channels, with 11 Mbps capacity each, 802.11a supports 8 nonoverlapping
channels, with 54 Mbps capacity each (approximately 432 Mbps in a given coverage area). Hence there is
a significant increase in the available bandwidth.
The 802.11a WLAN utilizes 300 MHz of bandwidth in the 5GHz spectrum in unlicensed national
information infrastructure (UNII). The available bands are further divided into different realms and are
useful for various in-building or building-to-building communications depending upon the power outputs
of the individual bands. In contrast, only 83 MHz of spectrum in ISM band is available to 802.11b and
that too is a shared band with other devices like cordless phones and microwave ovens etc.
The above two enhancements in 802.11a drastically improve the scalability of the 802.11a as compared to
802.11b. Another aspect, which enhances the scalability of 802.11a, is the orthogonal frequency division
multiplexing (OFDM) used by it, unlike 802.11b, which uses direct sequence spread spectrum (DSSS).
OFDM serves a two-fold purpose, i.e., besides providing higher data rates, it also improves the signal
distortion and transmission echo.
The 802.11a standard has 8 channels of 20-MHz width with 52 subcarriers inside each channel giving its
scalability a great boost. The squeezing of so much of information per transmission might lead to data
loss and the designers of 802.11a have envisioned such scenario and have provided the feature of forward
error correction (FEC) in 802.11a. FEC consists of sending a secondary copy of data along with the
primary data and if some primary data is lost then the lost data could be recovered by the receiver using
the secondary copy, thus avoiding the need for retransmissions.
The data speeds available in 802.11a are 6, 9, 12, 18, 24, 36, 48, and 54 Mbps and 802.11a has the
provision of auto-rate scaling down from 54 Mbps depending upon the distance and the network load.
283
284
Anand R. Prasad, Wireless LANs, Status Today and Visions for Future.
Joanie Wexler and Steven Taylor, Wireless LAN, State of the Market Report, February 2004.
Aaron Vance, WLAN: Trend and Analysis.
Intel, 802.11a Scalable 5 GHz Wireless LAN.
Louis E. Frenzel, Two-Chip Transceiver Set Will Eliminate 802.11b/Bluetooth Interference.
Steve Thomas, Interference Detection and Mitigation.
Philips, How 802.11b/g WLAN and Bluetooth Can Play Together.
Interference and signal propagationWi-Fi signals may be subject to interference from cordless
telephones, microwave ovens, Bluetooth devices, remote garage door controlsand may also be
impacted as they penetrate buildings and walls
285
Wi-Fi link distances are dependent on deployment specifics, but in general, are as follows:
o
Greater link distances may be achieved with smart, high-gain antenna systems, but this adds
complexity and costs to the system and is more likely to be used in commercial rather than
consumer deployments.
The main elements of a basic wireless LAN include the following: a Wi-Fi LAN card for the laptop PC or
other portable device; an access point that transmits the wireless signals to multiple user PC Wi-Fi LAN
cards; and a broadband access connection from a wired network into the access point, such as digital
subscriber line (DSL) or broadband cable. In addition, the user PC may have client software to provide
authentication and other security functions. Other elements may include a secure server (for Web or email access) and a router/firewall. These elements could be part of a standalone wireless network or part
of the wired network that the access point interconnects to. It should be noted that some of the newer
laptop PCs might have Wi-Fi LAN capabilities built into the PC chipsets, as opposed to separate cards.
Typical wireless LAN configurations, depicted in Figure 1, include the following:
1.
2.
3.
4.
WAN
2- 4 Sq Miles
WAN
Access
Points
Gateway
WAN
Campus Coverage
Bridging
Server
286
Firewall
Ron Landi
Basic security features (i.e., suppress broadcasting of system IDs) often are not enabled by the users
Data and management frames and headers are transmitted in clear-text mode, raising the potential
for the capturing of MAC addresses and encryption information (i.e., public keys)
Present standards do not support mutual authentication between access points and users
802.11 wired equivalent privacy (WEP) encryption, the existing security standard, is not secure
private keys are static and WEP employs a simple encryption methodology
Some examples of possible Wi-Fi security breaches include the following: war drivers who target
and identify open or uncontrolled systems; network attackers who pose as valid users, gaining access
via frame spoofing and session hijacking techniques; and rogue access points, resulting in hot-spot
forgeries. To combat these security weaknesses, a number of current solutions can be implemented.
Users can implement the Wi-Fi Alliances wireless protected access (WPA)this is an interim
solution until the IEEE 802.11i standard is completed. WPA provides port control and dynamic key
distribution features to enhance security capabilities. Enterprise customers can use virtual private
network (VPN) client software to create secure IP tunnels across the wireless LAN. In addition,
additional security features can be enabled in the access points to enhance encryption and frame
control functions. The longer-term solution to Wi-Fi security issues is the implementation of the
802.11i standard. It should be noted that this standard was just approved by the IEEE in late June
2004. The 802.11i security specification replaces WEP and will provide a higher level of security.
Some of its features include: access control (mutual authentication); improved encryption key
management; frame authentication; advanced encryption standards (AES); and isolation of the
wireless LAN from the wired infrastructure. The transition from existing Wi-Fi security methods
(WEP, WPA) to equipment that supports 802.11i and AES will take time and will require hardware
and software upgrades.
1.3 Wi-Fi Market Overview
The wireless LAN market is very fragmented, with 802.11 becoming the dominant wireless LAN
solution. Equipment and chip vendors are building 802.11 interfaces directly into devicessuch as
laptop PCs, personal digital assistants (PDAs) and wireless 2.5G/3G telephones. The increased
volume of 802.11 devices, in addition to the entry of multiple vendors into the equipment market, has
resulted in lower equipment costs. Wi-Fi equipment is now within the reach of the consumer markets,
being deployed in the home and the small office/home office (SOHO) environments. In addition,
cellular and personal communications system (PCS) service providers are beginning to integrate
802.11 capabilities into their wireless servicesoffering PCS/digital services and Wi-Fi capabilities
in hot spots or zones. The IEEE is addressing critical areassuch as security, quality of service,
streaming data, roamingto enable new Wi-Fi services. Wi-Fi has attracted the attention of major
vendors, service providers and venture capital funds. While the Wi-Fi market has received much
287
attention and publicity in recent yearswide-scale commercial system deployments, solid revenue
estimates, and compelling business cases for wireless LAN services have yet to be seen. Some of the
general market drivers for Wi-Fi include the following: cost and convenience over wiring; the
extension of and convenience for enterprise systems; enhanced productivity for personnel working in
the field; and public Internet connectivity, via hot spots. Presently, in the United States, there are an
estimated 30,000 to 40,000 hot spotsmainly in retail outlets, hotels, and airports; in addition, the
number of free community hot spots, funded in part by local government entities, continues to grow.
The estimated number of wireless LAN users, accessing these hot spots, is estimated at between 13
and 15 million users.
Some of the challenges faced by Wi-Fi include the following: proprietary vendor solutions add
complexity and lessen plug-and-play installation opportunities; standardized roaming does not exist;
there is a lack of strong quality-of-service capabilitieslimiting applications such as streaming video
and voice over the Internet (VoIP); there are no external network management capabilities; available
bandwidth is shared and could limit the ability of wireless LANs to scale and grow; and the IEEE WiFi standards are still being worked, potentially resulting in the change out of early adopter equipment
and the use of proprietary/noninteroperable solutions.
1.4 Wi-Fi Network Configurations
There are three wireless LAN options of interest to the communication carrier. These alternatives are as
follows:
The incumbent local exchange carrier (ILEC) and interexchange carrier (IXC) service providers have
shown interest in the enterprise wireless LAN alternative. The EWLAN option allows them to extend
their enterprise and business customers wired LANs. The enterprise wireless LAN becomes a secure
extension of the wired customer premise network. The service providers can offer managed services, in
addition to bundles of current and future servicessuch as frame relay access, Internet, firewall, and
managed data servicesover the EWLAN. Refer to Figure 2 for a depiction of an enterprise wireless
LAN with multiple access points.
288
Ron Landi
Wired Network
Secure
Server
To WAN
Wireless Access Point (AP)
10/100 Base T
Wireless
Access Point
(AP)
Secure
Edge
Device
10/100 Base T
Router/Firewall
10/100 Base T
Wireless Access Point (AP)
Wireless
Access Point
(AP)
To wired
network
10/100 Base T
Wireless Access Point (AP)
Wireless
Access Point
(AP)
Hot spots
Hot spots enable handheld and laptop users to connect to the Internet from public places such as
airports, cafes, restaurants, hotels, and other public places (i.e., libraries). They are 802.11
wireless local networks connected to the Internet via broadband access. Hot spots are used by
wireless Internet service providers (WISPs) such as T-Mobile and Wayport to provide users with
Internet and e-mail access.
There has been much activity regarding wireless hot spots in recent years. There are a number of large
entities that have built wireless hot spots and made related infrastructure investments in airports, hotels,
cafes, restaurants, and business centers. Some of these entitieswhich are partnering with
communications companies and systems integratorsare wholesaling, reselling, or retailing wireless
LAN connectivity and Internet access services. One example of a high-profile but unsuccessful venture
was Cometaan AT&T and IBM joint venture, with plans to build out more than 20,000 hot spots
throughout the country. Cometa recently announced plans to terminate its efforts.
The cost of backhaul and Internet service appear drive the majority of hot-spot deployments. In addition,
there is minimal cross system operation for roaming between hot spots provided by different vendors.
Refer to Figure 3 for a generic configuration of a typical wireless LAN hot spot.
289
Wireless
Access Point
(AP)
Secure
Server
Router
10/100 Base T
10/100 Base T
To WAN
Edge Router
Hot Spot
Local and city government entities are showing much interest in campus and last-mile access to provide
public wireless access to their citizens and enhance the capabilities of public safety departments. Campuslike or hot-zone environments (i.e., hospitals, corporate parks, malls, revitalized downtown areas) and
suburban/rural residential areas lacking broadband access are examples of where last-mile wireless LAN
solutions might be used. In addition, service providers are looking at wireless connectivity to and within
resort and other high-traffic areas, in addition to specialized applications (i.e., sea ports). Radio-frequency
interference, in addition to building penetration and coverage, remains a key challenge for wider-area
coverage using wireless LANs. The ability of a service provider or other entity to control geographic
areas should help mitigate some of these deployment issues. Refer to Figures 4 and 5 for examples of
campus and last-mile wireless LANs. Figure 4 shows an example of a wireless LAN using partial mesh
architecture. Figure 5 shows another example using smart antenna technology.
290
Ron Landi
Open-Area Service
- campus-like environments
- suburban residential coverage
- minimum infrastructure for coverage
Peered APs
nxMb Ethernet
nxMb Ethernet
Wired APs
Router
Other Considerations
- interference
- building penetration
- more APs / higher costs
Peered APs
Wired APs
Router
Peered APs
Peered APs
Peered APs
Edge Router
Internet Dedicated
T1 (nxMb)
TDM NxT1
TDM NxT1
To WAN
Open-Area Service
ile
1m
ius
rad
- campus-like environments
- suburban residential coverage
- minimum infrastructure for coverage
Other Considerations
Smart / Directional
Antenna AP
nxMb Ethernet
Router
- propagation
- interference
- building penetration
- more smart antennas / higher costs
Edge Router
TDM NxT1 (nxT1s)
To WAN
Internet Dedicated
T1 (nxMb)
critical mass of hot-spot locations in order to attract a sufficient number of wireless LAN customers. But
without a sufficient number of wireless LAN customers, it becomes difficult to justify the capital
investments in hot spots. In addition, there may be more than one service provider competing for
customers in the same hot spot. The campus and last-mile access models also require a sufficient number
of customers to justify the capital outlays. Figure 6 displays, in chart form, relative costs for a few
different Wi-Fi network configurations. It should be noted that these costs represent rough-order-ofestimate ranges and will vary by the specific deployment.
Configuration (1)
Total Capital
Internet +
Access
Total Monthly
Cost
Not included
$7K - $12K
$15K - $24K
Not
included
$400 - $600
$2K - $4K
$1.1K - $1.3K
$8K - $13K
$1.3K - $1.6K
$104K - $173K
$35K - $59K
$200 - $300
(1) = costs are for illustrative purposes only and will vary based upon specific configuration, associated requirements, vendor
equipment and scale/size
(2) = Internet access only
(3) = Internet + last mile access
(4) = includes utility pole lease + power
292
Ron Landi
802.11 Spaces
802.11 Opportunities
Infrastructure
Provider
Branded
Resale
Back Office
Systems &
Services
Network
Provider
Enterprise WLAN
Hot Spots
Campus
293
Pronto Networks
One of the major challenges in offering a carrier-grade wireless fidelity (Wi-Fi) service is to choose and
deploy an effective OSS infrastructure. This paper discusses the requirements for customers, service
providers, network operators, and venue owners and provides recommendations for service providers and
network operators evaluating different OSS options.
The OSS Challenges for WLAN Services
Wireless local-area network (WLAN) access is finally coming of age. The number of hotspots is growing
rapidly around the world (with more than 10,000 in the United States alone), as users are increasingly
eager to use the service not only in airports, coffee shops, and hotels, but also on trains and in
underground stations, in RV parks, and marinas. As WLAN access expands to new types of hotspots or to
wider hot zones, and as multiple networks and service providers share the wireless infrastructure, more
complex WLAN topologies have started to appear.
The WLAN Infrastructure Is Growing in Complexity
Airports are a good example of this ongoing transformation. Initially, WLAN access was limited to an
airline lounge or a couple of gates, and a single service provider offered the service through a monthly
subscription or a credit -card payment. The network was typically used only for public access and the
over-the-air link was left unprotected. This relatively simple WLAN infrastructure is becoming more
complex, as the number of services increases and as the requirements from users, service providers and
venue owners have become more sophisticated:
Coverage. Areas covered by the WLAN have expanded, often to include entire terminals or airports.
Roaming. The increasing number of roaming agreements results in several service providers
offering public access through the same network operator.
295
Operations Support System (OSS) Requirements and Solutions for Carrier-Grade Wireless LAN Services
Security. Robust security solutions, such as wireless protected access (WPA), are now available
to protect all users, but they require direct involvement of service providers and network
operators to enable mutual authentication and encryption keys management.
Consistent service. Service providers want to label the service, charge the user through a
single bill, and offer the same level of service and client interface that is available within
their own network.
Location-based services. The airport may request to provide location-based information and
services (e.g., information about flight departures, airport services) to the travelers.
Client interface. Users increasingly demand a robust, easy-to-use interface and a secure
connection, while retaining the ability to select their favorite service provider or use alternative
security solutions, such as their companys virtual private network (VPN).
Multiple virtual networks. Besides public access, WLANs may provide services to airport
concessions, airlines, and security agencies, thus creating the need for multiple virtual networks,
all supported by the same WLAN infrastructure.
The effects of the increased complexity at the hotspot level are magnified by the growth of the domestic
footprint and its international expansion. The establishment of roaming agreements among service
providers based in different countries has suddenly given international access to subscribers. It has also
increased the back-end burden for service providers and network operators alike.
Compelling, Carrier-Grade Services Are Critical
Several mobile carriers, fixed operators, wireless Internet service providers (WISPs), and, in some cases,
venue owners, are offering or planning to offer Wi-Fi access, and are becoming increasingly aware of the
challenges that WLAN services pose. Once they have established a footprint, by building their own
infrastructure (e.g., T-Mobile, BT Openzone, Wayport) and/or by entering roaming agreements (e.g., Sprint,
AT&T Wireless), service providers need to ensure that they are offering a compelling, carrier-grade service
to their customers. This is a crucial capability for service providers to attract subscribers, differentiate
themselves from competitors who may offer free access, and avoid commoditization of the service.
A Robust OSS Is a Prerequisite for a Carrier-Grade WLAN Service
The task of managing the WLAN infrastructure and services to the subscriber has proved to be more
complex than initially expected. As a result some subscribers find it too difficult or time consuming to get
connected at hotspots, and service providers have only limited back-end functionality. Adoption of an
effective WLAN OSS is a crucial step to ensure that WLAN can become a carrier-grade service that has
the same reliability and level of customer support as the other services offered by the service provider.
WLAN OSS Is Challenging and Increasingly Complex
The challenges that a WLAN OSS faces are several. WLAN access is a service that is not yet mature
and that is rapidly evolving. Relying on a relatively new technology that is used in multiple
environments (e.g., at home and at work, in addition to public hotspots), users often need additional
support to learn how to use the service in a new environment and avoid software conflicts. The
potential for value-added services (VAS) creates additional requirements for future compatibility that
are often difficult to assess. The lack of widely accepted standards for authentication, authorization,
and accounting (AAA) and security and the need to integrate with the OSS for existing services (e.g.,
voice services) adds complexity to the adopted OSS solution. To complicate matters further, the
presence in the market of different players (e.g., venue owners, fixed operators, mobile operators,
296
WISPs) translates into different integration needs, as the WLAN OSS typically coexists with
different legacy systems. The emerging complex roaming infrastructure increases the number of
players that need to exchange user data information and to process payments. At the wholesale level
there is still considerable uncertainty as to the metric (volume, time-metered, or daily access) to be
used, requiring network operators to collect different sets of data for different service providers.
Finally, the need to establish interworking with cellular networks, will create additional requirements
and a closer cooperation between service providers and network operators.
This white paper examines the OSS requirements that service providers, network operators,
aggregators, and clearinghouses have to meet to address these challenges and provides
recommendations on how to evaluate and select an OSS. The requirements are assessed from the
viewpoint of the value chain actors: subscribers, corporate clients, venue owners, network operators,
and service providers. Recommendations are based on the outlined requirements and on the demand
that we expect to arise from future services.
OSS for WLAN: A Definition
To clarify the scope of the white paper, it is useful to define the five functional areas that WLAN OSS
encompasses (see Figure 1). They include billing and customer management that are sometimes defined
as business supporting services (BSS).
Billing
Customer
Management
Network
Management
Service
Provisioning
Service
Assurance
Event management
CRM
Network planning
Order management
Rating, discounting
Customer service
Resource
management
Workflow
management
Roaming wholesale
billing
Call center
management
Configuration
management
Service activation
Trouble ticketing
Network provisioning
SLA management
Trouble reports
Network inventory
QoS implementation
Policy management
Testing
Invoicing
Taxation
Remote management
Security and
authentication
management
Performance
management
Fault management
Collection
Fraud management
Micropayments
Operations Support System (OSS) Requirements and Solutions for Carrier-Grade Wireless LAN Services
Aggregators that bring together network operators and service providers through roaming
arrangements geared to increase traffic and optimize the use of the wireless infrastructure.
Clearinghouses that facilitate the transmission of usage and billing information and offer financial
settlement services.
Hotspot
Venue owner
Aggregator
Network operator
Clearinghouse
Router
Internet
Access
controller
Service provider
Radius AAA
server
Switch
Access
points
Billing, customer
management, provisioning
Mobile
client
Subscriber
298
Quick negotiation of connections. The user needs to be able to select the preferred service set
identified (SSID) and service provider and get authenticated very quickly (less than a minute).
Easy to use interface for AAA. Regardless of the local network operator, the subscriber needs to
find a familiar, consistent, easy-to-use interface provided by the service provider.
Robust security. The over-the-air link has to be secured while allowing the subscriber to use
alternative security solutions, such as VPN (see Figure 3). It is worth noting that robust security
solutions such as Wi-Fi protected access (WPA) and, in the future, Microsoft wireless
provisioning services (WPS) require support from the network operator and service provider and
thus, unlike VPN, cannot be adopted unilaterally by the subscriber.
Single bill. When roaming on a partners networks, all the fees incurred have to be charged to
subscribers main account.
Information about hotspot locations and prices. When multiple options are available (e.g.,
multiple network operators, or service available through roaming agreements from different
service providers), the user needs to have access to information about pricing, service features,
and coverage area and be able to select the preferred connection option.
Pricing flexibility. Different pricing options are necessary to attract users with different
requirements. Heavy travelers may choose a flat fee price, while the occasional user may prefer a
per-session or per-minute charge or a prepaid account. Subscribers with high usage requirements
may also opt for premium services enabled by QoS, which will give them a higher priority in
using the bandwidth available.
Management of own account on-line. Subscribers find it valuable to access information about
their account and their usage of the service on-line.
Effective customer support. Customer support needs to be available initially to educate the user
and later to address more complex problems.
WPA/802.1x
802.11i
VPN
WPA is a Wi-Fi Alliance de-facto standard that overcomes the security vulnerabilities of WEP. It
provides support for mutual authentication through 802.1x, dynamic encryption keys through
TKIP, and for AAA functionality through RADIUS.
IEEE 802.11i is a proposed standard that will be backward compatible with WPA and will
provide support for a more powerful encryption scheme (Advanced Encryption Standard, AES).
Ratification of the standard is expected by the end of 2004.
VPN is a tunneling protocol that protects data in transit over a wireless or wired link through
encryption. It is currently widely used to secure WLAN connections, after the initial
authentication.
EAP
PEAP
Extensible authentication protocol that can be used in conjunction with EAP to provide further
protection.
WPS
Microsoft provisioning platform that includes WPA, 802.1x and PEAP, that provides increased
security and an improved interface. WPS is expected to be available in 2004.
299
Operations Support System (OSS) Requirements and Solutions for Carrier-Grade Wireless LAN Services
Security. While corporations may in the longer term want to make specific corporate services
available to their employees while working remotely, at present their major preoccupations are to
ensure secure connections and adherence to company policies (e.g., with regards to Internet access).
The ability of a service provider to offer secure authentication and access and to allow the use of
corporate VPN tunneling is necessary to gain acceptance, as a security breach may compromise not
only the employee using WLAN services, but it could jeopardize the entire company.
Wide footprint. Corporate customers will need to have access to a wide network of hotspots,
including most airports, convention centers, and hotels. At all locations, a consistent service is
expected. As not all service providers will be able to have their own infrastructure in those
locations, it is necessary they develop roaming partnerships with the network operators that
manage these hotspots to offer an attractive service.
Customer support. Effective customer support has to be available to address the corporate users
needs. This is key to reduce the impact that the service will have on corporate internal IT support
and to ensure customer satisfaction.
Advanced account management ability. Corporate accounts will require detailed usage and billing
information both at the individual user and at the company level for internal monitoring purposes.
Venue Owner Requirements: Reach the Subscriber and Minimize Infrastructure Maintenance
Venue owners welcome WLAN hotspots on their property for three primary reasons:
Revenue opportunity
A service to their customers
A means to reach visitors or customers and offer information and services.
Venue owners are increasingly less involved in the direct management of hotspots, relying instead on
network operators to deploy and manage the WLAN. As a result, they want to minimize the intrusion of
the wireless infrastructure on their operations. To meet their expectations, WLAN services need to meet
these requirements:
300
Hassle-free infrastructure. High reliability and remote network management are necessary to
ensure that the service runs smoothly in the background, without too many inquires from
subscribers or visits from the network operator.
Location-based services and branding. The venue owner may want to brand the service, along
with the service provider, in the splash page, or offer relevant information and specific services
(e.g., check out in a hotel, flight information at an airport) to its customers. If micropayments are
required for those services, the OSS must support a framework for charging the subscriber
through the service provider bill. As some location-based services may be free to all customers
and may not require authentication, virtual separate networks may be required.
Integration with internal billing service. The WLAN infrastructure may be used for locationbased services that are tied directly to the billing system of the venue. The OSS role is to facilitate
the integration of the local Wi-Fi infrastructure with venue owner back-end systems, such as
property management systems in hotels that enable services like on-line checkout.
Pricing flexibility. If the venue owner gets a share of the revenues, it is keen on ensuring high
traffic levels. Pricing flexibility is conducive to higher traffic levels as it more efficiently captures
the potential demand among hotspot visitors. In particular, some venue owners may be interested
in retaining the ability to market the service to visitors or to provide free access to customers (for
instance through coupons that allow customers to have free WLAN access for a limited time).
Multiple virtual networks. In some locations, the WLAN infrastructure may be used by the venue
owner and its tenants for their own internal services, in addition to public access and locationbased services. In this case, the network operator will be required to manage multiple virtual
networks and to ensure that the traffic is kept separate for security reasons and to ensure that
sufficient bandwidth is available to all users.
Network Operator Requirements: A Robust and Effective OSS to Manage Their WLAN Infrastructure
Network operators occupy a central role in ensuring that WLAN is a carrier-grade service. They need to
work with venue owners to ensure that they have their support and access to the facilities, and with
aggregators, clearinghouses, and service providers to ensure that the service runs smoothly. The
responsibility for the network installation, maintenance, and operations mostly lies with the network
operator. Service providers are instead responsible for billing, customer care and some provisioning
functions. Requirements include the following:
Scalable and extensible system. The OSS has to handle effectively any extension to a single
hotspots coverage area and any increases in the number of hotspots in the footprint.
Easy installation. Installation of the WLAN infrastructure has to be streamlined and require a
minimum of staff resources at the hotspot location.
Hardware independent network management. An OSS that is not tied to specific hardware
implementations and that provides multivendor hardware support will give network operators the
necessary flexibility in selecting the hardware that is best suited to their needs.
Remote network management. Expensive truck rolls may quickly become one of the major cost
items in deploying and managing a WLAN. Remote network management allows the operator to
keep operating costs associated with configuration, provisioning, and maintenance low.
Detailed traffic and usage monitoring. Collection of detailed records allows service providers to
offer multiple pricing options and network operators to implement revenue-sharing agreements
effectively. Once WLAN and cellular technologies become integrated, real-time billing may be
needed by cellular carriers to prevent subscribers from exceeding their time or volume allocation.
Efficient use of the WLAN infrastructure. In some instances, the WLAN infrastructure may be
used by multiple networks (in the airport example at the beginning, for public access, security
301
Operations Support System (OSS) Requirements and Solutions for Carrier-Grade Wireless LAN Services
agency communications, airlines internal operations). This increases the utilization of the network
deployed and may be conducive to higher revenues.
Robust AAA capabilities. As the network operator deals with multiple aggregators,
clearinghouses, and service providers, it is important that it adopts a best practice AAA solution
compatible with remote access dial-in user service (RADIUS)1. In particular, secure
authentication is necessary to gain the trust of service providers in implementing roaming
agreements. Different service providers may prefer specific authentication methods (for instance,
global system for mobile communications [GSM] and other cellular carriers may want to use the
extensible authentication protocol with the subscriber identity module [EAPSIM]
authentication) and expect that their roaming partners support them.
Offer security to protect network and users. Implementation of a robust security framework is
necessary to protect the users and the network, and to gain trust of the other players. Network
operators need to support security standards and solutions (see Figure 3) that their roaming
partners and the subscribers use.
QoS management. Advanced services such as voice over WLAN (VoWLAN) and high traffic
levels require the introduction of QoS and load balancing to ensure an appropriate and efficient
distribution of resources.
Service Provider Requirements: Integrate the WLAN OSS into Their Legacy Systems
Wi-Fi access is often offered as complementary to existing services: cellular carriers offer Wi-Fi access
alongside voice and cellular data, broadband providers as an add-on to fixed broadband access, data
service providers as a complement to dialup or fixed broadband access to the road warriors. As such, it is
critical that the WLAN OSS can be easily and fully integrated with the existing OSS, both to improve the
customer experience and to keep the costs down, by avoiding unnecessary duplication.
In addition, subscribers will often use Wi-Fi services in hotspots managed by roaming partners. The service
provider will want to brand services at these locations and will need to have access to information about the
network to provide effective customer support. This entails the exchange of information between network
operator and service provider, possibly through an aggregator to coordinate these efforts.
Finally, Wi-Fi needs to be a carrier-grade service to complement or compete successfully with other
cellular technologies and to allow service providers to generate revenues from it. To ensure that this is the
case, an OSS needs to meet these service provider requirements:
Full roaming capability, with infrastructure supporting flexible wholesale contracts. Roaming
gives service providers access to a wider footprint, which in turn makes the service more
attractive to subscribers. Roaming, however, may be an expensive service to offer (especially if
roaming premiums are low or nonexistent) as it requires the establishment and maintenance of
several roaming partnerships. While aggregators and clearinghouses may help keep costs under
control, a billing system that includes full support for roaming is necessary. The billing system
needs to provide functionality to manage wholesale relationships with multiple roaming partners
and clearinghouses (both for selling and buying access at a wholesale basis i.e., providing
access to subscribers from other networks and paying for the charges incurred by its own
subscribers accessing the roaming partners hotspots) and to bill subscribers.
In addition to RADIUS servers, solutions such as Microsofts Active Directory and Suns Identity Server offer
RADIUS compatibility.
302
Integration with existing billing and customer-management systems. As WLAN access will in
most cases be a new, add-on service, it is typically more cost effective for service providers to
integrate WLAN billing and customer-care systems with the existing ones, than to switch to new
ones. As a result, it is crucial that the OSS allows the service providers to easily interface with
established OSS for cellular and fixed operators and broadband providers.
Flexibility in setting multiple or tiered pricing options. It is too early to know which pricing
model will prevail, and it is likely that different service providers will prefer different models. In
all cases, they will want to keep their options open and be able to experiment and integrate Wi-Fi
more or less tightly with other services offered. For instance, a cellular operator may want to
allow subscribers to use their monthly minutes for Wi-Fi connections as well as voice. When QoS
becomes available, tiered services may be linked to different pricing options.
Secure authentication and data transmission. The OSS needs to offer secure connectivity to
subscribers who require it and to allow other users to avail themselves of their preferred security
solutions (e.g., VPN). In cases where the subscriber connects at a roaming partners location, it is
important that the user is authenticated directly against the service provider authentication servers
and that the partner network operator does not have access to the subscriber authentication
credentials. Providing secure connections will attract users (especially those who do not have VPN
at their disposal) and constitute a differentiating factor from hotspots that offer free connectivity.
Branding of service. Branding of the visited hotspot enables service providers to offer a
consistent service, with a single interface and to add VAS that may increase subscriber retention
and average revenue per user (ARPU). As venue owners and franchises will also require branding
of the service provided at their hotspot(s), branding information from different sources needs to
be integrated in the splash page.
Access to information on roaming partners networks. While the service provider will not need
extensive visibility into the partners network, it will need to have sufficient information to
provide customer support.
Ability to manage increasing complexity. The OSS needs to offer the flexibility, extensibility and
scalability required to manage an increasingly complex footprint and the subscriber base demands
for advanced functionality. The OSS needs to be developed with an eye towards advanced
services and must facilitate the introduction of new services.
Support for standards and widely adopted solutions (e.g., WPA and VPN for security, RADIUS
for AAA, or billing formats such as internet protocol detail record [IPDR], transferred account
procedure [TAP], cellular intercarrier billing and exchange roaming record [CIBER] and, in the
future, mobile exchange protocol [MXP]) (see Figure 4). A standard-based OSS is crucial to
303
Operations Support System (OSS) Requirements and Solutions for Carrier-Grade Wireless LAN Services
ensure interoperability with roaming partners, to facilitate integration with existing billing and
customer management systems, and to meet the security and usability requirements of subscribers
and corporate accounts managers.
Offer an easy-to-use, effective, and consistent interface to the user. Users need to rely on an
interface that is straightforward to use, without assuming any knowledge of the operations of
the underlying technology. Furthermore, it must enable them to monitor and manage their
accounts on-line.
Provide a cost-effective solution. The OSS should allow service providers to avoid duplication of
OSS functionality (e.g., in billing and customer care) and to streamline the deployment and
management of the OSS infrastructure.
IPDR
IPDR NDM-U (Network Data Management - Usage) protocol has been specifically developed for
IP services, both for wired and wireless services.
TAP
CIBER
MXP
RADIUS
RADIUS is the AAA protocol that provides the data to be converted into IPDR, TAP, CIBER and
MXP.
widen, thus increasing the importance of an effective communication channel among venue owners
(pushing for most local services), network operators and service providers. If services are billed to the
subscriber account, the OSS has to offer the functionality to process micropayments.
VoWLAN
VoIP is making quick progress in the enterprise and in the fixed telecommunications markets in North
America, and VoWLAN is one of the hot applications for WLANs in the enterprise. The public hotspot
market is not yet ready for VoWLAN, both because the devices are not yet for sale (although several
manufacturers are working on handsets using 802.11b) and because QoS is necessary, but no standardsbased solution is yet available. IEEE 802.11e, when ratified, will provide support for QoS and will
facilitate adoption of VoWLAN. It is still uncertain whether cellular carriers will want to use the WLAN
infrastructure to route voice calls when subscribers are in hotspots: while WLAN may provide a cheaper
over-the-air link, and it may help relieve congestion in the cellular network, coexistence of cellular voice
service with VoWLAN makes service provisioning more complex. It is also possible that other service
providers, such as WISPs, may elect to offer VoWLAN for free, with the user simply paying for access
charges, as voice effectively becomes an additional data service that is inexpensive to provide.
On-Line Gaming
While the predominant applications for WLAN public access over the next few years will still be e-mail,
Web surfing and Internet connectivity, on-line gaming is one of the most interesting emerging
applications, as it is targeted at the consumer users, a segment that will grow quickly once WLAN
becomes more widely embedded in laptops and personal digital assistants (PDAs). Games have proven to
be a successful application for cellular devices; WLAN higher bandwidth and the possibility to establish
peer-to-peer connections make on-line gaming is a likely winner. It is not yet clear, however, how a
WLAN service provider can generate revenues from it, other than from the increase in traffic that gaming
may generate. Gaming will put additional requirements on the WLAN infrastructure, as it will require a
close monitoring of performance and possibly the introduction of QoS. Support for micropayments will
also be necessary if revenues are to be extracted from downloading or playing games.
Content Downloads and Streaming
Today subscribers can already download content and stream video or audio, either for free or by paying
the content providers directly. Because traffic on WLANs is still on average below capacity, downloads
do not cause problems. In an environment where usage is high, it may be necessary to use QoS, load
balancing, and more sophisticated network management tools. In addition, specific types of content (e.g.,
music, films) may require a micropayment from the user and specific tools for digital rights management.
Interworking with Other Wireless Technologies
While not yet in demand, interworking with other wireless technologies, including cellular networks,
IEEE 802.16 (WiMAX) and IEEE 802.20, will become a requirement within a few years. While
todays users typically access a WLAN from a laptop, an increasing share of traffic will come from
PDAs, phones and other mobile devices. The increased mobility that these devices promote will make
interworking with other wireless technologies more useful, as users will often walk or drive beyond the
WLAN area of coverage.
Final Recommendations
As WLAN services become more widely available and popular, the requirements for service providers,
network operators, and intermediaries, such as clearinghouses and aggregators, are becoming better
defined and more stringent. Managing the WLAN infrastructure is becoming more demanding, because of
the growing number of hotspots and their more complex architecture (wider coverage, multiple networks,
support for roaming and security). The increasing availability of roaming and the need to provide a
305
Operations Support System (OSS) Requirements and Solutions for Carrier-Grade Wireless LAN Services
carrier-class service that can be integrated with existing ones poses new challenges for service providers
and intermediaries alike.
The availability of standards like 802.1x and WPA for authentication and security, IPDR, TAP, and
CIBER for billing information, and of widely used solutions like RADIUS for AAA, or VPN for security
makes it possible to have a robust infrastructure that enables service providers to offer WLAN as a
carrier-grade service. To achieve this goal, however, network operators and service providers need to
evaluate the OSS solutions available carefully and ensure that they meet the requirements for WLAN
access. This paper identifies requirements that need to be satisfied to ensure that the subscribers receive a
compelling service and that all the value chain players can work together toward this goal.
The initial challenge for WLAN service providers was to be able to offer connectivity across a
sufficiently wide footprint. The rapid growth in the number of hotspots and the progress toward
establishing a wider network of roaming agreements show that service providers and network operators
are moving in the right direction. The next challenge is to ensure that the infrastructure deployed can be
used to offer a service that users will find attractive and easy to use, while being cost effective for
network operators and service providers. The adoption of robust, yet flexible OSS will be crucial to meet
this challenge.
Copyright 2003 Senza Fili Consulting LLC. All rights reserved.
This white paper was commissioned by Pronto Networks, Inc. Pronto Networks retains the right to
redistribute the integral version of this report. No selection of this material can be copied, photocopied,
duplicated in any form or by any means, or redistributed without express written permission from Senza
Fili Consulting LLC. While the white paper is based upon information that we consider accurate and
reliable, Senza Fili Consulting LLC makes no warranty, express or implied, as to the accuracy of the
information in this document. Senza Fili Consulting assumes no liability for any damage or loss arising
from reliance on this information. Names of companies and products here mentioned may be the
trademarks of their respective owners.
306
307
Figure 1
Point-to-Multipoint Wireless Network
A high-capacity wireless network was required to extend the fiber-optic networks coverage area.
Motorola Canopy was selected because of its superior capabilities, ease of implementation and affordable
cost. Seven tower sites were constructed to host the Motorola Canopy access points thus extending the
community networks broadband footprint throughout the entire city. All sites are synchronized using a
unique approach. Broadcast and receive cycles are synchronized using global positioning system (GPS).
Tower sites are built or located on existing city properties or on partner sites. As an example, a 100foot water tower was used to install an omnidirectional set of Canopy access points (see Figure 2).
Canopy backhaul at the water tower is consolidated by the use of a Cisco 2940 series switch and
rebroadcasts Ethernet over the fiber-optic network to the head-end network switch. Seeking out a
qualified Motorola dealer versed in both radio and networking was identified as a critical success factor
in the deployment and maintenance of the wireless network. Eastern Wireless, a local Motorola dealer,
was selected to supply, install, and maintain the network. In return, Eastern Wireless is invited to use
the over capacity to offer Internet service provider (ISP) services to the community. BrunNet, another
local ISP, also resells the citys network to small business and to homes. In this way, e-Novations acts
as a wholesale ISP to the community.
308
Mike Richard
Figure 2
Fred-e-Zone Wireless Fidelity (Wi-Fi) Network
Fred-eZone is comprised of a unique combination of leading-edge technologies. The recent 802.11g
wireless technology standard has allowed for low-cost, reliable equipment that is now interoperable.
Deploying isolated access points to create a hotspot has become relatively easy and is now quite common.
Deploying a larger metropolitan Wi-Fi zone, however, has unique and difficult challenges, which include
the following:
With a state-of-the-art, city-wide network as a foundation, a metropolitan Wi-Fi project became possible.
In Fredericton, Wi-Fi installations are comprised of the integration of all levels of its existing network
technology. A typical configuration consists of one or more Cisco 1200 g radios (Wi-Fi access points)
consolidated on a single Cisco 2940 (8 port) switch. In general, 110degree panel antennas are deployed
outdoors for Wi-Fi coverage. Ceiling omniantennas are used for indoor coverage. Antennas are connected
to Cisco radios using LMR 400 cabling. The consolidated signal is then backhauled to the core network
via Motorola Canopy. The Canopy equipment is also consolidated on the same 2940 switch.
Ciscos Wireless LAN Solution Engine is used to manage over 120 Cisco 1200 Wi-Fi radios. This allows
for remote management, automatic firmware updates, power and load balancing, and rogue access-point
309
detection. Trouble areas are logged using SNMP and network administrators are automatically advised of
faulty equipment and many other potential issues the system has been configured to detect.
Fred-eZone is made up of over 120 access points deployed in over 40 geographical locations. All Wi-Fi
2940 switches are identically configured and tag all Fred-eZone traffic on a dedicated virtual local area
network (VLAN). When Fred-eZone traffic is consolidated at the core, the VLAN is redirected to a
Solutions Inc. proxy server.
When a user launches a browser, all DNS requests are redirected to the proxy server forcing the user to
the Fred-eZone portal. At this point users learn more about Fred-eZone and are asked to agree to an
acceptable use policy (AUP) and enter account information. Future use of Fred-eZone is then managed by
the Solutions Inc. server by login account and password. User information and MAC addresses are
managed and stored on a central database. In this way the Solutions Inc. proxy server also acts as a
remote authentication dial-in user service (RADIUS) server.
The City of Fredericton as Its Own ISP
Fred-eZone traffic is passed to the community network Cisco 3750 router. The community network
destined traffic remains on the core network at maximum speeds and minimum network hops. This
creates a truly low latency, high-speed network. For example, users with an 802.11a card can theoretically
realize full 54 Mbps within the network. Internet-bound traffic is directed to a dedicated port on the
community network Internet gateway switch. Again this is accomplished using VLAN tagging.
Bandwidth is shaped on this port to a 4 Mbps maximum of full symmetrical bandwidth. This approach
minimizes the impact of possible denial of service attacks or inequitable use of the full community
network Internet pipe by Wi-Fi users.
Users on the network can receive and send mail through the Community Network mail relay. Users can
also allow the Solutions, Inc. proxy server to automatically relay their sent items without requiring a
change of simple mail transfer protocol (SMTP) settings.
A large number of routable IP addresses are required to facilitate a project of this size. The community
network has peering agreements with two ISPs using border gateway protocol (BGP). The large IP
requirement along with the peering agreements and number of users qualified the community network to
apply for and receive its own IP allocation from American Registry for Internet Numbers (ARIN).
Conclusion
A unique combination of city infrastructure and leading-edge technology has been combined with
superior network engineering to accomplish a city-wide Wi-Fi zone. This has been realized through the
partnership of city staff, local expert resources, and the vendor community. It has taken four years, from
start to finish, to form a nondominant carrier company, construct the fiber-optic network, deploy the
point-to-multipoint wireless network, and, finally, deploy more than 120 Cisco Wi-Fi access points in the
field. All this has been done using existing budgeted dollars by reinvesting the citys telecom savings
back into the network and by accepting commercial subscribers on the citys network. The private sector
has also contributed cash and in-kind services representing a total value in excess of $250.000. The
resulting infrastructure has made Fredericton stand out as a truly smart city and has built an intellectual
infrastructure that is seen by many as leading edge and visionary.
310
311
The first step for this environment is to allow nomadic roaming between home and foreign hotspots.
Some of the main challenges associated with roaming across multiple service providers include
transparency of the service, network security, service authentication, and management of billing and
accounting records. Roaming centers are introduced to facilitate roaming services while resolving
previously mentioned challenges. A user may experience a captive portal page that includes optional
services and he/she has to select roaming service and provide his/her credential. Roaming centers have the
responsibility of redirecting the call into the user home network for authentication and authorization.
The next step for most service providers is to integrate hotspots with 3G networks and allow seamless
roaming between the two networks. There are two main architectures for deploying hotspots within 3G
networks, namely the following: loosely coupled architecture and tightly coupled architecture. In the
tightly coupled architecture, the hotspot access gateway (HAG) would have to behave to the 3Gcore
network as if it is another base station within the radio access network. In loosely coupled architecture, on
the other hand, the hotspot is indirectly connected to the core network mainly for access of authentication
and authorization. In this case, moving from/to a hotspot to/from 3G may lead to a change in the Internet
protocol (IP) address, connection protocol, and authentication requirement. Consequently, basic IP cannot
provide support for roaming between these two networks. Despite that, loosely coupled architecture is
almost always preferred due to the simplicity of integrating both networks, the minimal modification
needed on the 3G network, and the simplicity of the future addition of new hotspots. Tightly coupled
network is very rigid, less flexible, and more difficult to implement. Figure 1 provides high-level diagram
of an illustrative case of loosely coupled hotspot/3G CDMA2000 networks.
312
Aladdin Saleh
The authentication, authorization, and accounting (AAA) server referenced in Figure 1 is used to provide
a single sign-on authentication for customers subscribed for both 3G and hotspots services where same
credential is used in both cases. The AAA server is also used to provide RADIUS capability for
authenticating 802.1xbased hotspot users. Additionally, it generates accounting records that enable
postpaid and prepaid billing support.
3. Hotspot/ 3G Roaming
As indicated before, basic IP may not be suitable for supporting roaming services between hotspots and
3G networks. Mobile IP (MIP), which is originally defined by the Internet Engineering Task Force
(IETF), has been regarded as the most satisfactory protocol for serving this heterogeneous environment.
Figure 2 provides high-level diagram of such multiaccess networks including additional nodes needed to
support MIP with particular reference to a 3G CDMA 2000 network.
313
process between hotspots and 3G networks using MIP does not undermine the efficiency of the whole
network. Figure 3 shows latency for sample case during soft handoff from 3G CDMA2000 network
into WLAN network.
Aladdin Saleh
IEEE Std. 802.11b-1999, Local and Metropolitan Area Networks Specific Requirements-Part
11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications:
Higher-Speed Physical Layer Extension in the 2.4 GHz Band, 1999.
[2]
IEEE Std. 802.11a-1999, Local and Metropolitan Area Networks Specific Requirements
Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications:
High-Speed Physical Layer in the 5 GHZ Band, September 1999.
315
[3]
IEEE Std. 802.11g-2003, Local and Metropolitan Area Networks Specific Requirements
Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications
Amendment 4: Further Higher Data Rate Extension in the 2.4 GHz Band, 2003.
[4]
[5]
[6]
A. Saleh, Broadband Wireless Access Strategy, Internal Report, Bell Canada, 2004.
[7]
A. Saleh, IEEE 802.11 WLAN Integration in Multiaccess Environment, Submitted for Pblication.
[8]
[9]
C. Perkins, Ed.,IP Mobility Support for IPv4, IETF RFC 3344, August 2000.
316
especially at the lower layers such as MAC and PHY layers. This will ensure that existing 3G networks
will continue to function. Section 6 introduces session initiation protocol (SIP) mobility. SIP mobility can
be used for real-time communication over UDP.
2. 3G Standardization and Deployment
Second-generation (2G) wireless systems include global system for mobile communications (GSM),
IS136, and IS-95 code division multiple access (CDMA) [6, 7, 8]. GSM networks have the highest
penetration worldwide. The International Telecommunications UnionRadio (ITUR) developed the
International Mobile Telephony2000 (IMT2000) specifications [9]. IMT2000 is a set of standards for
creating a global 3G network that includes terrestrial systems, satellite systems, along with fixed access
and mobile access networks. The following groups are involved in the international standardization effort
for IMT2000: European Telecommunications Standards Institute (ETSI) Special Mobile Group (SMG),
Research Institute of Telecommunications Transmission (RITT) in China, Association of Radio Industries
and Businesses (ARIB) and Telecommunication Technology Committee (TTC) in Japan,
Telecommunications Technology Association (TTA) in Korea, Telecommunications Industry Association
(TIA) and T1P1 in the United States. ETSI SMG identified usage of wideband CDMA (WCDMA) for
3G networks [10]. China decided to deploy Synchronous Time Division CDMA (TDSCDMA) for 3G
networks [11]. ARIB, Japan decided to use WCDMA for 3G networks. TTA, Korea presented two
schemes, one similar to WCDMA and the other close to the TIA CDMA2000 approach. In the United
States, TIA presented several proposals for 3G UMTS, UWC-136 [12] as an evolution of IS-136,
CDMA2000 [13] as an evolution of IS-95 and WCDMA for GSM networks.
GSM and IS-136, being time division multiple access (TDMA) systems, will evolve to CDMA systems in
a number of steps. GSM networks must incorporate GPRS for evolving to 3G UMTS network capabilities
[14]. The next step for GSM networks will be to add enhanced data rates for GSM Evolution (EDGE)
capabilities [15]. Enhanced data rates for GSM evolution (EDGE) allows GSM operators to use existing
GSM radio bands while offering high bandwidth data services. EDGE will offer 384 kbps data rates for
pedestrian and low-speed environment. EDGE will offer 144 kbps for high speed vehicular environment
and two Mbps for indoor office environment. UMTS is a new radio access network based on WCDMA.
UMTS will offer 384 kbps in wide area and up to two Mbps in local areas. GSM operators have two
complementary options to upgrade their networks to 3G. The first option will be to use GPRS and EDGE
in the existing radio spectrum. The second option will be to deploy UMTS in new two gigahertz (GHz)
bands. CDMAOne (TIA IS-95) allows for channel aggregation to provide data rates in the range of 64
115 kbps. This simplifies migration of CDMAOne systems to CDMA2000 systems.
3. Evolution towards 3G UMTS HSCSD, GPRS, and EDGE
We describe key technologies for transitioning to 3G UMTS networks. HSCSD provides for high
bandwidth data rates by co-allocation of multiple full rate traffic channels (TCH/F) in GSM networks.
GPRS is a packet data service using transmission control protocol/internet protocol (TCP/IP) and X.25 to
offer speeds up to 115 kbps. EDGE is essentially a radio interface improvement scheme. EDGE is an
enhancement for both HSCSD and GPRS modes of operation. EDGE packet switched mode of operation
is known as Enhanced GPRS (EGPRS). EDGE circuit switched mode of operation is known as Enhanced
CSD (ECSD).
3.1 High-Speed Circuit Switched Data (HSCSD)
HSCSD provides for high bandwidth data rates by co-allocation of multiple TCH/F in GSM networks
[16]. Baseline data rates have increased from 9.6 kbps to 14.4 kbps due to reduction in error correction
overhead of GSM radio link protocol (RLP). Multiple 14.4 kbps time slots may be combined to offer
access rates up to 57.6 kbps. Multiple time slots are dynamically allocated based on network operator
318
Syed A. Ahson
policies and the data transfer needs of the user. HSCSD requires software upgrades to the base station
(BS) and the mobile switching center (MSC). The end-user experience will be similar to an internet
service provider (ISP) that offers fast secure dial up using their mobile equipment. HSCSD does not
require any changes to existing mobility management procedures. Simultaneous handoff should take
place for all time slots comprising the HSCSD connection. It should be noted that multiple time slots for
HSCSD connections will probably be available in off-peak network usage times.
3.2 General Packet Radio Service (GPRS)
GPRS is a packet data service using TCP/IP and X.25 to offer speeds up to 115 kbps [17]. GPRS provides
short connection setup times, virtual connections and users are charged for actual data transmitted.
Network resource and bandwidth are only used when data is actually transmitted. Bandwidth can be
shared efficiently and simultaneously among several users. A GPRS core network is defined in parallel to
the existing GSM core network. Two new types of nodes are introduced in GPRS, the serving GPRS
support node (SGSN) and the gateway GPRS support node (GGSN). The GGSN is responsible for the
connection to other packet switched networks and stores information about location of GPRS users. The
GGSN encapsulates TCP/UDP packets and forwards them to the SGSN using GPRS tunneling protocol
(GTP). The GGSN may also offer packet filtering services. The GGSN is connected with SGSN via an
IPbased GPRS backbone network. The home location register (HLR) is enhanced to store GPRS
subscription data such as IP address of mobile users and routing information. The HLR also maps each
subscriber to one or more GGSN. SGSN and GGSN nodes interface with the HLR through signaling
system seven (SS7) links. The SGSN is responsible for authorization, authentication, admission control,
charging, and mobility management of mobile users. The SGSN encapsulates TCP/UDP packets and
forwards them to the GGSN using GTP. The SGSN is connected to the Base Station system by Frame
Relay. Figure 1 illustrates a GPRS enhanced GSM network.
Voice Network
VLR
VLR
MSC
BTS
MSC
HLR
BSC
SGSN
GGSN
IP Backbone
Data Network
319
Mobile user packet data session is known as a packet data protocol (PDP) context. At power up,
mobile users perform a GPRS attach. At GPRS attach, the mobile users profile is downloaded from
the HLR to the SGSN. The mobile user must perform PDP context activation before it can send or
receive IP packets. The SGSN validates the PDP context activation request against the subscription
information downloaded from the HLR during GRPS attach. The GGSN that should be used for
TCP/UDP traffic routing is identified by a domain name service (DNS) query of the access point
name (the destination) [18]. A GPRS tunnel is created using GTP between the SGSN and the
destination GGSN.
3.3 Enhanced Data Rates for GSM Evolution (EDGE)
EDGE is designed to offer 3G services in existing spectrum bands. EDGE is essentially a radio interface
improvement scheme. The major enhancement in EDGE is the introduction of a new modulation system
known as eight phase shift keying (8 PSK). 8 PSK will coexist with the existing Gaussian minimum
phase shift keying (GMSK). 8 PSK will provide higher data rates in a reduced coverage area. GSM/GPRS
protocols are reused wherever possible. EDGE is an enhancement for both HSCSD and GPRS modes of
operation. EDGE packet switched mode of operation is known as enhanced GPRS (EGPRS). EDGE
circuit switched mode of operation is known as enhanced CSD (ECSD).
4. Universal Mobile Telecommunications Service (WCDMA)
A 3G UMTS network consists of three interacting domains: core network (CN), UMTS terrestrial radio
access network (UTRAN) and user equipment (UE). A UMTS system is divided into a set of domains and
reference points that interconnect them. The 3G UMTS protocol structure is based on the principle that
the layers and planes are logically independent of each other. Following is a description of the 3G UMTS
core network architecture, reference points, UTRAN, and protocol structure.
4.1 UMTS Core Network Architecture
The UMTS core network (UCN) is composed of a circuit switched (CS) domain and a packet switched
(PS) domain. The CS domain consists of a mobile switching center (MSC) and a gateway MSC (GMSC).
Figure 2 illustrates the entities present in a UMTS core network. The PS domain contains the SGSN, the
GGSN, domain name server (DNS), dynamic host configuration protocol (DHCP) server, packet charging
gateway and firewalls. The HLR interfaces with both domains over SS7 links. Other components required
for operation of the UCN include billing systems, provisioning systems, and service or element
management systems. The 3G MSC is responsible for mobility management. The 3G MSC handles IMSI
attach, authentication, HLR updates, Serving Radio Network Subsystem relocation and intersystem
handovers. The 3G MSC handles call setup, messages from or to mobile users, and provides
supplementary services such as call waiting. The 3G MSC also provides circuit switched data services for
services such as fax.
320
Syed A. Ahson
SSP
Voice Network
SCP
STP
CircuitSwitched
G-MSC
VLR
PacketSwitched
3G-MSC
DNS/DHCP
HLR
Firewall
SMSC
Packet
Charging
Gateway
GGSN
3G-SGSN
UMTS RAN
Data Network
Cu
User
USIM Domain
Uu
Iu
Equipment Domain
Core Network
Domain
Core Network
Iu
Iu
Iur
RNS
RNS
RNC
Iub
Node B
Iub
Node B
Node B
322
RNC
Node B
Syed A. Ahson
The RNC enables autonomous radio resource management (RRM) by the UTRAN. The RNC assists in
soft handover of the user equipment (UE) when a mobile user moves from one cell to another. The RNC
combines and splits Iub data streams received from multiple Node Bs. The RNC is also responsible for
frame synchronization, outer loop power control, and serving RNS (SRNS) relocation.
Node B is physical unit of radio transmission and reception with cells. It can support both time division
duplex (TDD) and frequency division duplex (FDD) modes and can be collocated with GSM base
transceiver system (BTS) to reduce implementation costs. It connects to the user equipment via the Uu
interface and the RNC via the Iub interface. The main task of Node B is the conversion to and from the Uu
radio interface, including forward error correction (FEC), rate adaptation, WCDMA spreading/despreading, and quadrature phase shift keying (QPSK) modulation on the air interface. It measures quality
and strength of the connection and determines the frame error rate (FER), transmitting these data to the
RNC as a measurement report. The Node B also participates in power control, as it enables the user
equipment to adjust its power using down link (DL) transmission power control (TPC) commands via the
inner-loop power control on the basis of up link (UL) TPC information. The predefined values for innerloop power control are derived from the RNC via outer-loop power control.
4.4 UTRAN Logical Interfaces
The general protocol model for UTRAN Interfaces is shown in Figure 5. The structure is based on the
principle that the layers and planes are logically independent of each other [1921]. The protocol structure
Control Plane
Radio
Network
Layer
User Plane
Application
Protocols
Data
Streams
Transport Network
Control Plane
Transport
Network
Layer
Signaling
Bearer(s)
ALCAP(s)
Data
Bearer(s)
Signaling
Bearer(s)
Physical Layer
323
consists of two main layers, radio network layer (RNL), and transport network layer (TNL). All UTRAN
related issues are visible only in the RNL, and the TNL represents standard transport technology that is
selected to be used for UTRAN, but without any UTRAN specific requirements. The control plane includes
radio access network application part (RANAP) at Iu, radio network subsystem application part (RNSAP) at Iur,
or Node B application part (NBAP) at Iub, and the signaling bearer for transporting the application protocol
messages. Among other things, the application protocol is used for setting up bearers (i.e. Radio Access Bearer
or Radio Link) in the RNL. The user plane includes the data stream(s) and the data bearer(s) for the data
stream(s). The data stream(s) are characterized by one or more frame protocols specified for that interface.
The transport network control plane does not include any RNL information, and is completely in the
transport layer. It includes the access link control application part (ALCAP) protocol(s) that are needed to
set up the transport bearers for the user plane. It also includes the appropriate signaling bearer(s) needed
for the ALCAP protocol(s). The transport network control plane is a plane that acts between the control
plane and the user plane. The introduction of transport network control plane is performed in a way that
the application protocol in the radio network control plane is kept completely independent of the
technology selected for data bearer in the user plane.
The UMTS Iucs logical interface interconnects the UTRAN to the UMTS circuit switched core network. The
circuit switched protocol architecture on the Iucs interface is illustrated in Figure 6. The radio network layer
control plane consists of RANAP. The transport network user plane consists of SS7 protocols. Signaling
connection control part (SCCP), message transfer part (MTP3B) and signaling asynchronous transfer mode
(ATM) adaptation layer for network-to-network interface (SAALNNI) is present in the transport network
user plane. SAALNNI is divided into service specific coordination function (SSCF), service specific
connection oriented protocol (SSCOP) and ATM adaptation layer 5 (AAL5). SSCF and SSCOP are designed
for signaling transport in ATM networks. AAL5 is responsible for segmenting data into ATM cells.
Control Plane
Radio
Network
Layer
Transport
Network
Layer
User Plane
Radio Access
Network
Application Part
Iu Data
Stream
SCCP
ALCAP
MTP3-B
MTP3B
SSCFNNI
SSCFNNI
SSCOP
SSCOP
AAL5
AAL5
ATM
Physical Layer
AAL2
Syed A. Ahson
The UMTS Iups logical interface interconnects the UTRAN to the UMTS packet switched core network.
The packet switched protocol architecture on the Iups interface is illustrated in Figure 7. SCCP, message
transfer part user adaptation layer (M3UA), simple control transmission protocol (SCTP) and IP is present
in the transport network user plane. AAL5 is responsible for segmenting data into ATM cells.
Control Plane
Radio
Network
Layer
Radio Access
Network
Application Part
Iu Data
Stream
Transport
Network
Control Plane
SCCP
M3UA
Transport
Network
Layer
User Plane
GTPU
SCTP
UDP
IP
IP
AAL5
AAL5
ATM
ATM
Physical Layer
Physical Layer
325
is that the IWU present as single point of failure and a potential bottleneck in the 3G UMTS IEEE
802.11b WLAN interconnected network.
Internetworking Relay
Mobile
Station
Access Point
PDCP
GTP-U
802 LLC
UDP
IP
802.11
MAC
802.3
MAC
802.3
MAC
AAL5
802.11
PHY
802.3
PHY
802.3
PHY
L1
3G SGSN
IEEE
802.11
WLAN
ATM
3G GGSN
GTPU
GTPU
GTPU
UDP
UDP
UDP
IP
IP
IP
AAL5
L2
L2
L1
L1
L2
UMTS
Core
Network
ATM
L1
L1
Data Network
326
Syed A. Ahson
802.11b terminals are treated as 3G UMTS users. 3G UMTS mobility management schemes will keep
track of the IEEE 802.11b users irrespective of the network they are connected to. The main advantage of
this interconnection scheme is that mobility management, roaming, billing and security are taken care of
by existing 3G UMTS procedures. Minimum changes are required to the existing networks. The main
drawback of this scheme is that the IWU presents a single point of failure and a potential bottleneck in the
3G UMTS IEEE 802.11b WLAN interconnected network.
Internetworking Relay
Mobile
Station
Access Point
PDCP
GTPU
802 LLC
UDP
IP
802.11
MAC
802.3
MAC
802.3
MAC
L2
802.11
PHY
802.3
PHY
802.3
PHY
L1
3G SGSN
IEEE
802.11
WLAN
3G GGSN
GTPU
GTPU
GTPU
UDP
UDP
UDP
IP
IP
IP
AAL5
L2
L2
L1
L1
L2
UMTS
Core
Network
ATM
L1
L1
Data Network
327
802.11b WLAN
MS
AP
Router
Router
Virtual
Access
Point
Local-Area Network
Internet
UMTS
Core Network
UTRAN
MS
GGSN
3GSGSN
RNC
Figure 10: 3G UMTS IEEE 802.11b Internetworking by Virtual Access Point Scheme
328
Syed A. Ahson
802.11b WLAN
MS
AP
Router
Router
Local-Area Network
Mobility
Gateway
Internet
UMTS
Core Network
UTRAN
MS
GGSN
3GSGSN
RNC
329
handover when it moves out of the IEEE 802.11b WLAN network coverage and detects the presence of a
3G UMTS network. The 3G UMTS network is treated as a foreign network by the mobile device. When a
mobile user roams into a foreign network, it registers with a foreign agent to obtain a care-of-address. The
mobile device is allocated a care-of-address by a foreign agent on the 3G UMTS network. Mobile users
inform their home agents of their care-of-address. The mobile device home agent on the home IEEE
802.11b network is informed of the care-of-address by a mobile IP registration procedure. Datagrams are
always routed to the mobile user by its home agent. The home agent receives all datagrams addressed to
the mobile device and encapsulates them using IPinIP. These encapsulated datagrams are tunneled to
the mobile devices foreign agent. The Foreign Agents removes the IPinIP header of the datagrams
and delivers them to the mobile device. While the mobile device is attached to the 3G UMTS network, it
constantly searches for IEEE 802.11b signal. The mobile device could initiate a handover when it detects
an IEEE 802.11b WLAN network while connected to the 3G UMTS network. The foreign agent in the 3G
UMTS network is deactivated. The home agent is informed by the mobile device that it no longer requires
IPinIP tunneling. This scheme suffers from a triangular routing overhead.
802.11b WLAN
MS
AP
Router/
Foreign
Agent
Router/
Home Agent
Local-Area Network
Internet
UMTS
Core Network
UTRAN
MS
GGSN/
Foreign
Agent
3GSGSN
RNC
330
Syed A. Ahson
802.11b WLAN
MS
AP
Router/
Visited
Registrar
Router/
Home Registrar
Local-Area Network
Internet
UMTS
Core Network
UTRAN
MS
GGSN/
Visited
Registrar
3GSGSN
RNC
7. Conclusion
Seamless wireless data and voice communication is fast becoming a reality. IEEE 802.11b WLAN
networks have been widely deployed in offices, homes and public hot spots such as coffee shops and
hotels. However, IEEE 802.11b is limited by small coverage area (100300 feet). 3G UMTS networks
aim to create a convergence of existing networks to a global network based on one international standard.
If the user is under the coverage of an IEEE 802.11b WLAN network, his or her communication device
can access high bandwidth data service using the IEEE 802.11b WLAN network. If IEEE 802.11b
WLAN service is not available, the user may handover to the 3G UMTS network. This chapter describes
3G UMTS networks and internetworking between IEEE 802.11b WLAN and 3G UMTS networks. It
presents 3G standardization efforts and a description of strategies for 2G networks and the key
technologies (HSCSD, GRPS and EDGE) for transitioning to 3G UMTS networks. It further describes the
3G UMTS core network, reference points, UTRAN and protocol structure. Five possible network layer
architectures, which make no major changes to existing networks and technologies, are presented for
internetworking and handover between IEEE 802.11b WLAN and 3G UMTS networks. Mobile IP
interconnection architecture provides an efficient method for internetworking of heterogeneous packet
oriented networks. SIP mobility support reduces triangular routing overhead associated with mobile IP.
SIP cannot support TCP connections. SIP mobility can be used for real-time communication over UDP
and mobile IP for TCP connections.
8. References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
332
IEEE 802.11, 1999 Edition (ISO/IEC 8802-11: 1999) IEEE Standards for Information Technology
Telecommunications and Information Exchange between Systems Local and Metropolitan Area
Network Specific Requirements Part 11: Wireless LAN Medium Access Control (MAC) and
Physical Layer (PHY) Specifications
IEEE 802.3 2002, IEEE Standard for Information technology Telecommunications and
information exchange between systems Local and metropolitan area networks Specific
requirements Part 3: Carrier Sense Multiple Access with Collision Detection (CSMA/CD) Access
Method and Physical Layer Specifications
Daniel Collins, Clint Smith, 3G Wireless Networks, McGraw-Hill Professional, September 18,
2001
Heikki Kaaranen, Siamk Naghian, Lauri Laitinen, Ari Ahtiainen, Valtteri Niemi, UMTS
Networks: Architecture, Mobility and Services, John Wiley & Sons; 1st edition (August 14, 2001)
Daniel Collins, Carrier Grade Voice over IP, McGraw-Hill Professional, September 17, 2002
Mouley, M. and Pautet, M.B., The GSM System for Mobile Communications, Palaiseau, France 1992
Garg, V. K. and Wilkes, J. E., Wireless and Personal Communications Systems, Prentice
Hall, NJ 1996
TIA/EIA IS-95, Mobile StationBase Station Compatibility Standard for Dual-Mode Wideband
Spread Spectrum Cellular System, PN-3422, 1994
ETSI SMG, Proposal for a Consensus Decision on UTRA, ETSI SMG Tdoc 032/98
Harri Holma (Editor), Atti Toskala (Editor), WCDMA for UMTS, 2nd Edition, John Wiley &
Sons, September 12, 2002
TDSCDMA, http://www.tdscdma-forum.org/
TIA TR45, Proposed RTT Submission (UWC 136), TR-45.3/98.03.03.19, March 1998
TIA/EIA IS-2000-1 Introduction to cdma2000 Spread Spectrum Systems, Nov. 1999
ETSI Technical Specification GSM 02.60 GPRS Service Description Stage 1 version 5.2.1, July 1998
ETSI Tdoc SMG2 95/97, EDGE Feasibility Study, Work Item 184; Improved Data Rates through
Optimized Modulation, version 0.3 Dec. 1997
Syed A. Ahson
[16] Digital Cellular Telecommunication System (Phase 2+), High Speed Circuit Switched Data
(HSCSD), Service Description, Stage 2, GSM 03.34
[17] RFC 793, Transmission Control Protocol, September 1981
[18] RFC 1034, Domain Names Concepts and Facilities, November 1987
[19] 3GPP Technical Specification 25.410 UTRAN Iu Interface: General Aspects and Principles
[20] 3GPP Technical Specification 25.420 UTRAN Iur Interface: General Aspects and Principles
[21] 3GPP Technical Specification 25.430 UTRAN Iub Interface: General Aspects and Principles
[22] RFC 3220, IP Mobility Support for IPv4, January 2002
[23] RFC 2543, SIP: Session Initiation Protocol, March 1999
333
Maria Palamara
IMS Mobility Offer Manager
Lucent Technologies
Jim Starkey
IMS Mobility Marketing Manager
Lucent Technologies
This white paper addresses the following:
The need (as a result of primary market research conducted by Lucent among consumers) to
provide service providers with new solutions to meet their end-customer requirements for unique
and value added applications and services
The market-driven architectural approach of the Internet protocol multimedia subsystem (IMS)
335
It is a tall order to deliver a wide array of blended services cost effectively to target specific user segments
with the services that fit their lifestylesboth personal and business. Operators are now deploying hot
services based on point solutions to capitalize on early market opportunity quickly in spaces such as push
to talk, instant messaging, and IP private branch exchange (PBX)/Centrex. They, however, are quickly
discovering problems and opportunities associated with this approach of rolling out services.
For example, mobile operators are quickly realizing that it is difficult for subscribers to manage their
contacts across all of their various communications methods when they have to administer separate
contact lists for push-to-talk, instant messaging, voice calls, and voice messaging. If they could have a
single presence-enabled contact and group list management tool that could be leveraged by all these
applications, they believe that there would be much more utility to each of the services and
significantly less complexity in managing subscriber data. In addition, they see that presence-enabling
the contacts is very useful to subscribers so that they have more control in their communications, only
initiating a session if their buddies are on-line, or having an on-line transition alert stimulate the
beginning of a session.
Wireline operators have also discovered the problems associated with deploying point solutions as they
want to enhance their services. For example, if they would like to get a unified messaging solution from
Vendor A, and an IP PBX/Centrex solution from Vendor B, it is often difficult to get these solutions to
work well together, especially if Vendor B also offers unified messaging. So they are looking for a
standards-based service architecture that allows them to leverage things like common subscriber data and
underlying infrastructure cost effectively.
In addition, such a service architecture will be a key enabler to design and develop more lifestyle targeted
services that target niche markets cost effectively. Lucent believes that this is where the real subscriber
value will be unleashed to service providers. Without good service architecture, however, most of these
lifestyle services cannot be deployed profitably. Blending services together to create lifestyle service
offerings will be the logical next step for service providers to take, after implementation of a standards
based service architecture.
The standards-compliant IMS architecture is an excellent fit with service provider desires to create and
offer lifestyle services cost effectively. This enables individual voice and data services to be blended
together into more powerful services that are able to intelligently interact with each other by sharing
common information such as subscriber profiles, presence and location data, and buddy lists. It also
enables the interoperability of these services between subscribers on different access technologies
(Internet, 802.11, code division multiple access [CDMA]/universal mobile telecommunications system
[UMTS]) using different devices (laptops, personal digital assistants [PDAs], mobile phones, etc.) while
still providing backward compatibility with existing circuit-switched voice and data networks.
Only with such an intelligent, blended services delivery environment can service providers begin to offer
the high value, easy to use lifestyle services desired by end users. Lucents AccelerateTM IMS Solution
delivers such an IMS architectureenabling blended lifestyle services that deliver real value to end users
and profitable revenue for service providers.
Building on 3GPP and 3GPP2 Standards-Based Architecture
The IMS service intelligent architecture is based on IP core networking standards developed by the 3GPP
and the 3GPP2 bodies that establish the standards for 3G networks, providing an open environment that
allows efficient delivery of services to end-users while managing operators operations expenditures
(OpEx) and capital expenditures (CapEx).
336
The IP multimedia subsystem offers a structured multimedia service architecture, which adds service
intelligence to new and existing wireline and wireless networks. The IMS standard service architecture
supports a complete network solutions approach to voice over IP (VoIP) across all layers of the network. It
is a unified architecture that supports a wide range of services enabled by the flexibility of session initiated
protocol (SIP) while continuing to support H.323 and circuit endpoints. The IMS architecture separates and
extracts service and control types from access and client types making services available on both wireless
and wireline environments. Application-development partners can access the open IMS architecture, which
allows service providers to rapidly roll out new features and functionality (see Figure 1).
Access layer
Session/call management layer
Application layer
Each layer is buffered from the others by standard protocols that allow introduction of new devices and
functionality without drastic impact on the other layers.
Access Layer
The access layer consists of media gateways and end-points. It initiates and terminates SIP signaling to
set up sessions and provide bearer services such as conversion of voice from analog or digital formats to
packets using real-time transport protocol (RTP). The media gateways convert the VoIP traffic to public
switched telephony network (PSTN)/public land mobile network (PLMN) format where necessary, and
support media conversion between the control and payload processes.
The media resource server (MRS) also resides in the access layer and provides traditional telephony
services including conferencing, announcements, prompt and digit collection, speech recognition, speech
synthesis, etc. The media server resources are shared across all applications. Consequently, each
application that requires playing announcement, collecting digits, etc., can use a common server.
337
By using a common pool of media servers across multiple services, service providers can optimize their
footprint and CapEx investments. The media resource function controller (MRFC), which resides in the
session-management layer, controls the media stream resources needed to provide tones, announcement,
and conferencing.
Session-Management Layer
The session management layer contains the call session control function (CSCF), which provides the
registration of the endpoints and routing of the SIP signaling message to the appropriate application
server. The CSCF interworks with the access and transport layer to guarantee quality of service (QoS)
across all services. There are three types of CSCFs:
The serving CSCF manages all SIP sessions and coordinates with other network elements for
session control
The proxy CSCF is the first contact between the devices and the IMS core
The interrogating CSCF routes requests for connections within an operators network
The session-management layer includes the home subscriber server (HSS) database that maintains the
unique service profile for each end user. The end users service profile stores all of the user service
information and preferences in a central location. This includes an end users current registration
information, voice mail box options, (greetings), etc. By centralizing this data, applications can share
information to create unified personal directories, multiclient type presence information and blended
services. This centralized arrangement also greatly simplifies the administration of user data and insures
consistent views of active subscribers across all services.
The session-management layer also includes the media gateway control function (MGCF) and controls
establishment of bearer reservations for sessions that require interworking between the IMS and the
PSTN. The MGCF manages the distribution of sessions across multiple gateways. The breakout gateway
control function (BGCF) selects the PSTN/PLMN network in which the call termination is to occur or is
the entry point for incoming calls from the legacy PSTN/PLMN.
Application-Server Layer
The application-server layer contains the application servers, which provides the end-user service logic.
The IMS architecture and SIP signaling is flexible enough to support a variety of telephony and
nontelephony application servers. For example, SIP standards have been developed for telephony services
and IM services.
IMSsupported portfolio solutions should incorporate a comprehensive set of platforms and softwarebased, plug-and-play applications designed to support enhanced voice and data services. As networks
become more intelligent, operators have the opportunity to harness applications that create more robust
and personalized services for their subscribers.
Examples of solutions provided by a product portfolio might be as follows:
338
Telephony-Application Server
The IMS architecture supports multiple application servers for telephony services. The telephony
application server (TAS) is a back-to-back SIP user agent that maintains the call state. The TAS contains
the service logic which provides the basic call-processing services including digit analysis, routing, call
setup, call waiting, call forwarding, conferencing, etc. The TAS provides the service logic for invoking
the media servers to provide the appropriate call progress tones and announcements. If the calls are
originating or terminating on the PSTN, the TAS provides the SIP signaling to the MGCF to instruct the
media gateways to convert the PSTN TDM voice bit stream to an IP RTP stream and to direct it to the IP
address of the corresponding IP phone.
As part of executing the telephony call model, the TAS provides the advanced intelligent network
(AIN) call trigger points. When a call progresses to a trigger point, the TAS suspends call processing
and checks the subscriber profile to determine if additional services should be applied to the call at this
time. The subscriber profile identifies which application servers should be invoked. The TAS formats a
SIP IP multimedia service control (ISC) message and passes call control to the appropriate application
server. This mechanism can be used to invoke legacy AIN services or to invoke new SIP based
applications servers.
A single IMS can contain multiple TASs that provide specific features to certain types of endpoints. For
example, one TAS might provide the IP Centrex business features (example, private dialing plans, shared
directory numbers, multiple call appearances, automatic call distribution [ACD], attendant services, etc.).
Another TAS might support PBXs and provide advanced virtual private network (VPN) services. The
multiple application servers can interwork using SIPI signaling to complete calls between the different
classes of endpoints.
IP Multimedia Services Switching Function (IMSSF)
The IP multimediaservices switching function (IMSSF) provides the interworking of the SIP message
to the corresponding customized applications for mobile networks enhanced logic (CAMEL), ANSI-41,
intelligent network application protocol (INAP), or transaction capabilities application part (TCAP)
messages. This interworking allows the IP phones supported by IMS to access services such as calling
name services, 800 services, local number portability (LNP) services, one number services, and more.
Supplemental Telephony Application Servers
The application layer can also contain standalone independent servers that provide supplemental telephony
services at the beginning of a call, end, or in the middle of a call, via triggers. These services include clickto-dial, click-to-transfer, click-to-conference, voicemail services, interactive voice response (IVR) services,
VoIP VPN services, prepaid billing services, and inbound/outbound call-blocking services.
Non-Telephony Application Servers
The application layer can also contain SIPbased application servers that operate outside of the telephony
call model. These application servers can interwork with endpoint clients to provide services such as
instant messaging (IM), PTT, presence-enabled services, etc. By implementing these nontelephony SIP
based services in a common IMS architecture, it is possible to interwork telephony and nontelephony
services to create new blended communication services.An example would be a converged click-tocontact buddy list that displays end users presence and availability information, and provides a point and
click interface across multiple communication services (telephony, IM, and PTT). Another example
would be to use a single prepaid services account for telephony and VoD services.
339
For greatest effificency, a centralized application server can be deployed as host to multiple applications,
decreasing the need for multiple servers. This generic hosting environment is called a service delivery
environment (SDE) and has been implemented already in the IT environment. In the IMS architecture, the
SDE provides centralized support for SIP VoIP applications as well as legacy protocols for interworking
with the PSTN and existing mobile networks. As a high-availability application-server platform, SDEs
are designed to enable rapid development of applications that can reside on it in the network.
The SDEs provide switch and media independence so it can support deployments in various
environments such as multimedia (wireline, wireless, and data), multivendor, multiprotocol, and
multitechnology environments.
Another application supporting IPbased multimedia capabilities is the messaging server. This
application would support speech-enabled, next-generation, Web-based applications such as real-time
stock transactions and meeting-location services. This is provided along with voice, fax, and text
messaging. These systems provide end users with a single network based mailbox with the ability to
manage their communications of anytime, anywhere using wireless or wireline telephone, PDA, or
Internet Web browser for their voice-messaging, unified-messaging, and unified-communications needs.
Open Service Access Gateway (OSAGW)
The IMS architecture allows service providers the flexibility to add services into their VoIP networks by
interacting with legacy applications or by integrating SIPbased application servers that they purchase or
develop themselves. In addition, service providers want to allow their customers to develop and
implement services that leverage the VoIP network resources. For example, an enterprise may want to
voice-enable or IMenable some back-office operations to automatically initiate a call or an IM if an
order is about to be delivered. This could be triggered by the location information of a wireless PDA
carried by the delivery person. Frequently the enterprise application developers have IT backgrounds,
however, and are not familiar with the variety of complex telephony signaling protocols (example, SS7,
ANSI41, CAMEL, SIP, ISDN, etc.). To provide a simple application programming interface (API) for
communications services, the Parlay Forum working closely with the 3GPP and ETSI standards
development organizations have jointly defined a parlay API to telephony networks. The interworking
between SIP and the parlay API is provided in the OSAGW that is part of the application layer of the
3GPP IMS architecture. As described above, the other application servers provide the interworking
between SIP and the telephony protocols (ANSI41, CAMEL, INAP, TCAP, ISUP, etc.). The OSAGW
allows the enterprise-based parlay applications to access presence and call state information, set up and
tear down sessions, and manipulate legs of a call. Presence and availability servers will likely have the
broadest impact to communications applications and generate the most signaling related traffic on the
network. The OSAGW implements the parlay framework, which allows the enterprise application
servers to register with the network and manages access to network resources.
The OSAGW handles the task of interacting with legacy applications and SIPbased applications. The
OSAGW is an OSA/parlay standards-compliant mediation gateway that offers a rich set of mediation
and policy-management capabilities designed to enable application developers to incorporate network
presence easily and add availability, call control, location, messaging, billing, and authentication
functions into their applications.
The OSAGW makes it possible for operators to be more responsive to emerging trends by reducing the
time-to-market for the introduction of new applications and services. The complexities of the mobile
network are hidden from the application developer, who can rapidly develop new applications by
employing mainstream programming languages and commercial tools.
340
341
Emilio Fontela
Dean
The Nebrija University, Spain
Former Professor of Econometrics
Geneva University
1. Towards Third Generation (3G) and Beyond
Looking into the future requires some understanding of the past and an interpretation of the present.
Mobile telecommunications have received considerable attention from all possible points of view, by
technologists, economists, and social scientists. The sector is closely followed by financial analysts all
over the world. The number of publications dealing with the characteristics of the mobiles explosion is
unlimited. Furthermore, statistics are fully exploiting the on-line characteristics of the operations of the
sector, and, stimulated by the regulatory bodies of the telecommunication sector, are producing more
quantitative information about the development of mobiles than for practically any other sector of
economic activity. The data for such a New Economy activity is more detailed, more easily available, and
perhaps even more reliable than for any other important sector of the Old Economy (like steel products,
chemicals or motor cars).
The following analysis of the broad economic and social impacts, and of the new challenges faced by
mobile telecommunications, moves above this set of information and does not attempt another detailed
description of the activities of this sector.
1.1 The Social and Economic Impacts of Mobile Communication
Innovation is knowledge transformed into wealth. For centuries, the creation of wealth was linked to the
use of land, energy, and natural resources, through hard human work. After an Agricultural Era, came an
Industrial Era; great innovations allowed for the development of manufacturing. Currently society is
witnessing a new process of change towards a Human Resources Era and towards a Knowledge Society,
more ethereal and services oriented. In this new stage of economic growth, the information and
343
communication technologies (ICT) are expected to provide the core of tools necessary for a full diffusion
of knowledge.
When knowledge becomes the key asset of humanity, wealth is necessarily linked to the capacity for
innovation. This is why modern societies are driven by the innovation process.
Some innovations change the way we produce goods and services, others change the characteristics of
these goods and services and the way we consume them. Innovations increase both productivity and
quality of life.
Among the great innovations of the second half of the last decade of the 20th century, the mobile phone
ranks at the top of the scale, and changes both consumption patterns and production processes. The
mobile phone, like the fixed phone, provides at the same time consumer services and business services.
The Story of Mobiles
As early as 1947, the Bell Labs had developed cellular phones. But as often happens, there is a long delay
between invention and innovation. Practical applications of mobile cellular phones started in Sweden in
1981, and in Chicago in 1983.
The explosion of the demand and use of mobile phones had still to wait for a decade, and was a clear
output of the processes of liberalization of telecommunication markets that were progressively adopted in
all Organization for Economic Cooperation and Development (OECD) countries; combined with a
normative standardisation scheme, most public telecommunication monopolies were privatized, fully or
partially, and competition has been implemented through deregulation.
The success of mobiles is a prototype of the contribution of competition to economic development. With
the advent of mobile phones all communications have drastically increased, including those with fixed
installations.
The social impact of mobiles is rather obvious: it has brought voice communications to situations where it
was clearly needed, but was not technically available. Consequently it has changed important aspects of
life styles; parents can keep permanent contact with their children, friends can exchange ideas in all
circumstances. The care of the older, the handling of accidents, the fear of loneliness have been
substantially modified, for the best. The increased possibilities for human communications by mobiles
have undoutbly contributed to well-being. Social life has been transformed. We cannot anymore envisage
a world without mobiles, just as we cannot think of living without electricity.
However, the economic impact of mobiles is often underestimated, because it is certainly less obvious.
But a second look at this innovation process will show that mobiles have impacts at both the structural
mesoeconomic level where sectors explain the creation of wealth, as well as at the macroeconomic level
that explains economic growth.
A New Industry
Mobiles stimulate the development of a complex new industry, including both the advanced technologies
required by the terminals and networks, and the organizational and human capabilities required by the
services to the final users.
344
345
Table 1
In terms of penetration rates, we start again from almost zero, and an epidemiological process is to be
expected, in a similar way as already observed for mobile voice transmission.
Sometime in the future this new market for mobile data should also observe a saturation of terminals and
lines, but in this case it becomes more difficult to envisage a saturation in terms of contents. While in the
case of voice telecommunication the demand is constrained by the social or economic behaviour of the
users, in the case of data and information the driving force is on the side of supply and of its capacity to
interpret hidden needs of the potential users.
The sequence adopted for the introduction of the new 3G services is seen in Figure 1:
Figure 1
346
Obviously we could expect a migration process for a significant number of the current 2G subscribers to
3G services, much along the lines that were observed at the turn from 1G (analogic) to 2G (digital)
systems. The migration rate will be a crucial factor of success or failure for the 3G systems.
Most corporations are playing heavily on the expectations of these new developments, even more that
what they played in the early stages of the mobile phone. They have run the risks of unusually high levels
of indebtness to afford the rights to contend in the new markets. They have discounted very high rates for
future returns. They are implicitely convinced that the S-curbs for the new system will behave at least as
was the case for mobile phones, that epidemiological processes will again run the show. In 2001, the
growth observed in all countries for short written messages (SMS) is certainly encouraging at this respect.
However, with 2.5G technologies the industry is still using most of the infrastructure of the 2G
technologies, and the change in terminals, even if important both in terms of components and prices, is
not yet as radical as requested by the 3G quantum jump to the broad band system.
Under these conditions, as soon as the first symptoms of a slowdown of the New Economy and of the
demand for voice communication have become noticeable, there has been substantial increase in the level
of uncertainty. Maybe the future is not as bright as expected. Maybe the key links with Internet will rather
flow towards fixed PCs. Maybe the passage from user needs to user effective demand will force
prices down and seriously affect the level of expected returns.
For the industry, this is the right moment to look into the future, and this is why there is a frenetic set of
meetings and expert reports that intend to provide insights about future developments.
This effort of looking into the future is conducted by the sector itself: the suppliers of terminals and
infrastructure, the regulatory agencies, the specialized consultants, the companies producing
transmission services or contents all are practicing forecasting, and foresight, and are attempting to
explore future developments.
Forecasts are being elaborated at a vertiginous path; most of them however are proprietary, i.e. available
only to clients of the corresponding forecasting service and requested to be treated confidentially.
We have decided to concentrate on the Universal Mobile Telecommunications System (UMTS) Forum
reports in order to reflect the conventional wisdom forecast at the present moment (we will mainly refer
to the UMTS Third Generation Market Study Survey of August 2001, and the Long Term Potential
Remains High for 3G Mobile Data Services report of February 2002.
The UMTS Forum basic positions related to 3G services are the following:
Complex services will take time to develop and services will be adopted slowly through 2005 in
the developed countries.
Issues involving handsets and 3G service delivery are short-term and resolvable and will not
affect the longer-term viability of 3G services.
The demand for 3G mobile data services is real. Consumers and business users have consistently
demonstrated strong interest in trying new services that combine mobility with content and
customisation.
The challenge is to create compelling services that take advantage of the promise of 3G
technology and provide satisfactory end-user experiences.
347
These positive views are the result of many market studies and of the continuous monitoring of
development in production and consumption.
In developing their forecasts, UMTS Forum/Telecompetition have considered a number of country
specific factors for each service, such as the following:
The sequence adopted for the introduction of the new 3G services is as shown in Figure 1.
A synthesis of the forecasting work of UMTS Forum is provided below in Table 2, in which the total
amount of 3G connections is estimated, as well as the resulting revenues, under rather conservative price
forecasts. (Average revenues per unit, while higher on average to those of 2G services, are expected to
decline during the next years, along the lines followed by all innovations).
Table 2
With an expected European Union (EU) population of 375 million in 2005 and also in 2010, the UMTS
Forum/Telecompetition Inc., forecasts a penetration rate of 3G of 13%, in 2005, growing to reach 43% by
the end of the decade. That is an evolution rather similar to the one we have observed in the 90s for 1G
and 2G mobile phone connections. The revision made by UMTS Forum in February 2002 slows down the
growth process during the first years but does not change the final 2002 projection.
The bulk of the expected connections to use 3G lines is expected to be provided by simple voice, a
fact that confirms that this new technology already relies on markets developed by previous
348
technologies. In a sense, while the penetration rate of 3G is expected to increase, the penetration rate
of 2G should start decreasing.
In terms of subscriptions, customised infotainment ranks second and multimedia messaging service
(MMS) and mobile Intranet/Extranet access are in third place. The picture is rather different if we look at
revenues, in Table 5.
In the case of future revenues, voice revenues remain obviously important, but customised infotainment
produces the same amount of revenue, and the business applications for Intranet-Internet access are
clearly in third place. The reason for this change is simple, and has to do with the average revenue per
subscriber. It is as low as 150 Euros for voice, and as high as 270 Euros for Intranet-Internet. The revenue
will therefore greatly depend on the value-added content of the services provided.
Table 3
349
Tables 4 and 5
Consider any large technological innovation such as mobile telecommunications. It is applied in many
fields. But how these applications develop? It is a result of decision-making constraint by circumstantial
situations. Some applications come first because they find either the appropriate supplier or a demand
adapted to the market price conditions. Sometimes the application leads to a fast diffusion process,
because the right marketing approach is adopted, or because the demonstration effect of new consumption
is very high.
In any case, it is obvious that we live only one of possible futures of the past. Today most motorcars use
oil, but with a different set of circumstances we might be using solar energy.
As a result of this we may never be sure that the applications of a given innovation, that are taking place
at a certain moment, could not be different, perhaps very different, even when starting with the same
knowledge and technology.
Morphological analysis is a tool of futures research designed to explore the possibilities offered by a
given technological field, using the combinations of several dimensions characteristic of the system
under study.
350
351
It can rapidly be seen that these three dimensions are interacting between themselves. The cost for the
user is necessarily related to the value added he receives form the service; the technology is there to make
possible the development of the service, and the computation of its cost; value-added services require
specific technologies. The fact is that these three dimensions should be analysed together if we want to
understand the deep morphology of mobile communications.
2.2 The Cube of Mobile Communications
The interaction of the three dimensions described above, the technological dimension (data rate), the
economic dimensions (cost) and the content dimension (value added) can be portrayed in a tridimensional space, in a cube. The inside of this cube has been divided into tri-dimensional components,
and in order to simplify the analysis, three segments have been considered for each dimension:
The data rate deals with speed and width of the flows. The three segments retained correspond to the
mobile generations:
The cost dimension can be materialised by the average revenue per user (ARPU), and in this case the
segments correspond to the following:
Finally the value added dimension has been analysed with some characteristics of the contents of the
flows, with the following three segments:
If we portray geometrically these three dimensions, we obtain a cube, the inside elements of which
correspond to a combination of the dimensions (see Figure 2). All the elements of the cube are not
perfectly covered by existing services, that is, there are some empty market segments that, in terms of
morphological analysis, consists of possibilities yet to be explored.
The empty mini-cubes correspond to potential market segments which are not evidently covered by
current or future relevant groups of services. The detailed tables following the cube provide a spatial
distribution of identified services. Such a distribution is certainly subject to discussion. However, there is
no doubt that the entire market potential is far from being addressed by the services that are currently
offered or envisaged by the industry, and it is clear that much remains to be done in terms of targeted new
services able to be economically provided adequately to each category of current and future customers.
352
Figure 2
In Table 6 through Table 8, we summarise the content of the mini-cubes at each of the three data rates
levels.
353
Tables 6 8; Figure 3
In Table 6, it is not surprising to see a lack of high added value services offered here due to the limitation
of the 2G technology. However, the relative marketing failure of the WAP offerings is also to be seriously
studied as an example of a lack of perception of the market actual demand in terms of Price/performance
acceptance. The price war between operators to capture subscribers has been the dominant factor behind
354
the spectacular growth of the number of declared subscribers. However, as mentioned earlier, the weight
of pre paid customers is already close to 50% and a major objective for the operators is to turn this
volatile market into steady revenue generating users.
There is a consensus regarding the emergence, thanks to the deployment of the 2.5G technology, to a
brand new set of added-value services for both business customers and consumers as portrayed in Table
7. The clear challenge here relates to the costs affordable by the customers compared to the cost of
provision of such services. We introduce here some m-commerce services (in B2), assuming that the
industry will be able to convince the public about the security and confidentiality of commercial
transactions via the mobile network. Access to Internet portals will have to be properly designed in order
to provide a quick and customised answer to users needs. The 2.5G performances and the ergonomic of
any handset imply a major attention to the types of solvent services to be actually paid for by the general
public. Operators should carefully consider to avoid the repetition of the WAP fiasco to preserve their
credibilityat the time of arrival of the 3G services.
Notwithstanding the postponement concerning the actual deployment of 3G networks and the pending
crucial questions related to the economic balance of this revolutionary technology, there is no doubt that
sooner or later it will be put on the market on a large scale. Already some operators are engaged in such
process and the existing competitive environment will force all major players to follow. The key
challenge here will be to really create a public perception that 3G products and services offerings bring a
true revolution in term of services. This cant be based only on some attractive gadgets the costs and
usefulness of which will rapidly appear as prohibitive, but on the provision of truly innovative functions
impacting in a significant way to the business and private life of the general public as it has been the case
with the emergence of cellular telephony.
2.3 The 3G Challenge
It seems obvious that with the 3G infrastructure deployment, operators and manufacturers will be able to
offer the complete set of products and services already provided with the previous technologies (i.e: 2G
and 2.5G). In some cases more sophisticated services will replace the former ones at a similar cost or even
cheaper. In other cases, only the cost will be lowered thanks to the enhanced performance of the system.
The process is rather similar to the one observed from analogic to digital technologies: the number of
analogue mobile connections is continously decreasing while digital connections are increasing. The
density of the services offered is therefore expected to move from the 2G level to the 3G level for simple
reasons of overall efficiency of the communications system. The rhythm of the migration will depend on
the renewal of the stock of cellular phones, and this process itself is a function of the attractiveness of the
new services offered by 3G.
Is there an area in which 3G new services is yet to be explored?
To project the morphological cube into the 3G surface is to examine the resulting situation should all
services that today characterize 2G and 2.5G were offered by 3G.
The projected surface is now a matrix (Figure 4) showing all possible services at the 3G level. Simple
inspections of the result shows that the morphological approach ends up with one market segment
uncovered by any kind of identified service. It is all the more interesting that such an empty case
pinpoints a crucial market segment, that is, low cost, high added value services. To ensure the
profitability of 3G infrastructures deployment, operators have to capture not only the high income and
business market segment, but also the mass market. This is the one which has to be offered services fitting
with the identified white square materialised in the Figure 4.
355
Figure 4
Of course there are at least two ways to develop products for this empty market segment. One is to lower
the cost of high value added services. Another is to increase the value added of low cost services. In the
first case, we could obviously think of lowering the cost of m-commerce or of adult infotainment; in the
second case one should think about increasing the value added content of rich voice services (perhaps
along the lines of future videophones). Using the UMTS Forum revenue forecasts (see Table 5), it is
possible to quantify the matrix (shown in Figure 4) for the services specifically provided with 3G
adopting the correspondances shown in Table 9 (see Figure 5) .
Table 9
356
Figure 5
Final Consideration
The analysis of characteristic dimensions of the m-communications system has been based in the
interaction between technological performance (characteristics of the different generations), cost for the
user, and value added. The interaction of these dimensions allows for the positioning of all mcommunications services into a morphological cube (see Figure 2). The result of the systematic study of
the contents of this cube shows that the supply of new 3G services is specially weak in high value added
and low-price applications (see Figure 4), that are characteristic of mass consumer markets.
There are at least two ways to develop products for this empty market segment. One is to lower the cost
of some existing high value added services; another is to increase the value added of existing low cost
services. In the first case, we can think of lowering the cost of m-commerce or of adult infotainment; in
the second case one can think about increasing the value added content of rich voice services (perhaps
along the lines of future videophones).
357
UMTS 3G Technology in
Broadband Wireless Applications
Tom Flak
Vice President of Product Marketing
SOMA Networks, Inc.
359
key market requirements for broadband wireless deployments and then follow with a technical analysis
on the feasibility of 3G technologies in broadband wireless applications.
Until recently, earlier attempts at broadband wireless deployments, particularly those for the mass market,
met with limited success. While the technology used in these deployments worked on a technical level,
most failed to meet the basic business case requirements of the service providers that wanted to deploy
them. Capital and operational costs were too high, and the revenue that could be generated from the
services offered was too low. The shortcomings of these earlier-generation broadband wireless systems,
however, are very instructive for identifying the key requirements for a profitable service provider
broadband wireless business case: reducing the initial capital investment required to get into business and
allowing for a pay-as-you-grow model; minimizing the operating costs necessary to run the business; and
maximizing the revenue opportunities supported by the network.
Requirement #1: Large Cell Radius
One key factor in a service providers business case is the number of cell sites required to establish
market coverage. The number of cell sites drives the initial costs for market entry and the operational
costs associated with tower rentals, backhaul, and other site-related expenses. Minimizing the number of
cell sites implies the need to maximize the cell area.
Requirement #2: Base Station Capacity/Area > 675 kbps/sq-km
Capacity delivered over the coverage area varies according to local market conditions, but 675 kbps per square
kilometer is a typical figure derived from reasonable assumptions. Factors in this estimate are as follows:
Throughput/area = (bandwidth/sub)(market share)(broadband penetration)/(HH/sq-km)
The bandwidth per subscriber factor is a rule of thumb derived from broadband operators empirical
experiences provisioning cable and DSL service. It is related to the relatively low duty cycle for data
services that enable high ratios of overbooking available bandwidth. A fairly aggressive figure is
currently considered to be around 15 kbps per subscriber. A reasonable objective for the market share
factor is 1/3, assuming a competitive market where cable, DSL, and fixed-wireless service share the
broadband business. Current broadband penetration varies, but in high penetration markets, 45 percent is
a typical figure. Lastly, household density ranges widely, but a typical suburban scenario is approximately
300 households per square kilometer. Putting these factors together gives the 675 kbps per square
kilometer target:
Throughput/area = (15 kbps)(1/3)(45%)(300/sq-km) = 675 kbps/sq-km
Requirement #3: Unlimited Network Capacity
Traditional point-to-multipoint microwave systems suffered from finite capacity limits. For broadband
wireless technology to reach mass adoption levels, it must offer the means to add capacity to the network
to satisfy customer demand. The mobile wireless industry, for example, accomplishes this objective
through spatial reuse of spectrum in cellular coverage architecture.
Requirement #4: No Truck Rolls
Past attempts to deploy broadband wireless technology in advanced economy markets failed, at least in
part, due to the high cost of service installation. One characteristic of developed economies is that labor
costs are high, so the loaded costs per installation for technicians, trucks, tools, and management can add
hundreds of dollars to each subscriber installationextending payback by several months. For broadband
wireless to go mainstream, the technology must eliminate the need for professional installation.
360
361
Scheduling multiple users packets onto a shared medium at a fast rate enables the efficient use of that
medium. The medium is never idle as long as at least one user has traffic to send. This is commonly
called a packet-switching (PS) paradigm. This is in contrast to the older circuit-switching (CS) paradigm
that establishes physical circuits between the source and destination to carry traffic. The biggest problem
with circuit switching is that it often leaves the medium idle when the single user assigned to the medium
has no traffic to send. 3GPP Rel99 uses dedicated channels with a circuit-switching paradigm to carry
sporadic Internet traffic. This was a very inefficient use of the scarce air interface bandwidth resource.
Switching to a shared channel that is shared by many users dramatically increases the utilization of the air
interface bandwidth.
A potential drawback of scheduling onto a shared medium is an increase in the latency for a single user,
as one must wait for the medium to become available. While it is impossible to reduce latency to zero,
designing the channel physical layer transmission time interval (TTI) to be as small as possible can
dramatically reduce latency. 3GPP Rel5 reduced the HSDSCH TTI to 2ms from the standard 10ms TTI
used for dedicated channels.
The throughput of the HSDSCH is further increased relative to dedicated channel through the use of
adaptive modulation and coding during each TTI. The HSDSCH can use quadrature phase shift keying
(QPSK) or 16 quadrature amplitude modulation (QAM) in each TTI at a variety of coding rates. This
flexibility allows the scheduler to choose a modulation and coding scheme that will yield the highest data
rate during a given TTI.
The scheduler chooses a particular modulation order and code rate for each TTI based upon channel
feedback from the user to be scheduled during a particular TTI. The user periodically sends information
on its measured channel quality to the base station, and the base station uses this information to choose
the modulation order and code rate that yields the highest data rate the channel can currently carry. This
adaptation loop ensures that a user is always receiving data at the highest rate possible.
In spite of the short TTI and fast channel quality feedback, it is still possible that the receiver will
incorrectly decode the received signal. Hybrid ARQ (HARQ) was added to the HSDSCH to provide an
extra layer of protection to the data steam. In the event of a decoding error at the subscriber device, the
HARQ algorithm sends a message back to the transmitter indicating that the last transmission was in
error. Depending upon the configuration, the transmitter then either sends additional parity information
that the subscriber device can use with the first transmissions data to decode the original message or the
transmitter sends the message again. HARQ is fast because it operates on TTI time scales and hence is
invisible to the end-to-end Internet protocol (IP) traffic. Ensuring channel reliability is particularly
important for transfer control protocol (TCP)based Internet traffic as TCP is extremely sensitive to
channel errors.
With a short TTI, channel feedback information from each user, and a multiplicity of users on the system,
it is possible to conceive of many shared-channel scheduling disciplines at the base station. The round
robin (RR) scheduler is in the simple category, servicing the various users in order, until each has been
serviced. It then goes back to the first user and starts over. This scheduler does not take advantage of the
channel feedback information available at the base station. It is possible to construct scheduling
disciplines with higher aggregate throughput than the RR scheduler. The proportional fair (PF) scheduler
improves the aggregate throughput by priority scheduling the users who happen to be able to achieve
higher data rates during any particular TTI. The scheduler includes a discipline that allows the scheduling
of users who cannot achieve high data rates to ensure that everyone gets a fair share of the channel.
To facilitate fast packet scheduling and fast channel feedback, the architecture of 3GPP Rel5 had to be
slightly modified relative to previous releases. Some of the packet scheduling features that previously
362
were handled in the radio network controller (RNC) had to be moved to the radio base station (Node B) to
reduce the control loop latency. Backward compatibility to previous releases was considered in the
architecture change.
High Data Rates and Broad Coverage
Wireline access has set the bar on user data-rate expectations. Broadband access has been defined by DSL
and cable-access technologies to be greater than 1 Mbps. In addition to being broadband, modern data
services must also have broad coverage. The dual requirements, broadband and broad-coverage, are key
enablers for the success of a retail access system.
Wireless systems that are accessible to retail consumers have been challenged to meet this dual goal.
Retail systems have achieved the peak data rate goal with local area network (LAN) technologies such as
the IEEE 802.11 family and with point-to-point (PTP) or point-to-multipoint (PMT) proprietary
technologies. But peak rates have only been achieved in either small-scale spotty coverage LAN
deployments or in expensive-to-install and retail market inaccessible WAN deployments. The current
wireless LAN hot-spot market is a good example of spotty coverage. And the fragmented PTP and PMT
market with proprietary niche technologies has never been structured to address the retail market needs.
Todays wireless market requires a technology that can fill the gap between these two technologies: one
that achieves broadband access data rates and broad coverage.
Mobile access technologies have been attempting to bring data coverage to the retail consumer since the
inception of the cellular system. Old analogue mobile phone service (AMPS) systems, the birthplace of
the retail mobile phone, had add-on, dial-up, data service. They satisfied the broad coverage requirement
but delivered mediocre data rates at high cost. One cannot even consider applying the term broadband to
such a system.
One must remember, however, that early mobile access systems were not specifically designed to provide
data services. Voice was the killer application, and it is, by its very nature, narrowband. Voice is still the
killer application for mobile service, but it is no longer the only application. The demand for data services
such as e-mail and Web browsing is rapidly growing. Add to that bandwidth-consuming applications such
as video and voice, and those are the forces that have driven the evolution from old narrowband voice
systems to the high bandwidth data systems required today.
The UMTS mobile standards have always been designed to enable broad coverage for retail users. Now,
with 3GPP Rel5, the high data rate requirement has also been satisfied. The addition of HSDPA has
increased the peak data rates in excess of 10 Mbps. Several key leading-edge technologies have been
added to the standard to achieve broadband access rates. Adaptive modulation and coding, high order
modulation, hybrid ARQ processing, and channel quality feedback mechanisms have been specified.
By evolving to keep pace with customer demands in the mobile retail market, the worlds leading mobile
standard has managed to create the solution that satisfies both the data rate and coverage requirements
demanded by the retail broadband wireless access market.
IP Layered Design
The Internet has become the killer platform on which to deploy services. It is widely used in the business
place and has increasingly become a must-have service in the residential market. Nothing can escape its
grip. Even traditional voice services are migrating toward delivery on IP infrastructure with the help of
enabler applications such as the session initiation protocol (SIP), an application layer protocol that can
establish multimedia sessions or Internet telephony calls, and modify or terminate them. Wireless access
technologies have begun to embrace this trend by designing the systems for IP from the ground up.
363
A key attribute of an IP layered design is the capability to support multiple classes of service, including
real-time streaming media, voice, guaranteed data, and best-effort data. Each of these classes may use a
unique management protocol and require specific quality of service (QoS) support during transport. To
support such services, a wireless system needs to be effectively integrated with the IP access and transport
QoS mechanisms.
In wireless system design, support for QoS is built from the ground up starting with the PHY and media
access control (MAC) layers. Physical layer (PHY), packet-switched, wireless channels with MAC
priority scheduling, based upon IP QoS attributes, is the best way to ensure that IP end-to-end QoS is
maintained. Fortunately, HSDPA has set the foundation for making this possible.
UMTS 3GPP Rel5 has embraced the IP services framework with the introduction of the IP multimedia
subsystem (IMS). IMS is a framework that can support the integration of IP services into the 3GPP
architecture. Services such as real-time and nonreal-time traffic, instant messaging, streaming media, and
gaming are but a few examples considered in the formulation of IMS. They have also embraced a variant
of SIP as a key element of the IMS architecture.
With the HSDPA high-speed data channels at the PHY and MAC layers and the IMS architecture at the upper
layers, 3GPP Rel5 is well positioned to provide always-on IP connectivity to the residential consumer.
Cellular Frequency Reuse
Cellular frequency reuse concepts have been the key enabler for the success of cellular networks.
Frequency reuse allows the same carrier frequency to be reused multiple times in a relatively small
geographic area. This increases the geographic spectral efficiency measured in bps/Hz/area.
Mobile cellular systems have evolved from 1:7 frequency reuse patterns where clusters of seven cells
used seven carriers before repeating the frequency pattern, to 1:4 reuse patterns in TDMA systems, to
todays highly efficient 1:1 reuse patterns supported by CDMA systems. CDMA systems have the highest
geographical spectral efficiency of any mobile or broadband wireless access scheme to date.
In addition to having the highest spectral efficiency, the CDMA 1:1 frequency reuse schemes are the
simplest to upgrade should more base stations be needed to cover hot spots. The new base station requires
no frequency planning, as it uses the same frequency as the adjacent base station. This is dramatically
simpler than an upgrade process in a 1:N frequency reuse deployment where one may have to redeploy
frequencies over a number of cells to create the space for the new cell in the deployment.
UMTS WCDMA systems inherit the time-proven 1:1 frequency reuse scheme, and can continue to use it
in the deployment of HSDPA.
The proprietary PTP and PMP systems still, for the most part, use technologies that require frequency planning
should they attempt a cellular deployment. In fact, many PTP systems never considered using a cellular
architecture and instead opted to deploy a single macro cell to cover the entire city. This kept infrastructure
costs low as only one base station was deployed, but it limited the total number of users in the city to the
capacity of the single base station. Without a cellular architecture, the systems often failed as the service
became popular because extra capacity to support new customers could not be added to the deployment.
Collision-based wireless access technologies such as 802.11 also have difficulties in 1:1 frequency reuse
deployments. The number of media access collisions increases dramatically when more users are added in
an adjacent location at the same frequency, and this in turn dramatically reduces the available bandwidth.
These systems have to use 1:N frequency reuse plans to control the total interference and to prevent
precipitous drops in available bandwidth.
364
In fact, wireless access technology needs to be designed from the ground up with 1:1 cellular frequency reuse
to work effectively in such deployments. It is not a feature that can be added on later. CDMAbased systems
are currently the only time-proven systems built from the ground up to support 1:1 frequency reuse.
User-Deployable CPE
One of the major roadblocks that has traditionally kept wireless access out of the residential market is the
cost to professionally install the customer premise equipment (CPE). Point-to-point and point-tomultipoint systems for the most part have been designed to require an experienced technician to locate an
appropriate spot for the CPE antenna on the outside of the customers premises, install the antenna, and
then manually point the antenna towards the serving base station. This helps the link budget (and hence
the coverage) that is particularly important in low-power transmitter systems. But the cost of this process,
which often exceeds the cost of the CPE, restricts the market to businesses and well-healed consumers in
the residential market. Professionally installed CPEs have to be eliminated to open the door to the
residential market.
To be user deployable, a CPE must have several attributes. It must be deployable inside the customers
premises. Deploying an external antenna and running a cable inside the house is definitely beyond the
desire and often beyond the skill of most residential consumers. In addition, the indoor deployment must
not be overly restrictive. Consumers should have the option of placing the unit on their work desk, in their
bedroom, or even in the basement. Finally, the unit should have the ability to provide feedback to the user
on the quality of the deployment. Without a doubt, the signal strength will vary according to the location
in the house. An attic deployment will generally provide higher signal strengths than a basement
deployment. The CPE should help guide the user through the deployment by giving ongoing feedback
that can be used to optimize the placement of the unit within a chosen room.
Nonline-of-sight operation is key to the operation of high data-rate services with indoor deployments.
Indoor deployments are by their very nature rich-scattering and reflective environments that create
multiple transmission paths from the base station transmitter to the CPE receiver. The signal is
reflected by multiple interior walls and scattered by near-by furniture. The environment outside of the
house also contributes to the scattering and adds additional reflections. The neighbors house across the
street can act like a big reflector. Trees around the house and in the neighborhood scatter the signal. All
of these environment factors create a complex radio propagation environment that the system must be
able to handle.
To combat these effects, the receiver needs to employ advanced signal processing algorithms such as an
equalizer to combine the multiple reflections back into a single coherent signal from which the data
stream can be recovered.
Mobility systems based upon CDMA technology traditionally have not used CPE equalizers for several
reasons. They add cost to the CPE, they consume power, and they are not strictly needed for CDMA
narrowband voice systems. The traditional Rake receiver has been the best choice for narrowband
services as it is inexpensive, simple and consumes minimal power and has acceptable radio performance.
The advent of HSDPA in 3GPP Rel5 changes the narrowband operation assumption. Using a Rake
receiver for high-order modulation like 16QAM on low spreading factor channels (e.g., SF=16), while
possible, is suboptimal. In the power-consumption-sensitive mobility handset market, many
manufacturers will most likely stay with the Rake receiver. But the broadband wireless market is uniquely
positioned to build more complex power-consuming receivers with equalizers that yield the best
performance in NLOS deployments.
365
Interestingly, many of these future 3GPP design objectives and planned innovationswhich enable the
performance required in broadband wireless deploymentsare also envisioned for inclusion in the
developing WiMAX standard. Should the WiMAX standard gain real market traction and demonstrate
that its performance attributes meet the service providers business case requirements in the coming years,
a system following this 3GPP+ strategy would be well positioned to transition to WiMAX. Therefore, a
broadband wireless system with innovations built on top of a base standard such as 3GPP is in the
enviable position of being able to take advantage of the benefits of standardization in the wireless
broadband space, either from the 3GPP or WiMAX camps.
Companies such as SOMA Networks have recognized the numerous benefits of 3Gbased systems and
have been working to engineer such a system. By using the industry standards that exist today, but
building on top of those standards to meet and exceed the performance requirements of service providers,
these type of systems offer a commercial-ready solution that delivers a superior service provider business
case. The 3GPP+ solution enables competitive local exchange carriers, incumbent phone companies, rural
service providers, carriers in rural and emerging economies, and long-distance and backbone companies
to achieve higher revenues with the benefit of lower capital and operating costs. In particular, these types
of solutions are optimized to mitigate the significant operational costs associated with traditional
broadband infrastructure and first-generation broadband wireless systems that require truck rolls,
professional installations, and frequent customer-care intervention. 3GPP+ solutions can meet the
broadband wireless market requirements by delivering intelligent subscriber terminals with an internal
antenna and over-the-air self-activation capability. With plug-and-play capabilities, service providers can
implement a true retail distribution model that offers customers immediate gratification and ensures
instant customer gratification, high customer satisfaction, minimal churn, and rapid time to revenue. A
large cell size enables service providers to begin deployments with limited upfront capital costs and then
expand capacity incrementally as the demand for service grows.
The future of broadband wireless is bright, and the technology to deliver high-speed data and IPbased
voice services is finally here. The key is to take advantage of the best-of-class technology that will most
effectively meet the service providers business case requirements. The existing standards have yet to
catch up, but that is only a matter of time.
Bibliography
1.
2.
3G Americas white paper, The Evolution of UMTS3GPP Release 5 and Beyond, June 2004,
www.3gamericas.org.
3.
T. Kolding et al, High Speed Downlink Packet Access: WCDMA Evolution, IEEE Vehicular
Technology Society News, February 2003.
4.
3GPP, Physical layer aspects of UTRA High Speed Downlink Packet Access (Release 4),
TR25.848, v4.0.0, March 2001.
5.
3GPP, UTRA High Speed Downlink Packet Access, TR25.950, v4.0.0, March 2001.
6.
367
A Mobility-Management Scheme
in AllIP Integrated Network
Md. Abdur Razzaque
Department of Computer Science and Engineering
University of Dhaka
Zia Ush-Shamszaman
Department of Computer Science
Bhuiyan Computers, National University
Samiul Bashar
Department of Computer Science
Bhuiyan Computers, National University
existing and emerging IPrelated technologies and services [1]. The large-scale support of data services
and their integration with legacy services are the common objectives of all wireless efforts termed 3G and
beyond. Mobile operators are transitioning toward 3G networks and beyond in order to provide highspeed data access and sophisticated services, predominantly based on IP. Many operators have also
expressed an interest in deploying several access technologies in a seamless converged IPbased network.
For example, 3G cellular access, based on the code-division multiple access technology (either
widebandcode division multiple access [CDMA] or cdma2000), may be used to support users who
desire higher mobility over wider coverage areas, and broadband access based on the IEEE 802.11 [1]
specification to support users with relatively lower mobility over smaller geographical areas. The
remaining challenge is the design of such a transport infrastructure, which takes full advantage of IP
based technologies to achieve the desired mobility between the various access technologies, and at the
same time provides the necessary capabilities in terms of quality of service (QoS), robustness, and
manageability to unleash the potential of emerging 3G services.
This paper will attempt to address this issue, with focus pertaining to the aspect of mobility management.
Mobility management allows a roaming user to seamlessly switch between the different service areas.
The aim of this work is to design a mobility-management scheme for integrated networks. We analyzed
global system for mobile communications (GSM), general packet radio service (GPRS), and mobile IP
networks in this phase and sorted out strengths and weaknesses of existing wireless communication
system (WCS) systems such as GSM, GPRS, and Mobile IP.
The allIP integrated network overcomes those limitations exits in various network of mobile
communication and combines the strength of each network. Finally, a mobility management scheme for
integrated networks is proposed.
What Is the AllIP Strategy?
In current wireless communications, different services are provided over different networks. For example,
a circuit switched network is used for voice services and an additional packet network for data services.
This is illustrated in Figure 1. If these networks can be integrated into a single network such that all
services have common facilities, there will be an increase in efficiency and cost savings. Hence there is a
migration toward allIP, which is an integrated global communications network based on IP for providing
video, voice, and data services to the cellular handset or handheld Internet appliances. The allIP network
is illustrated in Figure 2. Referring to Figure 2, an allIP network can be defined as having the following:
370
An IP core network architecture that consists of a network unit that manages and controls
network resources, a call agent that handles all call services and a mobility management unit that
deals with the location update process.
IP radio access network (RAN) architecture with a radio network controller that provides access
for the various networks to the core IP network.
Md. Abdur Razzaque, Zia Ush-Shamszaman, Samiul Bashar, Jobeda Khatoon Sumi, and Safina Showkat Ara
Circuit
Switch
Network
Radio
Access
Network
Packet Core
Network
MT
GW
Networ
GW
Packet
Node
ISDN
PSTN
Internet
Intranet
Services
Unified
Communication
Voice
features
ECommerce
Web
Hosting
Location
Services
Core Network
Internet
Core IP
Network Call
Management
Mobility
Management
Access Network(s)
Other
Feature
ISDN
WLAN
GSM
CDMAOne
UMTS
CDMA
OFDM
An allIP network allows fast access to all services from a single device such as a cellular
handset or handheld Internet appliance. This is the main reason for developing a network, which
is capable of providing services of different natures within a same network.
371
An allIP wireless core network provides a wide variety of services for the users such as voice,
data, and multimedia. There is an increasing growth in wireless data traffic relative to demand for
voice traffic. An allIP network is able to cater to this need.
An allIP network supports many different access technologies, such as wideband code-division
multiple access (wCDMA), Bluetooth, and HyperLAN. Hence the core IP network can be
developed independently from the access network and allows seamless movement between
wireless and wired environments.
An IPbased network offers cost advantages. It eliminates the need for much of the equipment
typically found in a current wireless network, such as the circuit-switched core, because IP
system equipment can be used to support both the voice communication service and packet
communication service. Thus, there is reduced network building and operating costs.
It is also easier to develop applications for an allIP network as it handles voice and data on a
single allIP core network. Thus, Internet service providers will be less dependent on cellular
operators and will be able to develop applications more freely.
An allIP network removes the need for SS7 signaling that consumes a large part of the network
bandwidth even when there is no signaling traffic. IP networks use other less bandwidthexpensive mechanisms to achieve reliability.
372
Mobile terminal (MT) that consists of the mobile equipment and a module that store the home
identifier address (HIA) and a secret key for authentication; the HIA is used to identify the MT
uniquely in the network
Home identifiers address (HIA) and a secret key for authentication; the HIA is used to identify the
mobile terminal uniquely in the network; each MT will also have a binding cache that keeps track
of the location of each of its correspondent hosts
Access control center (ACC), which is a router connecting the access network to IP core
network; it has many access points to which a MT can attach; it also has binding cache that
stores the HIA, network care of address (NCoA) and the point care of address (PCoA) or each
MT visiting this access network; an NCoA is the address that identifies this access network
where the MT is currently in. A PCoA is the address that identifies the exact location of the
terminal in the access network
Home agent (HA) is a router, which connects the MTs whose home network is this particular
access network; it has a binding cache that stores the HIA of each MT connected to it, NCoA that
identifies the access networks where the MT is currently located, and PCoA
An authentication center (AC) is a database that stores information used for HIA security
authentication; it verifies a users identity for network access and authorizes requests for services
Md. Abdur Razzaque, Zia Ush-Shamszaman, Samiul Bashar, Jobeda Khatoon Sumi, and Safina Showkat Ara
Access Network
Voice
Manager
Access
Control
Home
Agent
Authentication
Center
IP Core Network
Transport Gateway
Access Network
Access
Point
Access
Control
Mobile
Terminal
PSTN
Gateway
PLMN
Gateway
Home
Agent
Authentication
Center
PSTN
PLMN
Transport gateway (TGW) that consists of two gateway functions, the public switched telephone
network (PSTN) gateway and public land mobile network (PLMN) gateway, which are used to
provide access for other communication networks. The former allows PSTN to connect to the IP
core network while the latter performs the same functionality for PLMN. This allows users who
are still using PSTN and PLMN to be able to access to the IP core network.
Voice manager (VM) provides voice over IP (VoIP) services that are built on and extended from
the mobile IP architecture. It sets up, controls all sessions and basic voice calls and manages their
states for each user. Within a service session, it allocates, modifies, and deallocates resources
accordingly in order to support the session.
Location Update
The location update process is divided into two levels, intramobility and intermobility.
Intramobility
When an MT moves from one access point to another within the same access network, it will be assigned
a new PCoA from the access point to which it is attached. Because the MT is only moving within the
same access network, the NCoA remains the same. After receiving the new PCoA, the MT will send two
updates. The first update is sent to each of the local corresponding hosts (CHs) that are in the same access
network. This is to update them of the new binding between the moving MTs HIA and its new PCoA.
The second update is sent to the ACC to update the ACC with the new binding between the moving
mobile terminals HIA and its new PCoA. In intramobility, all updates occur within the access network.
No updates are sent out to the other access networks through the IP core network (see Figure 4).
373
MT
ACC
Figure 4: Intramobility
Intermobility
When a MT moves from one access network to another access network, it will be assigned a new PCoA
and a new NCoA. After receiving the new CoAs, the MT will send three updates. The first update is sent
to each of the local CHs that are in the same access network. This is to update them of the new binding
between the moving MTs HIA, NCoA and its new PCoA. The second update is sent to the ACC to
update the ACC with the new binding between the moving MTs HIA and its new PCoA. Authentication
is carried out to validate the identity of the MT, and an acknowledgement is sent back to the MT if it is
accepted. The third update is sent to its home agent (HA) and each of the external CHs that are outside of
this access network. This is to update them of the new binding between the moving MTs HIA and its
new NCoA (see Figure 5).
MT
ACC
HA
Acknowledgement
New NCoA, HIA
Figure 5: Intermobility
Packet Delivery
Routing of packets involves routing to mobile terminal and from mobile terminal.
Routing to MT
For the routing of packets from local CHs to MT, which is within the same access network, the local CH
will look up its binding cache using the MTs HIA to obtain the PCoA. Packets are then delivered directly
to the MT if there is a corresponding entry in the cache. Otherwise, a binding update request will be sent
to the MTs home agent, which will in turn update the local CHs cache with the new binding. Packets are
374
Md. Abdur Razzaque, Zia Ush-Shamszaman, Samiul Bashar, Jobeda Khatoon Sumi, and Safina Showkat Ara
then directly delivered to the MT. In both cases, the MTs binding cache will be updated with the new
binding of the local CHs HIA, NCoA, and PCoA.
For the routing of packets from external CHs to MT, which are in different access networks, the external
CH will look up its binding cache using the MTs HIA and obtain the NCoA if there is a corresponding
entry in the cache. Otherwise, a binding update request will be sent to the MTs home agent, which will in
turn update the external CHs ache with the new binding. In both cases, the MTs binding cache will be
updated with the new binding of the external CHs HIA and NCoA. Packets will be delivered to the
access control center (ACC) identified by the NCoA. ACC will look up its binding cache and obtain the
MTs PCoA. It will then tunnel the packets to the PCoA of the MT.
Routing from MT
Packets sent by the MT are routed directly to their destination, without any need for tunneling, with the
ACC acting as a router for all the packets generated by the MT.
Analysis of Proposed Mobility-Management Scheme
Strengths of Proposed Scheme
Location management is divided into two levels, intra- and intermobility. It reduces the amount of signaling
traffic. The exact location of the MT is transparent to its external CHs it reduces the number of location
updates. It allows for scalability. Packets follow the shortest path from the CHs to the mobile terminal
except for the first time. This solves the triangle routing problem, and measures of security are good.
Weaknesses of Proposed Scheme
This refers to the lack of proper support for integrating voice services. The VM in the scheme utilizes
VoIP to handle transmission of voice over the IP core network. Using a third-party protocol may
introduce inefficiency and latency as compared to having a fully integrated voice transmission protocol. It
is necessary to provide QoS for VoIP.
Conclusion
In this paper, we have proposed a mobility-management scheme in an integrated network allIP. The
analysis have been carried out about the related research done in mobility management of integrated
systems, and found out that an allIP network is able to best meet the requirements of an integrated system.
Finally, the attention is to specify the strength and weakness of the proposed scheme. The main strength of
the proposed scheme is that location management is divided into two levels, intra- and intermobility. This is
a significant advantage of GPRS as it helps in reducing the amount of signaling traffic [2], thus take this into
consideration in the design of the proposed scheme. One problem of mobile IP is that it generates a lot of
signaling traffic when a mobile terminal moves. The proposal deals with this problem by making the exact
location of the MT transparent to its external CHs. Only the NCoA of the access network in which it is in is
known by other external CHs [3]. Hence when the MT moves within the same access network, only local
CHs are updated. Building on the previous points where there is a reduction in signaling traffic due to the
transparency of intramobility to the external CHs, the proposed proposal allows for scalability. This is
because more nodes can be added without increasing as much signaling traffic as compared to mobile IP.
Hence, the proposed network can be easily expanded to provide services to an increasing number of users
without creating a bottleneck. The route optimization in IPv6 solves the triangle routing problem [4].
Similarly, in the proposal, a mobile terminal will send binding updates to the relevant entities the proposed
mobility management scheme is built on top of mobile IP and future works may include extensions to this
scheme, with focus on the idea of VoIP. AllIP is an area of immense research interest all over the world,
and some researches have been done. There is a scope for research on allIP and mobility management.
Some of the possible areas for future research are given below:
375
UMTS concept: A mobility management scheme can be proposed for future full architecture of
UMTS model [5]
Handover: Simulation for soft and hard handover in the proposed scheme
Focusing on VoIP: Future works may include extension to the proposed scheme with focus on the
idea of VoIP; simulation for soft and hard handover in the proposed scheme
References
[1]
[2]
Yi-Bing Lin, Imrich Chlamtac, Wireless and Mobile Network Architecture, Wiley Computer
Publishing, 2001.
[3]
[4]
D. B. Johnson and C. E. Perkins, Route Optimization in Mobile-draft-ietf-mobileip-optim05.txt, November 1996 (work in progress).
[5]
C. Perkins, A. Myles, and D. Johnson, IMHP: A Mobile Host Protocol for the Internet, Comp.
Networks and ISDN Sys., Vol. 27, No. 3, December 1994, pp. 47991.
376
377
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
the impact from intermodulation products and AWGN. Nominal values for cell-to-cell interference,
orthogonality, and activity factors are also included in the analysis. [1]
Cellular mobile units (or handsets) must provide for the reception of an entire allocated telephony band
because the handset does not have a priori knowledge of the channels to which it may be assigned. The
allocated telephony bands are typically from 40 to 60 MHz. There will be multiple service providers, with
transmitting base stations, within the allocated spectrum; and these signals can cause intermodulation
products in the radio frequency (RF) front end of the handset receiver. In the analysis presented herein, a
single cell sector is analyzed in the presence of AWGN and intermodulation products, produced by other
service-provider base stations within the allocated telephony bands. This model incorporates average
impact from other cells but not the impact that the cell under analysis (own cell) will have on other cells
as the own cell base station transmitter increases the transmission power to overcome the interference
from intermodulation products (IMP) experienced in the handset. The impact to cell capacity from IMPs
can be significant, but there will be additional capacity degradation in the adjacent cells as they
compensate for the increased power transmission in the own cell.
In this analysis, the impact on the load factor is computed for a single user, as a function of tones
producing intermodulation products with power levels of the tones ranging from -43 dBm to -21 dBm. As
will be shown, in CDMA systems with two-way power control, when a single handset is subjected to high
power intermodulation products (within the specifications of CDMA2000), the single handset can
consume a very high percentage of the cell/sector available traffic capacity. This will have significant
impact on the traffic capacity of the cell or the user will experience a dropped call. The analysis presented
herein is a conservative assessment of the impact on the capacity because the cell to cell interference will
create an additional impact to the capacity of the cluster of cells as one base station transmitter increases
the transmitter power to overcome the interference seen by the impacted handset.
Practical CDMA systems typically operate with maximum load factors (LF) (described below) between .5
and .7. In this analysis, the load factor of a single user is computed and then used to determine the
percentage of the cell traffic capacity that is used by the single user based on a maximum LF of .5 and .7.
The load factor is the percentage of the theoretical maximum capacity of the cell that is being used.
As an example, when a wideband CDMA (WCDMA) service is a 64 kbps service and the handset
experiences high levels of IMP interference, the load factor of the cell/sector can be driven by only one
user for one 64 kbps service, thus reducing the cell/sector capacity to almost zero if the impacted user is
not dropped. In most cases, the impacted user will be dropped. In the analysis presented herein, the
impact to the cell/sector load factor is computed beyond where the user would most likely be dropped, but
from the curves presented that point can be selected.
Five cases are presents herein, one for CDMA2000 voice, two for third-generation (3G) WCDMA, one
voice, and one data. An additional two cases are considered when WCDMA is designed for IMPs from
tones no higher than -49 dBm wherein the design does not support a low noise amplifier (LNA) bypass
mode as in CDMA2000. There are an almost unlimited number of combinations of data and voice
services that may exist simultaneously in 3G WCDMA. For simplification, one data and one voice case
are presented (along with the no LNA bypass mode). While 3G WCDMA is not required by standards
specifications to handle two-tone tests as high as that of CDMA2000, WCDMA will, in reality, be
subjected to the same level of interference in the field, but it will be more sensitive to intermodulation
products due to the wider bandwidth of the signal. If WCDMA handsets are not designed to handle the
greater interference, the impact on capacity is even greater. Within the model, many parameters can be
varied such as the NF and third-order intermodulation intercept point (IIP3) of the handsets. This model
assumes one IMP inband of the signal of interest (SOI), but in reality, there can be multiple interfering
IMPs. Note that the Finesse Wireless technology can remove multiple IMPs simultaneously.
378
Francis J. Smith
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
where S1( dBm) is the power in the first source signal and S2( dBm) is the power in the second source
signal and IIP3 is the system IIP3 of the handset (the cascaded IIP3 of all components). If the source
signals produce a wideband IMP, then only that portion of the energy that falls inband of the signal of
interest will interfere with the SOI. For this paper, the source signals will be assumed to be tones to
simplify the concepts presented, but as can be shown, the inband energy is the same. The Finesse
Technology works for any bandwidth IMP, even if it is wider than the SOI.
The interference requirements for GSM and WCDMA are shown in Figure 2. The two-tone, third-order
IMP, tests for GSM and WCDMA only require the systems to be designed for tones up to -49 dBm and
do not account for the very close in blocker at -43 dBm. As can be seen in Figure 2, there are specified
blockers as high as -26 dBm specified within the 60 MHz PCS receive band for North America. In reality,
as shown in Figure 1, WCDMA will be exposed to the same high-power interfering signals as specified
for CDMA2000.
Blocking Profile for North America
PCS 1900
Out-of-Band
0 dBm
Inband
(ii)
(i)
Out-of-Band
(iii)
(iv)
0 dBm
-12 dBm
-12 dBm
-26 dBm
...
...
...
- 33 dBm
...
- 43 dBm
- 43 dBm
...
- 33 dBm
...
...
0 dBm
0 dBm
-26 dBm
...
...
...
-99dBm
f0
Out-of-Band
(i)
0 dBm
0 dBm
Inband
(ii)
(iii)
-12 dBm
-12 dBm
20 MHz
...
Out-of-Band
-26 dBm
...
- 33 dBm
...
- 43 dBm
- 43 dBm
...
- 33 dBm
...
...
(iv)
0 dBm
0 dBm
...
...
...
-99dBm
f0
Francis J. Smith
the tones are at -21 dBm, the inband third-order IMP will be at -83 dBm, or 20 + dB above the GSM
sensitivity requirement for GSM. At the specified level of -49 dBm, the IMP generates no significant
interference. Field measurements readily show blocking signals are common at -30 to -35 dBm. Signal
levels between -30 dBm and -21 dBm are seen but not as often. As the density of base stations increases,
the elevation angle of the antennas is lowered and the potential interference becomes greater.
3. IMP Interference Mitigation Technology, System Simulations
A Matrix Laboratory (MATLAB) system simulation was built for a CDMA2000 receiver with performance
specifications that just met the high-power two-tone test. Figure 3 shows the SOI and the IMP generated by
the system that meets the sensitivity requirement of -94 dBm. Figure 4 shows the improvement when the
IMP is cancelled by the Finesse Technology. In the simulations, the SOI is shown as a tone so that the
relative strength can be seen. In reality, it is below the noise floor and spread by 21 dB.
In the real world the source signals, which generate the IMPs, are not tones but are modulated signals. An
additional simulation was run with simulated FM AMPS signal with a 30 kHz bandwidth. Figure 5 shows
the SOI plus the IMP prior to IMP suppression. Figure 5 shows the SOI after IMP suppression. As shown
in Figure 6, there is a small residual of the modulated IMP. This can be eliminated when the differential
group delay is taken into account. Even at the present level, when the IMP is spread by the CDMA2000
spreading gain of 21 dB, it will have no impact on the signal to interference ratio. The simulations are
predicting 18 to 20 dB of suppression for the modulated and tone IMPs respectively. As will be shown
below, significant sensitivity and system capacity improvements can be achieved with IMP suppression in
the handset or MS.
381
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
Figure 5: SOI Plus AMPS IMP from HighPower Two-Tone Signal Test
Francis J. Smith
For CDMA2000, an average cell capacity is around 13 to 16 users per sector with 3 sectors. This provides
the cell with around 50 users depending on the environment with a sectorization gain of 2.6. As will be
shown in the analysis that follows, if one user experiences high-power IMPs and the base station
increases power to support that user, a single user can consume 30 to 50 percent and up to 100 percent of
the cell capacity (assuming the user is not dropped at some point). If multiple IMPs are present, or more
users are impacted, the impact to capacity can be severe.
The average load factor can be computed by the equation below. [1]
Where:
N = the number of users j per cell sector
j = 1 to N
Dependent on service
Dependent on multipath
1 = full orthogonality
0 = no orthogonality
Nominal = .5
383
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
The following nominal values, as shown in Table 1, were used for voice and data.
Variable
AFj
Eb/No
Block Error Rate BLER
Bit Rate of User Rj
Data
1.0
5 dB
10 percent
64 kbps
3.84 Mcps
1.23 Mcps
7 dB
0 dBm
8.5 dBm
Voice
.58
7 dB
1 percent
12.2 kbps WCDMA
9.6 kbps CDMA2000
3.84 Mcps
1.23 Mcps
7 dB
0 dBm
8.5 dBm
Francis J. Smith
requires higher transmitter power from the base station and thus the system traffic capacity is impacted in
the own cell as well as the adjacent cells. If the LNA is not bypassed, the IMPs get even larger and have
a greater impact on the loading factor and the capacity of the cell.
The system IIP3 of a handset, as far as the two-tone test is impacted, is the cascaded IIP3 of all the analog
components in the receiver. For CDMA2000, the LNA requires a minimum IIP3 of 7.6 dBm to meet the
signal blocking tone (at -30 dBm) desense test, which results in the third-order cross modulation products
from the transmitter feed thru. The transmitter feed thru is filtered in the surface acoustical wave (SAW)
filter following the LNA and is not impacted by the down stream IIP3 of the amplifiers, mixers, baseband
filters and amplifiers. The system IIP3 as seen by the two-tone test will probably be about 10 dB lower
than the LNA alone. The analysis herein assumes a maximum LNA IIP3 of +10 dBm and a system IIP3
of 0 dBm. In the LNA bypass mode, a system IIP3 of +8.5 dBm is assumed given that the bypass mode
switch will have an IIP3 of around +15 to +20 dBm.
For this analysis, these same numbers for the CDMA2000 two-tone tests are used for WCDMA even
though the standards do not call for the same high-level two-tone test. This is done because in reality,
both systems will experience the same level of interference.
6.1 Case No. 1: CDMA2000 Voice Case (9.6 kbps)
For the CDMA 2000 voice case, the parameter values used are those shown in Table 1. The system IIP3
is assumed to be 0 dBm with the LNA having an IIP3 of +10 dBm (possibly optimistic). The thermal
noise floor is computed as -174 dBm/Hz x 1.23 MHz = -174 dBm/Hz + 61 dB-Hz = -113 dBm. With a
NF of 7 dB, the system noise floor is -106 dBm. With a required Eb/No of 7 dB and a coding gain of
(1.23E6/9.6E3) 21 dB, the SOI sensitivity = -106 -21 + 7 = -120 dBm with the LNA in and no interfering
signals (standards specification is -119.6 dBm). In the Table 2 below, the highlighted band in row 15 is
the point at which the LNA bypass mode is used because the IMP energy now is high enough to impact
the noise floor of the LNA, and the handset receiver sensitivity will be better in the bypass mode.
D
G
1
2 PowerM Thermal
3 Tone N Noise
MS
4
dBm
dBm
5
-43
-106.1
6
-42
-106.1
7
-41
-106.1
8
-40
-106.1
9
-39
-106.1
10
-38
-106.1
11
-37
-106.1
12
-36
-106.1
13
-35
-106.1
14
-34
-106.1
15
-32
-106.1
16
-31
-106.1
17
-30
-106.1
18
-29
-106.1
19
-28
-106.1
20
-27
-106.1
21
-26
-106.1
22
-25
-106.1
23
-24
-106.1
24
-23
-106.1
25
-22
-106.1
26
-21
-106.1
27
I
M
LNA in System
MS
Ioc
Spreading
IM3
Equals
Gain
IMP
Thermal
dB
dBm floor dBm
-129
-106.1
21.1
-126
-106.1
21.1
-123
-106.1
21.1
-120
-106.1
21.1
-117
-106.1
21.1
-114
-106.1
21.1
-111
-106.1
21.1
-108
-106.1
21.1
-105
-106.1
21.1
-102
-106.1
21.1
-96
-106.1
21.1
-93
-106.1
21.1
-90
-106.1
21.1
-87
-106.1
21.1
-84
-106.1
21.1
-81
-106.1
21.1
-78
-106.1
21.1
-75
-106.1
21.1
-72
-106.1
21.1
-69
-106.1
21.1
-66
-106.1
21.1
-63
-106.1
21.1
Target
Target Thermal
Signal
Signal
Noise
Based
Based
Bypass
on Ioc
on IM3
dBm
-120.2
-120.2
-101.1
-120.2
-120.1
-101.1
-120.2
-120.1
-101.1
-120.2
-120.0
-101.1
-120.2
-119.8
-101.1
-120.2
-119.5
-101.1
-120.2
-119.0
-101.1
-120.2
-118.0
-101.1
-120.2
-116.6
-101.1
-120.2
-114.6
-101.1
-120.2
-109.7
-101.1
-120.2
-106.9
-101.1
-120.2
-104.0
-101.1
-120.2
-101.0
-101.1
-120.2
-98.0
-101.1
-120.2
-95.1
-101.1
-120.2
-92.1
-101.1
-120.2
-89.1
-101.1
-120.2
-86.1
-101.1
-120.2
-83.1
-101.1
-120.2
-80.1
-101.1
-120.2
-77.1
-101.1
U
V
W
LNA By Pass Mode
Bypass
Ioc
Target
IM3
Equals Signal
IMP Thermal Based
on Ioc
-146 -101.1
-115.2
-143 -101.1
-115.2
-140 -101.1
-115.2
-137 -101.1
-115.2
-134 -101.1
-115.2
-131 -101.1
-115.2
-128 -101.1
-115.2
-125 -101.1
-115.2
-122 -101.1
-115.2
-119 -101.1
-115.2
-113 -101.1
-115.2
-110 -101.1
-115.2
-107 -101.1
-115.2
-104 -101.1
-115.2
-101 -101.1
-115.2
-98 -101.1
-115.2
-95 -101.1
-115.2
-92 -101.1
-115.2
-89 -101.1
-115.2
-86 -101.1
-115.2
-83 -101.1
-115.2
-80 -101.1
-115.2
X
Target
Signal
Based
on IM3
-115.2
-115.2
-115.2
-115.2
-115.2
-115.2
-115.2
-115.2
-115.1
-115.1
-114.9
-114.7
-114.2
-113.4
-112.1
-110.3
-108.1
-105.6
-102.8
-99.9
-97.0
-94.0
AB
Target
with
IMP
interfer
-120.2
-120.1
-120.1
-120.0
-119.8
-119.5
-119.0
-118.0
-116.6
-115.1
-114.9
-114.7
-114.2
-113.4
-112.1
-110.3
-108.1
-105.6
-102.8
-99.9
-97.0
-94.0
AD
AH
AJ
Optimal Selection w or w/o LNA
Increase
in Eb/No
Effective
Load Factor
With
req Eb/No per jammed
User
interfer
0.0
7.0
0.03
0.0
7.0
0.03
0.1
7.1
0.03
0.2
7.2
0.03
0.3
7.3
0.03
0.7
7.7
0.03
1.2
8.2
0.03
2.2
9.2
0.04
3.6
10.6
0.06
5.1
12.1
0.08
5.3
12.3
0.09
5.5
12.5
0.09
6.0
13.0
0.10
6.8
13.8
0.12
8.1
15.1
0.17
9.8
16.8
0.25
12.1
19.1
0.42
14.6
21.6
0.75
17.4
24.4
1.42
20.2
27.2
2.75
23.2
30.2
5.41
26.1
33.1
10.71
Table 2: CDMA2000 Voice-Case IMP Impact on Load Factor with LNA Bypass
In the bypass mode, the noise figure is assumed to be +12 dB and now the system noise floor without the
interfering IMP is -113 + 12 dB = -101 dBm. The SOI sensitivity = -101 + 7 -21 = -115 dBm without
interfering IMPs. The system IIP3 in the bypass mode is assumed to be 8.5 dBm. Table 2 gives the impact
of the IMPs on the load factor for a single user.
385
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
In column AJ, Table 2, the load factors for a single user that exceed .05 are highlighted in blue and those
that exceed .1 are highlighted in red. These numbers indicate the percentage of the pole capacity of the
system that will be consumed by one user with no other traffic on the channel. This is the percentage of
the theoretical maximum capacity.
In Table 3, the percentage of the available system capacity consumed by one jammed user, with a system
maximum load factor of .5 and .7 is presented. The cases wherein the single user consumes in excess of 5
percent of the total available system capacity are highlighted in blue. The cases where the signal user
consumes in excess of 50 percent of the available system capacity are highlighted in red. As the
percentage of the available system capacity consumed by one user approaches 100 percent, this becomes
a single-user system.
Table 3: CDMA2000 Voice Case: Percentage of Available System Capacity Consumed by a Single
Jammed User with Maximum System Load Factors of .5 and .7 as a Function of the Power in the Source
Tones that Generate the Third-Order IMP
Figure 7 shows the plots of the percentage of the system capacity that one CDMA2000 voice user will
consume as a function of the power in the signals (two tone) that generate IMPs and maximum load
factors of .5 and .7. Load factors for most practical CDMA systems fall between .5 and .7. Figure 8
shows the percentage of system capacity consumed with a maximum load factor of .5 and interference
suppression between 0 and 20 dB. Figure 9 shows the percentage of system capacity consumed with a
maximum load factor of .7 and interference suppression between 0 and 20 dB.
386
Francis J. Smith
-2
6
-2
8
-3
0
-3
2
-3
5
-3
7
-3
9
-4
1
-4
3
0%
Figure 7: CDMA 2000 Voice: Single-User Consumption of Available System Capacity for Maximum
Load Factors of .5 and .7 with No Interference Suppression
0 dB
10 dB
60%
15 dB
20 dB
40%
20%
0%
3
-4
1
-4
9
-3
5
-3
7
-3
2
-3
0
-3
8
-2
6
-2
4
-2
2
-2
Figure 8: CDMA2000 Voice: Percentage of System Capacity Consumed with a Maximum Load Factor of
.5 and Interference Suppression between 0 and 20 dB with LNA Bypass Capability
0 dB
10 dB
15 dB
-2
1
-2
4
-2
7
-3
0
-3
4
-3
7
20 dB
-4
0
-4
3
120%
100%
80%
60%
40%
20%
0%
Figure 9: CDMA2000 Voice: Percentage of System Capacity Consumed with a Maximum Load Factor of
.7 and Interference Suppression between 0 and 20 dB with LNA Bypass Capability
387
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
For Table 3 and Figures 7 through 11, the maximum available system capacity is the maximum load
factor times the pole capacity. When one user requires more power from the base station to overpower
inband IMPs, the load factor of the sector is increased and this increases the percentage of the system
capacity consumed by the one service. The available system capacity will be 1- percentage of capacity
consumed by the one user in question. Figures 10 and 11 show the available system capacity when one
user experiences interference from IMPs generated by signals from -43 to -21 dBm. The curves show the
ability to recover system capacity as a function of the level in interference suppression from 0 to 20 dB.
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
0 dB
10 dB
15 dB
-2
2
-2
6
-2
4
-3
0
-2
8
-3
5
-3
2
20 dB
-3
9
-3
7
-4
3
-4
1
% of System Capacity
Available
Figure 10: CDMA2000 Voice: Percentage of System Capacity Available for Two-Tone Power from -43
dBm to -21 dBm for a Maximum Load Factor of .5 and Interference Suppression from 0 to 20 dBm
120%
100%
0 dB
10 dB
15 dB
20 dB
80%
60%
40%
20%
-2
2
-2
4
-2
6
-3
0
-2
8
-3
5
-3
2
-3
7
-3
9
0%
-4
3
-4
1
% of System Capacity
Available
Figure 11: CDMA2000 Voice: Percentage of System Capacity Available for Two-Tone Power from -43
dBm to -21 dBm for a Maximum Load Factor of .7 and Interference Suppression from 0 to 20 dBm
Each of the four following cases address WCDMA for voice and data services with and without the LNA
bypass capability required in CDMA2000 due to the high blocking tones of -21 dBm. The results
presented were developed as shown above for CDMA2000. For efficiency, the cases for a maximum load
factor of .7 will be presented because this is the most optimistic.
6.2 Case No. 2: WCDMA Voice Case with LNA Bypass Capability (12.2 kbps)
For the WDMA voice case, the parameter values used are those shown in Table 4. The thermal noise
floor is computed as -174 dBm/Hz x 3.84 MHz = -174 dBm/Hz + 65.8 dB/Hz = -108 dBm. With a NF of
7 dB, the system noise floor is -101 dBm. With a required Eb/No of 7 dB and a coding gain of
(3.84E6/12.2E3) 25 dB, the SOI sensitivity = -101 -25 + 7 = -119 dBm with the LNA in and no
interfering signals. In the Table 4 below, the highlighted band in row 44 is the point at which the LNA
388
Francis J. Smith
bypass mode is used because the IMP energy now is high enough to impact the noise floor of the LNA,
and the handset receiver sensitivity will be better in the bypass mode.
D
G
29
30 PowerM Thermal
31 Tone N Noise
MS
32
dBm
33 dBm
-43
-101.2
34
-42
-101.2
35
-41
-101.2
36
-40
-101.2
37
-39
-101.2
38
-38
-101.2
39
-37
-101.2
40
-36
-101.2
41
-35
-101.2
42
-34
-101.2
43
-32
-101.2
44
-31
-101.2
45
-30
-101.2
46
-29
-101.2
47
-28
-101.2
48
-27
-101.2
49
-26
-101.2
50
-25
-101.2
51
-24
-101.2
52
-23
-101.2
53
-22
-101.2
54
-21
-101.2
55
I
M
LNA in System
MS
Ioc
Spreading
IM3
Equals
Gain
IMP
Thermal
dB
dBm floor dBm
-129
-101.2
25.0
-126
-101.2
25.0
-123
-101.2
25.0
-120
-101.2
25.0
-117
-101.2
25.0
-114
-101.2
25.0
-111
-101.2
25.0
-108
-101.2
25.0
-105
-101.2
25.0
-102
-101.2
25.0
-96
-101.2
25.0
-93
-101.2
25.0
-90
-101.2
25.0
-87
-101.2
25.0
-84
-101.2
25.0
-81
-101.2
25.0
-78
-101.2
25.0
-75
-101.2
25.0
-72
-101.2
25.0
-69
-101.2
25.0
-66
-101.2
25.0
-63
-101.2
25.0
Target
Target Thermal
Signal
Signal
Noise
Based
Based
Bypass
on Ioc
on IM3
dBm
-119.1
-119.1
-96.2
-119.1
-119.1
-96.2
-119.1
-119.1
-96.2
-119.1
-119.1
-96.2
-119.1
-119.0
-96.2
-119.1
-118.9
-96.2
-119.1
-118.7
-96.2
-119.1
-118.3
-96.2
-119.1
-117.6
-96.2
-119.1
-116.5
-96.2
-119.1
-112.8
-96.2
-119.1
-110.4
-96.2
-119.1
-107.7
-96.2
-119.1
-104.8
-96.2
-119.1
-101.9
-96.2
-119.1
-98.9
-96.2
-119.1
-96.0
-96.2
-119.1
-93.0
-96.2
-119.1
-90.0
-96.2
-119.1
-87.0
-96.2
-119.1
-84.0
-96.2
-119.1
-81.0
-96.2
U
V
W
LNA By Pass Mode
Bypass
Ioc
Target
IM3
Equals Signal
IMP Thermal Based
Noise
on Ioc
-146
-96.2
-114.1
-143
-96.2
-114.1
-140
-96.2
-114.1
-137
-96.2
-114.1
-134
-96.2
-114.1
-131
-96.2
-114.1
-128
-96.2
-114.1
-125
-96.2
-114.1
-122
-96.2
-114.1
-119
-96.2
-114.1
-113
-96.2
-114.1
-110
-96.2
-114.1
-107
-96.2
-114.1
-104
-96.2
-114.1
-101
-96.2
-114.1
-98
-96.2
-114.1
-95
-96.2
-114.1
-92
-96.2
-114.1
-89
-96.2
-114.1
-86
-96.2
-114.1
-83
-96.2
-114.1
-80
-96.2
-114.1
X
Target
Signal
Based
on IM3
-114.1
-114.1
-114.1
-114.1
-114.1
-114.1
-114.1
-114.1
-114.1
-114.1
-114.0
-114.0
-113.8
-113.5
-112.9
-112.0
-110.5
-108.6
-106.2
-103.6
-100.8
-97.9
AB
Target
with
IMP
interfer
-119.1
-119.1
-119.1
-119.1
-119.0
-118.9
-118.7
-118.3
-117.6
-116.5
-114.0
-114.0
-113.8
-113.5
-112.9
-112.0
-110.5
-108.6
-106.2
-103.6
-100.8
-97.9
AD
AH
AJ
Optimal Selection w or w/o LNA
Increase
in Eb/No
Effective
Load Factor
With
req Eb/No per jammed
User
interfer
0.0
7.0
0.01
0.0
7.0
0.01
0.0
7.0
0.01
0.1
7.1
0.01
0.1
7.1
0.01
0.2
7.2
0.01
0.4
7.4
0.01
0.8
7.8
0.01
1.5
8.5
0.02
2.6
9.6
0.02
5.1
12.1
0.03
5.2
12.2
0.03
5.3
12.3
0.04
5.7
12.7
0.04
6.2
13.2
0.04
7.2
14.2
0.06
8.6
15.6
0.08
10.6
17.6
0.12
12.9
19.9
0.21
15.6
22.6
0.38
18.4
25.4
0.73
21.3
28.3
1.42
-2
4
-2
6
-2
8
-3
0
-3
2
-3
5
-3
7
-3
9
-4
1
-4
3
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Figure 12: WCDMA Voice Case: Percentage of Available System Capacity Consumed by a Single
Jammed User with Maximum System Load Factors of .5 and .7 as a Function of the Power in the Source
Tones that Generate the Third-Order IMP
389
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
% of System Capacity
Available
120%
100%
0 dB
10 dB
15 dB
20 dB
80%
60%
40%
20%
-2
4
-2
2
-2
8
-2
6
-3
2
-3
0
-3
7
-3
5
-4
1
-3
9
-4
3
0%
Figure 13: WCDMA Voice: Percentage of System Capacity Available for Two-Tone Power
from -43 dBm to -21 dBm for a Maximum Load Factor of .7 and Interference Suppression
from 0 to 20 dBm with LNA Bypass Capability
6.3 Case No. 3: WCDMA Voice Case without the LNA Bypass Mode (12.2 kbps)
WCDMA and GSM are specified with the two-tone test to be conducted with the source tones at -49
dBm. This requirement does not require an LNA bypass mode. If WCDMA handsets are designed to
tolerate only the -49 dBm tones, then there will not be an LNA bypass mode and when the power in the
IMP exceeds the noise floor of the LNA, the impact to the capacity of the system will be even greater as
shown in Figures 12 and 13. The computational results for this case are shown in Table 5.
D
G
57
58 PowerM Thermal
59 Tone N Noise
60
MS
61 dBm
dBm
62
-43
-101.2
63
-42
-101.2
64
-41
-101.2
65
-40
-101.2
66
-39
-101.2
67
-38
-101.2
68
-37
-101.2
69
-36
-101.2
70
-35
-101.2
71
-34
-101.2
72
-32
-101.2
73
-31
-101.2
74
-30
-101.2
75
-29
-101.2
76
-28
-101.2
77
-27
-101.2
78
-26
-101.2
79
-25
-101.2
80
-24
-101.2
81
-23
-101.2
82
-22
-101.2
83
-21
-101.2
H
I
M
LNA in System
MS
Ioc
Spreading
IM3
Equals
Gain
IMP
Thermal
dB
dBm floor dBm
-129
-101.2
25.0
-126
-101.2
25.0
-123
-101.2
25.0
-120
-101.2
25.0
-117
-101.2
25.0
-114
-101.2
25.0
-111
-101.2
25.0
-108
-101.2
25.0
-105
-101.2
25.0
-102
-101.2
25.0
-96
-101.2
25.0
-93
-101.2
25.0
-90
-101.2
25.0
-87
-101.2
25.0
-84
-101.2
25.0
-81
-101.2
25.0
-78
-101.2
25.0
-75
-101.2
25.0
-72
-101.2
25.0
-69
-101.2
25.0
-66
-101.2
25.0
-63
-101.2
25.0
Target
Target
Signal
Signal
Based
Based
on Ioc
on IM3
-119.1
-119.1
-119.1
-119.1
-119.1
-119.1
-119.1
-119.1
-119.1
-119.0
-119.1
-118.9
-119.1
-118.7
-119.1
-118.3
-119.1
-117.6
-119.1
-116.5
-119.1
-112.8
-119.1
-110.4
-119.1
-107.7
-119.1
-104.8
-119.1
-101.9
-119.1
-98.9
-119.1
-96.0
-119.1
-93.0
-119.1
-90.0
-119.1
-87.0
-119.1
-84.0
-119.1
-81.0
U
V
W
LNA By Pass Mode
No By LNA By Pass
AB
AD
AH
AJ
Optimal Selection w or w/o LNA
Target
Increase
with
in Eb/No
Effective
Load Factor
IMP
With
req Eb/No per jammed
interfer interfer
User
-119.1
0.0
7.0
0.01
-119.1
0.0
7.0
0.01
-119.1
0.0
7.0
0.01
-119.1
0.1
7.1
0.01
-119.0
0.1
7.1
0.01
-118.9
0.2
7.2
0.01
-118.7
0.4
7.4
0.01
-118.3
0.8
7.8
0.01
-117.6
1.5
8.5
0.02
-116.5
2.6
9.6
0.02
-112.8
6.3
13.3
0.05
-110.4
8.8
15.8
0.08
-107.7
11.5
18.5
0.15
-104.8
14.3
21.3
0.29
-101.9
17.2
24.2
0.56
-98.9
20.2
27.2
1.11
-96.0
23.2
30.2
2.21
-93.0
26.2
33.2
4.39
-90.0
29.2
36.2
8.76
-87.0
32.2
39.2
17.46
-84.0
35.2
42.2
34.83
-81.0
38.2
45.2
69.48
Table 5: WCDMA Voice Case: IMP Impact on Load Factor without LNA Bypass Mode
In column AJ, Table 5, the load factors for a single user that exceed .05 are highlighted in blue, and those
that exceed .1 are highlighted in red. These numbers indicate the percentage of the pole capacity of the
system that will be consumed by one user with no other traffic in the channel. This is the percentage of
the theoretical maximum capacity.
In Figure 14, the percentage of the available system capacity consumed by one jammed user, with a
system maximum load factor or .5 and .7, is presented. As the percentage of the available system capacity
390
Francis J. Smith
consumed by one user approaches 100 percent, this becomes a single-user system at 12.2 kbps. Figure 15
shows the percentage of the system capacity that is available as function of the interference levels and the
interference suppression.
WCDMA Voice Single User Percentage of Available
System Capacity Without LNA Bypass
Solid Line Maximum Load Factor = .5
Dotted Line Maximum Load Factor = .7
120%
100%
80%
60%
40%
20%
0%
-43 -42 -41 -40 -39 -38 -37 -36 -35 -34 -32 -31 -30 -29 -28
Two Tone Power in dBm
Figure 14: WCDMA Voice Case Without LNA Bypass Mode: Percentage of Available System Capacity
Consumed by a Single Jammed User with Maximum System Load Factors of .5 and .7 as a Function of
the Power in the Source Tones that Generate the Third-Order IMP
120%
100%
0 dB
10 dB
15 dB
20 dB
80%
60%
40%
20%
-2
2
-2
4
-2
8
-2
6
-3
2
-3
0
-3
7
-3
5
-3
9
0%
-4
3
-4
1
% of System Capacity
Available
Figure 15: WCDMA Voice: Percentage of System Capacity Available for Two-Tone Power from -43 dBm
to -21 dBm for a Maximum Load Factor of .7 and Interference Suppression from 0 to 20 dBm without
LNA Bypass Capability
6.4 Case No. 4: WCDMA Data Case (64 kbps)
For the WDMA data case, the parameter values used are those shown in Table 1. The thermal noise floor
is computed as -174 dBm/Hz x 3.84 MHz = -174 dBm/Hz + 65.8 dB-Hz = -108 dBm. With a NF of 7 dB,
the system noise floor is -101 dBm. With a required Eb/No of 5 dB and a coding gain of (3.84E6/64.E3)
18 dB, the SOI sensitivity = -101 -18 + 5 = -114 dBm with the LNA in and no interfering signals. In
Table 6 below, the highlighted band in row 100 is the point at which the LNA bypass mode is used
because the IMP energy now is high enough to impact the noise floor of the LNA, and the handset
receiver sensitivity will be better in the bypass mode.
391
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
D
G
85
86 PowerM Thermal
87 Tone N Noise
88
MS
89 dBm
dBm
90
-43
-101.2
91
-42
-101.2
92
-41
-101.2
93
-40
-101.2
94
-39
-101.2
95
-38
-101.2
96
-37
-101.2
97
-36
-101.2
98
-35
-101.2
99
-34
-101.2
100
-32
-101.2
101
-31
-101.2
102
-30
-101.2
103
-29
-101.2
104
-28
-101.2
105
-27
-101.2
106
-26
-101.2
107
-25
-101.2
108
-24
-101.2
109
-23
-101.2
110
-22
-101.2
111
-21
-101.2
H
I
M
LNA in System
MS
Ioc
Spreading
IM3
Equals
Gain
IMP
Thermal
dB
dBm floor dBm
-129
-101.2
17.8
-126
-101.2
17.8
-123
-101.2
17.8
-120
-101.2
17.8
-117
-101.2
17.8
-114
-101.2
17.8
-111
-101.2
17.8
-108
-101.2
17.8
-105
-101.2
17.8
-102
-101.2
17.8
-96
-101.2
17.8
-93
-101.2
17.8
-90
-101.2
17.8
-87
-101.2
17.8
-84
-101.2
17.8
-81
-101.2
17.8
-78
-101.2
17.8
-75
-101.2
17.8
-72
-101.2
17.8
-69
-101.2
17.8
-66
-101.2
17.8
-63
-101.2
17.8
N
O
T
U
V
W
LNA By Pass Mode
Target
Target Thermal Bypass
Ioc
Target
Signal
Signal
Noise
IM3
Equals Signal
Based
Based
Bypass
IMP Thermal Based
on Ioc
on IM3
dBm
noise
on Ioc
-113.9
-113.9
-96.2
-146
-96.2
-108.9
-113.9
-113.9
-96.2
-143
-96.2
-108.9
-113.9
-113.9
-96.2
-140
-96.2
-108.9
-113.9
-113.9
-96.2
-137
-96.2
-108.9
-113.9
-113.8
-96.2
-134
-96.2
-108.9
-113.9
-113.7
-96.2
-131
-96.2
-108.9
-113.9
-113.5
-96.2
-128
-96.2
-108.9
-113.9
-113.1
-96.2
-125
-96.2
-108.9
-113.9
-112.4
-96.2
-122
-96.2
-108.9
-113.9
-111.3
-96.2
-119
-96.2
-108.9
-113.9
-107.6
-96.2
-113
-96.2
-108.9
-113.9
-105.2
-96.2
-110
-96.2
-108.9
-113.9
-102.5
-96.2
-107
-96.2
-108.9
-113.9
-99.6
-96.2
-104
-96.2
-108.9
-113.9
-96.7
-96.2
-101
-96.2
-108.9
-113.9
-93.7
-96.2
-98
-96.2
-108.9
-113.9
-90.8
-96.2
-95
-96.2
-108.9
-113.9
-87.8
-96.2
-92
-96.2
-108.9
-113.9
-84.8
-96.2
-89
-96.2
-108.9
-113.9
-81.8
-96.2
-86
-96.2
-108.9
-113.9
-78.8
-96.2
-83
-96.2
-108.9
-113.9
-75.8
-96.2
-80
-96.2
-108.9
X
Target
Signal
Based
on IM3
-108.9
-108.9
-108.9
-108.9
-108.9
-108.9
-108.9
-108.9
-108.9
-108.9
-108.8
-108.8
-108.6
-108.3
-107.7
-106.8
-105.3
-103.4
-101.0
-98.4
-95.6
-92.7
AB
AD
AH
AJ
Optimal Selection w or w/o LNA
Target
Increase
with
in Eb/No
Effective
Load Factor
IMP
With
req Eb/No per jammed
User
interfer interfer
-113.9
0.0
5.0
0.04
-113.9
0.0
5.0
0.04
-113.9
0.0
5.0
0.04
-113.9
0.1
5.1
0.04
-113.8
0.1
5.1
0.04
-113.7
0.2
5.2
0.04
-113.5
0.4
5.4
0.04
-113.1
0.8
5.8
0.04
-112.4
1.5
6.5
0.05
-111.3
2.6
7.6
0.06
-108.8
5.1
10.1
0.11
-108.8
5.2
10.2
0.12
-108.6
5.3
10.3
0.12
-108.3
5.7
10.7
0.13
-107.7
6.2
11.2
0.15
-106.8
7.2
12.2
0.18
-105.3
8.6
13.6
0.26
-103.4
10.6
15.6
0.40
-101.0
12.9
17.9
0.69
-98.4
15.6
20.6
1.26
-95.6
18.4
23.4
2.41
-92.7
21.3
26.3
4.70
Table 6: WCDMA Data Case IMP Impact on Load Factor with LNA Bypass Capability
In bypass mode, the noise figure is assumed to be +12 dB and now the system noise floor without the
interfering IMP is -108 + 12 dB = -96 dBm. The SOI sensitivity = -96 + 5 -18 = -109 dBm without
interfering IMPs.
In column AJ Table 6, the load factors for a single user that exceed .05 are highlighted in blue, and those
that exceed .1 are highlighted in red. These numbers indicate the percentage of the pole capacity of the
system that will be consumed by one user with no other traffic in the channel. This is the percentage of
the theoretical maximum capacity.
In Figure 16, the percentage of the available system capacity consumed by one jammed user, with a
system maximum load factor of .5 and .7, is presented. As the percentage of the available system capacity
consumed by one user approaches 100 percent, this becomes a single-user system at 64 kbps. Figure 17
shows the percentage of the system capacity that is available as a function of the interference levels and
the interference suppression.
WCDMA Data Single User Percentage of Available
System Capacity
Solid Line Maximum Load Factor = .5
Dotted Line Maximum Load Factor = .7
120%
100%
80%
60%
40%
20%
-2
4
-2
6
-2
8
-3
0
-3
2
-3
5
-3
7
-3
9
-4
1
-4
3
0%
Figure 16: WCDMA Data Case: Percentage of Available System Capacity Consumed by a Single
Jammed User with Maximum System Load Factors of .5 and .7 as a Function of the Power in the Source
Tones that Generate the Third-Order IMP with LNA Bypass Capability
392
Francis J. Smith
40%
20%
0%
-2
2
-2
4
-2
6
-2
8
-3
0
-3
2
-3
5
-3
7
-3
9
-4
1
20 dB
-4
3
% of System
Capacity
Available
100%
80%
60%
Figure 17: WCDMA Data: Percentage of System Capacity Available for Two Tone
Power from -43 dBm to -21 dBm for a Maximum Load Factor of .7 and Interference
Suppression from 0 to 20 dBm with LNA Bypass Capability
6.5 Case No. 5: WCDMA Data Case without the LNA Bypass Mode (64 kbps)
WCDMA and GSM are specified with the two-tone test to be conducted with the source tones at -49
dBm. This requirement does not require an LNA bypass mode. If WCDMA handsets are designed to only
tolerate the -49 dBm tones, then there will not be an LNA bypass mode and when the power in the IMP
exceeds the noise floor of the LNA, the impact to the capacity of the system will be even greater as shown
in Figures 16 and 17.
In column AJ in Table 7, the load factors for a single user that exceed .05 are highlighted in blue, and
those that exceed .1 are highlighted in red. These numbers indicate the percentage of the pole capacity of
the system that will be consumed by one user with no other traffic in the channel. This is the percentage
of the theoretical maximum capacity.
D
G
113
114 PowerM Thermal
115 Tone N Noise
MS
116
dBm
117 dBm
-43
-101.2
118
-42
-101.2
119
-41
-101.2
120
-40
-101.2
121
-39
-101.2
122
-38
-101.2
123
-37
-101.2
124
-36
-101.2
125
-35
-101.2
126
-34
-101.2
127
-32
-101.2
128
-31
-101.2
129
-30
-101.2
130
-29
-101.2
131
-28
-101.2
132
-27
-101.2
133
-26
-101.2
134
-25
-101.2
135
-24
-101.2
136
-23
-101.2
137
-22
-101.2
138
-21
-101.2
139
I
M
LNA in System
MS
Ioc
Spreading
IM3
Equals
Gain
IMP
Thermal
dB
dBm floor dBm
-129
-101.2
17.8
-126
-101.2
17.8
-123
-101.2
17.8
-120
-101.2
17.8
-117
-101.2
17.8
-114
-101.2
17.8
-111
-101.2
17.8
-108
-101.2
17.8
-105
-101.2
17.8
-102
-101.2
17.8
-96
-101.2
17.8
-93
-101.2
17.8
-90
-101.2
17.8
-87
-101.2
17.8
-84
-101.2
17.8
-81
-101.2
17.8
-78
-101.2
17.8
-75
-101.2
17.8
-72
-101.2
17.8
-69
-101.2
17.8
-66
-101.2
17.8
-63
-101.2
17.8
Target
Target
Signal
Signal
Based
Based
on Ioc
on IM3
-113.9
-113.9
-113.9
-113.9
-113.9
-113.9
-113.9
-113.9
-113.9
-113.8
-113.9
-113.7
-113.9
-113.5
-113.9
-113.1
-113.9
-112.4
-113.9
-111.3
-113.9
-107.6
-113.9
-105.2
-113.9
-102.5
-113.9
-99.6
-113.9
-96.7
-113.9
-93.7
-113.9
-90.8
-113.9
-87.8
-113.9
-84.8
-113.9
-81.8
-113.9
-78.8
-113.9
-75.8
U
V
W
LNA By Pass Mode
No By LNA By Pass
AB
Target
with
IMP
interfer
-113.9
-113.9
-113.9
-113.9
-113.8
-113.7
-113.5
-113.1
-112.4
-111.3
-107.6
-105.2
-102.5
-99.6
-96.7
-93.7
-90.8
-87.8
-84.8
-81.8
-78.8
-75.8
AD
AH
AJ
Optimal Selection w or w/o LNA
Increase
in Eb/No
Effective
Load Factor
With
req Eb/No per jammed
User
interfer
0.0
5.0
0.04
0.0
5.0
0.04
0.0
5.0
0.04
0.1
5.1
0.04
0.1
5.1
0.04
0.2
5.2
0.04
0.4
5.4
0.04
0.8
5.8
0.04
1.5
6.5
0.05
2.6
7.6
0.06
6.3
11.3
0.15
8.8
13.8
0.27
11.5
16.5
0.49
14.3
19.3
0.95
17.2
22.2
1.86
20.2
25.2
3.68
23.2
28.2
7.31
26.2
31.2
14.54
29.2
34.2
28.98
32.2
37.2
57.80
35.2
40.2
115.29
38.2
43.2
229.99
Table 7: WCDMA Data Case: IMP Impact on Load Factor Without LNA Bypass Mode
In Figure 18, the percentage of the available system capacity consumed by one jammed user, with a
system maximum load factor or .5 and .7, is presented. As the percentage of the available system capacity
consumed by one user approaches 100 percent, this becomes a single-user system at 64 kbps. Figure 19
shows the percentage of the system capacity that is available as a function of the interference levels and
the interference suppression.
393
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
WCDMA Data Single User Percentage of Available
System Capacity Without LNA Bypass
Solid Line Maximum Load Factor = .5
Dotted Line Maximum Load Factor = .7
120%
100%
80%
60%
40%
20%
0%
-43 -42 -41 -40 -39 -38 -37 -36 -35 -34 -32 -31 -30
Two Tone Power in dBm
Figure 18: WCDMA Data Case Without LNA Bypass Mode: Percentage of Available System Capacity
Consumed by a Single Jammed User with Maximum System Load Factors of .5 and .7 as a Function of
the Power in the Source Tones that Generate the Third-Order IMP
100%
80%
60%
40%
20%
0%
0 dB
-2
2
-2
4
-2
8
-2
6
-3
2
-3
0
-3
7
-3
5
10 dB
15 dB
20 dB
-3
9
-4
3
-4
1
% of System
Capacity Available
Figure 19: WCDMA Data: Percentage of System Capacity Available for Two-Tone Power from -43 dBm
to -21 dBm for a Maximum Load Factor of .7 and Interference Suppression from 0 to 20 dBm Without
LNA Bypass Capability
7.
394
Francis J. Smith
Sensitivity in dBm
-22
-26
-30
-35
-39
-43
0 dB
10 dB
15 dB
20 dB
-90.0
-95.0
-100.0
-105.0
-110.0
-115.0
-120.0
-24
-21
-34
-30
-27
-43
-40
-37
Sensitivity in dBm
395
Impact of the Interference from Intermodulation Products on the Load Factor and Capacity of Cellular CDMA2000
and WCDMA Systems and Mitigation with Interference Suppression
Sensitivity in dBm
-2
1
-2
4
-2
7
-3
0
-3
4
-3
7
-4
0
-4
3
0 dB
10 dB
15 dB
20 dB
Sensitivity in dBm
-2
1
-2
4
-2
7
-3
0
-3
4
-3
7
-4
0
-4
3
-80.0
-85.0
-90.0
-95.0
-100.0
-105.0
-110.0
-115.0
-120.0
0 dB
10 dB
15 dB
20 dB
Sensitivity in dBm
-60.0
-70.0
-80.0
-90.0
-100.0
-110.0
-120.0
-2
1
-2
4
-2
7
-3
0
-3
4
-3
7
-4
0
-4
3
0 dB
10 dB
15 dB
20 dB
396
Francis J. Smith
Harri Holma and Antti Toskala, WCDMA for UMTS, Radio Access for Third Generation Mobile
Communications, Second Edition (John Wiley & Sons, 2002), Chapter 8 Radio Network Planning.
[2]
Vijay K. Garg, IS-95 CDMA and cdma2000; Cellular/PCS Systems Implementation, (Prentice Hall
PTR, 2000), Chapter 13 Reverse and Forward Link Capacity for IS-95 CDMA Systems.
[3]
Clint Smith, Practical Cellular & PCS Design (McGraw Hill, 1997), Chapter 3 11 System
Design Guidelines.
[4]
Jorg Eberspacher, Hans-Jorg Vogal, and Christian Bettstetter, GSM Switching, Services and
Protocols, Second Edition (John Wiley & Sons, 2001).
[5]
[6]
Robert G. Meyer, Mark J. Shensa, and Ralph Eschexbach, Cross Modulation and Intermodulation in
Amplifiers at High Frequencies, Journal of Solid State Circuits, Vol. sc-7, No. 1, February 1972.
397
TDCDMA: Fusion of
Broadband and Mobility
James Teel
Director of International Wireless Product Management
UTStarcom
The ability to provide affordable and ubiquitous mobile broadband service is a key differentiator for
todays telecommunication providers, in their quest to attain profitable revenue growth in the face of
severe competition. The wireless industry has made significant technological progress in the recent past in
bringing together the once disparate worlds of mobility and media (voice and data). In an effort to provide
true convergence, tremendous strides have been taken to ensure that users derive the benefit of an
untethered society.
Fixed-line service providers and mobile operators around the globe are increasingly evaluating and
deploying time divisioncode division multiple access (TDCDMA) networks in an effort to provide a
low-cost mobile broadband data and voice over Internet protocol (VoIP) service. By satisfying the
increasing demand for advanced data applications and servicesuch as streaming audio and video,
videoconferencing, e-commerce, etc.in a mobile wireless environment, operators will be able to reduce
subscriber churn and increase network usage. The end goal being to increase the average revenue per user
effectively in a market where revenue is constantly declining for the majority of operators and service
providers, and bottom-line profitability, or lack thereof, is a matter of utmost concern.
TDCDMA Technology
Based on the specifications and guidelines set by the 3GPP standards body, TDCDMA is a variant or
subset of the universal mobile telecommunications system (UMTS) standard. Commonly known as
UMTS TDD, TDCDMA utilizes the unpaired UMTS spectrum and operates in 5MHz and 10MHz
channels. TDCDMA is a high-speed, low-latency, packet-based solution that allows for ubiquitous,
always-on broadband access by efficiently utilizing the available spectrum.
In a two-way communication system, two separate channels must be created within the available
bandwidth to allow for each direction of traffic. The method of separating the channels to accommodate
uplink and downlink traffic is known as duplexing. While uplink refers to the signal going from the
mobile unit to the network, downlink refers to the signal from the network to the mobile unit. Duplexing
can be achieved via two distinct methods: frequency division duplexing (FDD), which physically
separates the channels into two distinct frequencies and time division duplexing (TDD), which separates
the channels in time.
Voice traffic, which is relatively predictable and symmetrical, requires relatively constant bandwidth in
both uplink and downlink channels. FDD, with its equal partitions, is a highly efficient wide-area
coverage technology for transporting voice traffic. Data-heavy traffic, on the other hand, tends to be
399
highly asymmetrical in nature, with downlink traffic being far more heavily utilized than uplink and
with frequent bursts of data usage quite common. The TDD single channel is completely responsive to
changes in uplink/downlink demand for fully efficient usage of the available spectrum. While
customers experience an always-on connection, no system capacity is utilized unless data is being
transmitted or received.
TDCDMA takes optimal advantage of the synergies of duplexing, augmenting capacity, and providing
flexibility for deployment of high-use networks at extremely low costboth infrastructure and
operational. Optimal throughput is achieved through efficient on-demand assignment of resources, with
combined downlink and uplink data rates of roughly 13 Mbps. In a 10 MHz spectrum in a typical
configuration, TDCDMA allows a peak downlink rate of 11 Mbps and a peak uplink rate of 2 Mbps.
The prior mentioned throughput is halved when using 5 MHz spectrum. In addition, TDCDMA allows
for more compact chip design, which reduces the cost and complexity of the head-end infrastructure and
end-user equipment.
The technology works in a nonline-of-sight, multipath environment, allowing fast wireless data access to
users when outdoors or within buildings and while on the move in dense urban and/or suburban settings.
Depending upon the deployment topography, individual cell size can be as small as a microcell or as large
as 29 km. The network is highly scalable like any other cellular system, as additional base stations can be
added to increase coverage or capacity.
A TDCDMA network can be deployed in a very similar manner to a cellular network, with the
flexibility of omni- and three-sector cells. Existing cellular operators who have deployed wideband
CDMA (WCDMA) networks can leverage their core network infrastructure in deploying the TDCDMA
network, because both TDCDMA and WCDMA are based on the UMTS standard. The TDCDMA base
stations can also be co-located with UMTS FDD base stations, effectively lowering the operating costs
for the operator.
Target Market and Market Opportunity
TDCDMA solutions are a good fit for both mobile operators with allocated UMTS spectrum and
fixed-line service providers with available spectrum. TDCDMA networks can be deployed in
following spectrums: 19001920 MHz, 20102025 MHz, 25002690 MHz, and 34003600 MHz. Not
only does this provide mobile operators with an entry point to provide mobile broadband service, it also
provides fixed-line operators a beachhead to offer mobility-based services for broadband data and voice
over IP applications.
Deutsche Bank, in a recent study, has estimated the aggregate market opportunity for TDCDMA
infrastructure, excluding handsets, to be roughly $2 billion over the next five years. UTStarcoms belief is
that this estimate takes into consideration just the infrastructure opportunity for the traditional mobile
operators without accounting for the fixed-line service providers who have access to spectrum in the
25002690 MHz and 34003600 MHz bands. Traditional service providers and new entrants can leverage
TDCDMA to bypass incumbent broadband operators in offering broadband data service.
Mobile operators, especially those with unused UMTS spectrum, can utilize TDCDMA solutions to
deploy a low-cost third-generation (3G) network to compete with incumbent broadband providers while
providing the value-add benefit of mobility.
400
James Teel
Fixed Operators
Phone Based
Voice Centric
Limited Growth in Data
Declining ARPUs
FIXED OPERATORS
NEW OPERATORS
Figure 1
Conclusion
TDCDMA offers ubiquitous availability at broadband speeds with the added benefit of an IP core, which
enables deployment of value-add services previously not possible with time division multiplex (TDM)
based solutions. With an infrastructure cost, excluding customer premise equipment (CPE), of USD150200 per-subscriber, TDCDMA offers an affordable, low-cost solution for the operator and users, with an
assurance of quick return on investment (ROI) and potential payback period of between two and three
years for the operator.
As the telecommunication industry is evolving from fixed-line services to wireless services, from
narrowband networks to broadband networks, and from circuit-based topology to IP, TDCDMA is a step
towards the convergence of all the aforementioned evolutionary trends.
401
Kamila Piechota
Silicon & Software Systems
Abstract
This paper discusses quality of service (QoS) facilities being defined in the upcoming Institute of
Electrical and Electronics Engineers (IEEE) 802.11e standard. It analyzes the medium access control
(MAC) protocol enhancements as they are specified in the latest 802.11e draft specification [2], and
compares them with the original 802.11 standard [1]. This paper seeks to give the reader a taste for the
complexities and extent of changes in system architecture required for upgrading an existing 802.11 MAC
to provide standards compliant QoS. There are other QoS related functional blocks such as admission
control and schedule generation, which are logically above the MAC layer and as such are outside the
scope of this paper. The concepts described in this paper are in line with ongoing design activities within
Silicon & Software Systems to enhance the QoS functionality of wireless local-area networks (WLANs).
Introduction
802.11 is an IEEE standard for WLANs operating in the unlicensed 2.4 GHz and 5 GHz industrial,
scientific, and medical (ISM) bands. The 802.11b physical layer (PHY) amendment is an extension to the
original 802.11 direct sequence spread spectrum (DSSS) PHY and supports up to 11 Mbps data rate.
Currently it is the most deployed WLAN technology. Another PHY amendment, 802.11a, supports up to
54 Mbps using orthogonal frequency division multiplexing (OFDM) modulation techniques and operates
in 5 GHz band. Although 802.11a supports higher data rates, it is not as widely deployed, since it is a
later standard than 802.11b and the 5 GHz band is not available worldwide. A new PHY amendment,
802.11g, has just been ratified which achieves significantly higher data rates than 802.11b in the 2.4 MHz
band using OFDM modulation. It also seems to be more robust than 802.11a, mainly due to better
propagation characteristics in the 2.4 GHz band.
Of all the available WLAN technology, 802.11 has received the widest acceptance. According to market
researcher Allied Business Intelligence (ABI), wireless fidelity (Wi-Fi) chipset shipments are set to grow at a
403
compound annual growth rate (CAGR) of 43 percent between 2003 and 2007. If this forecast is correct it will
put shipment at about 148 million chipsets and revenues at $1.13 billion in 2007. In 2002 shipment of chipset
hit 23 million to 25 million units, up from 7.9 million in 2001. Where the initial expectation was 14 million to
15 million, 802.11 has experienced great success during the last two years mainly due to its simplicity,
robustness, reliability, and cost effective high mobile bandwidth. During the past year, prices have fallen
dramatically and further pressure on prices is expected, due to intensive competition in the market. It is likely
that 802.11 will play a major role in next-generation wireless communication networks.
802.11 is often considered as a wireless version of Ethernet, and it has become very popular
complementing and/or replacing Ethernet in many environments including corporate, public, and home.
Its popularity is growing, as the technology is adequate for basic connectivity. As 802.11 is being
incorporated into more and more application areas, the need for further feature enhancements is required.
One of these features is support for applications with quality of service (QoS). QoS issues of the original
standard and enhancements introduced by the 802.11e amendment are the subjects of this paper.
Section 2 of this paper defines and discusses QoS in WLAN. Section 3 introduces channel access
functions defined in the original standard and discusses its limitations in support of QoS. Section 4
presents the enhancements introduced by 802.11e. Changes in system architecture to upgrade an 802.11
MAC to provide QoS are discussed in Section 5. The paper is concluded with a summary in Section 6.
1. Quality of Service (Qos) in WLAN
In general the term QoS refers to set of qualitative and quantitative traffic characteristics (e.g. throughput,
service interval, packet size, delay, jitter, priority, type, etc.), which describe a traffic flow in support of a
specific application.
802.11 WLANs provide best-effort services similar to Ethernet. In wired Ethernet, QoS issues have been
neglected since data rate is very high and the physical layers error rate very low. However wireless
systems incur very high per-packet overhead with limited bandwidth, so the same does not apply to
802.11. For example, about 32% of available data rate is consumed by packet fragmentation, interframe
spacing, and MAClevel acknowledgment. If request to send /clear to send (RTS / CTS) is enabled, it can
approach 50%. In addition, heavy traffic load increases collisions and backoffs, so frame delivery time to
the peer station increases exponentially. Frequent retransmissions also cause unpredictable delays in the
order of tens to hundreds of milliseconds and transmission of pending frames could be blocked.
The quality of most multimedia services involving voice and video transmission deteriorate dramatically
if delay increases beyond certain level. This is a major stumbling block preventing 802.11 to enter new
markets such as consumers electronics (wireless audio visual distribution at home), or medical systems
(wireless patient monitoring).
For the wired network, Internet Engineering Task Force (IETF) Integrated Services (IntServ) [3] and
Differentiated Services (DiffServ) [4] architectures have been proposed to support guaranteed QoS and traffic
prioritisation respectively above the link layer. IntServ provides service guarantees to individual traffic flows.
A traffic flow is defined as a distinguishable traffic stream of datagrams from a unique sender to a unique
receiver that results from a single user activity and requires the same QoS. Signaling protocol, e.g. RSVP, must
be used prior to data transmission to reserve resources at every hop (IP router). IntServ provides high levels of
precision in resource allocation, however there are some scalability issues since the protocol requires
maintaining a soft state for each flow at every router along the reservation path.
In contrast, DiffServ mechanisms do not use per-flow signalling, and as a result do not consume per-flow
resources within the routing infrastructure. The DiffServ protocol allocates different service levels to
404
different groups of users, which means that all traffic is distributed into groups or classes with different
QoS parameters. This reduces maintenance overhead in comparison to IntServ, as state awareness is
required only at edge routers. At the edge router datagrams are classified as belonging to the certain
service class and then conditioned (marked, policed, and shaped) according to the service level
agreement. Once the datagram is in the DiffServ domain it gets the same treatment from all the routers
according to the service category with which it has been marked by the edge router.
When used to support applications with quality of service requirements, each 802.11 WLAN may provide
a single link within an end-to-end QoS environment that may be established between and managed by
higher layer entities. The 802.11 QoS facility requires the 802.11 MAC sublayer to incorporate features
that are not traditional for MAC sublayers. In addition, it may be necessary for certain higher layer
management entities to be "WLAN aware," at least to the extent of understanding that the available
bandwidth may be subject to sudden and unavoidable changes. In order to support both IntServ and
DiffServ protocols the 802.11e Task Group has proposed a number of enhancements to the MAC layer.
2. Original 802.11 MAC
The main purpose of the 802.11 MAC layer is to provide reliable data services to the user of the MAC
(i.e. higher layer protocols) and to control fair access to the shared wireless medium. This paper considers
an infrastructure basic service set (BSS) composed of an access point (AP) and a number of stations
(STAs) associated with the AP.
To provide reliable data services the 802.11 standard defines a frame exchange protocol. The minimum
exchange sequence consists of two frames: a frame sent from source to the destination and an acknowledgment
(ACK) sent from destination to source if the frame is received successfully. For every frame received at the
MAC the frame check sequence (FCS)(IEEE 32bit CRC) is checked. If the source does not receive the
expected ACK or the FCS fails, the frame is re-transmitted. This mechanism consumes some of the available
bandwidth, but it allows coping with error conditions such as interference in the wireless medium. In general it
is more efficient to ensure data integrity at the link layer than leaving it up to higher layer protocols, which rely
on mechanisms such as timeouts measured in seconds instead of milliseconds at the link layer.
In addition to this basic frame exchange sequence, an optional RTS/CTS mechanism is defined to
increase the robustness of the protocol and address problems such as the hidden node. A STA sends a
RTS frame to the destination before transmitting any MAC service data unit (MSDU). The destination
then responds with a CTS once it has correctly received a RTS. The source can then send the MSDU after
receiving the expected CTS response. Both RTS and CTS frames contain duration information about the
length of the MSDU and ACK transmission. Based on this information all surrounding STAs can update
an internal timer called network allocation vector (NAV) and defer any transmission until this timer
expires. Even if a hidden STA cannot hear the RTS from the source STA, it will be able to receive the
CTS response from the destination STA and update its NAV accordingly. This mechanism protects
transmission between STAs against unexpected transmissions from hidden STAs, and it is also used by
the 802.11g amendment to protect unexpected transmissions from legacy equipment.
The 802.11 standard specifies two channel access functions: distributed coordination function (DCF) and
point coordination function (PCF). Both channel access functions and their limitations in the support of
QoS traffic are discussed in the following sections.
2.1 Basic Access Method of the Original MAC DCF
The basic channel access function of 802.11 is DCF that allows sharing of the wireless medium between
compatible PHYs through the use of a carrier sense multiple access with collision avoidance (CSMA/CA)
protocol. This mechanism is mandatory for all STAs, including 802.11e QoS enhanced stations (QSTAs).
405
Carrier sense is performed using both physical and virtual mechanisms. Physical carrier sense means
before attempting any transmission a STA senses the channel condition and that the wireless medium is
idle for a certain time period. The duration of the time period varies, but the one used before a
transmission attempt under normal condition is called the DCF interframe space (DIFS). To avoid
collision between two simultaneously transmitting STAs, a backoff algorithm is used as well as deferral
of transmission for DIFS time. When there are pending transmission requests and the wireless medium is
detected busy, a STA would wait until the wireless medium is idle for DIFS time. Then it draws a random
number from a range of values, and this number is used to calculate an addition time period that it has to
wait. The range from which the random number is drawn is called the contention window (CW), which
varies depending on the number of previous re-transmission attempts. If the wireless medium is detected
to become busy during backoff, the count down is suspended. It is resumed once the wireless medium is
detected to be idle subsequently for DIFS time.
Collision avoidance is achieved by a virtual carrier sense mechanism. Each STA maintains an internal
timer, called the network allocation vector (NAV), which indicates when the wireless medium is busy. A
duration value is included in each frame that is transmitted by a STA, which indicates how long the
transmission lasts, including any subsequent acknowledgements and fragments. All STAs in the vicinity
receives the frame and uses the duration value to update its NAV. Then when a STA is about to start a
transmission, it first checks that its NAV is zero. If not, it must wait until it counts down to zero because
the NAV indicates that another STA has access to the wireless medium.
On winning contention to the wireless medium, a STA can transmit one MSDU. Then the STA waits for a
time period called short interframe space (SIFS) for the ACK from the MSDU recipient. SIFS is shorter in
duration than DIFS, which gives the ACK frame highest priority for accessing the wireless medium. This
ensures that any other STA would not start transmission while the ACK is expected. If the ACK is not
received after SIFS, retransmission is attempted until either the number of retransmission has exceeded certain
thresholds or the lifetime of the MSDU has expired. In these cases, the MSDU is discarded.
MSDUs can also be fragmented to increase the probability of successful transmission. However since
each fragment of the MSDU is individually acknowledged, fragmentation significantly increases
transmission overhead of the MSDU. An example of DCF frame exchanges is shown in Figure 1, with
STA D being a hidden node from STA A.
SIFS
DIFS
SIFS
SIFS
SIFS
SIFS
SIFS
DIFS
Backoff
(8 slots)
STA B
Backoff
(12 slots)
ACK
STA C
STA D
Data
fragment
Data
fragment
RTS
STA A
CTS
ACK
SIFS
Data
Resume
Backoff
(4 slots)
Data
ACK
406
New
Backoff
(15 slots)
Access
Point
(PC)
PIFS
Data
SIFS
Beacon
SIFS
Data +
CF-Poll
SIFS
CFPoll
Data +
CF-Ack
STA A
STA B
SIFS
PIFS
SIFS
CFPoll
CFEnd
No Response
after SIFS
ACK
SIFS
CFAck
t
Contention Period (CP)
CP
407
In summary, neither DCF nor PCF provide sufficient facility to support traffic with QoS
requirements.
3. 802.11e QoS Features
The wireless medium has fundamentally different characteristics from a wired medium. When providing
QoS, we should remember that the MAC endeavours to provide QoS service guarantees within this
inherently unpredictable medium. Therefore bandwidth and latency cannot be guaranteed like a wired
system, especially in unlicensed spectrum. However, we can always provide mechanisms that would
allow maximizing the probability that certain traffic classes will get adequate QoS, so that in a properly
controlled environment true QoS guarantees can be made.
To provide support for QoS the IEEE 802.11 Task Group E (802.11e) defines enhancements to the
original 802.11 MAC described earlier. 802.11e defines a superset of features specified in the 1999
edition of IEEE 802.11. These enhancements distinguish QSTAs from nonQoS stations (STAs), and
QoS enhanced access point (QAP) from nonQoS access point (AP). These features are collectively
termed QoS facility.
There are two main functional blocks defined in 802.11e. These are the channel access functions and
traffic specification (TSPEC) management. TSPEC management provides the link between the channel
access functions and higher layer QoS protocols such as IntServ or DiffServ. Optional features such as
block acknowledgement (BA), direct link protocol (DLP), automatic power save delivery (APSD), and
local multicast service class are not directly related to QoS, but improve the efficiency of the 802.11
MAC in general.
3.1 Channel Access Functions
QoS facility defines a new coordination function called hybrid coordination function (HCF) used only in
QoS enhance BSS (QBSS). HCF has two modes of operation: enhanced distributed channel access
(EDCA) is a contention-based channel access function that operates concurrently with HCF controlled
channel access (HCCA) based on a polling mechanism which is controlled by the hybrid coordinator
(HC). The HC is co-located with the QAP. Both access functions enhance or extend functionality of the
408
original access methods DCF and PCF. EDCA has been designed for support of prioritised traffic similar
to DiffServ, whereas HCCA supports parameterised traffic similar to IntServ.
The basic concept of these channel access functions is the transmission opportunity (TXOP). A TXOP is
a bounded time interval in which the QSTA is allowed to transmit a series of frames. A TXOP is defined
by the start time and a maximum duration. If a TXOP is obtained using the contention-based channel
access, it is called an EDCATXOP. If a TXOP is granted through HCCA, it is called a HCCA (polled)
TXOP. The duration of the EDCATXOP is controlled by the QAP and is distributed to nonAP QSTAs
in the beacon frames along with other EDCA related parameters. The duration of a HCCA (polled) TXOP
is passed to the nonAP QSTA directly by the HC as part of a QoS CFPoll frame, which grants the
HCCA (polled) TXOP.
8 0 2 .1 1 e p e rio d ic B e ac o n In te rv a l
C FP
CF-End
Beacon
TBTT
CP
B ro a d c a s t &
E D C A -T X O P
M u lt ic a s t F ra m e s R T S /C T S /D a ta /A C K
t
H C C A -T X O P
Q o S _ C F - P o ll/( D a ta /A C K )s
TBTT
C o ntro lle d A c c e s s P h as e (C A P )
409
TXOP limit
The maximum duration for which a QSTA can transmit after obtaining a TXOP
When data arrives at the MACUNITDATA service access point (SAP), the 802.11e MAC first classifies
the data with the appropriate AC, and then pushes the newly arrived MSDU into the appropriate AC
transmit queue. MSDUs from different ACs contend for EDCATXOP internally within the QSTA.
The internal contention algorithm calculates the backoff independently for each AC based on AIFSN,
CW, and a random number. The backoff procedure is similar to that in DCF, and the AC with the smallest
backoff wins the internal contention.
The winning AC would then contend externally for the wireless medium. The external contention
algorithm has not changed significantly compared to DCF, except that in DCF the deferral and backoff
were constant for a particular PHY. 802.11e has changed the deferral and backoff to be variable, and the
values are set according to the appropriate AC. One possible implementations of the external contention
mechanism is illustrated in the Figure 4.
AIFS
Busy
TXOP Limit
EDCF - TXOP
SIFS
410
EDCA
Higher
level SW
queues
AC BE
AC BK
AC VI
AC VO
TXOPLimit [AC]
AIFSN [AC]
CW [AC]
TXOPLimit [AC]
AIFSN [AC]
CW [AC]
TXOPLimit [AC]
AIFSN [AC]
CW [AC]
TXOPLimit [AC]
AIFSN [AC]
CW [AC]
TX Descriptor
TX Descriptor
TX Descriptor
TX Descriptor
EDCF internal
contention and
collision algorithm
Pointer associations
Link list
EDCA
Low level
SW queue
MAC
HW
TX Transmitted Entry
TX Current Entry
TX Backoff
TXOP Limit
TX Next Entry
411
The central concept of HCCA is controlled access phase (CAP), which is a bounded time interval and
formed by concatenating a series of HCCA (polled) TXOPs. Scheduling of HCCA (polled) TXOP and the
formation of CAP are performed by the HC. Figure 6 illustrates an example frame exchange sequence
during the CAP.
SIFS
PIFS
QAP
QSTA
SIFS
QoS Data
SIFS
SIFS
SIFS
SIFS
QoS
CFPoll
QoS Data
ACK
SIFS
ACK
QoS
Null
SIFS
SIFS
ACK
QoS
Data
ACK
QoS
Data
Durations
Durations
TXOP 2
TXOP 1
aSlotTime
aSlotTime
412
TSID = x
n
TX Descriptor
TX Descriptor
HCCA Scheduling
algorithm
Pointer associations
Link list
HCCA
Low level
SW queue
MAC
HW
HCCA Transmitted
413
QoS Management frames, primitives and procedures are defined for TSPEC negotiation, which is always
initiated by the SME of a QSTA, and accepted or rejected by the HC. Requested TSPEC is communicated
to the MAC via the MAC Layer Management Entity (MLME) SAP. This allows higher layer SW,
protocols, and application, such as RSVP, to allocate resources within the MAC layer. The MSC in
Figure 8 illustrates a typical TSPEC negotiation.
QSTA Higher Layer
QSTA MAC
HC MAC
HC Higher Layer
MLME-ADDTS.request
ADDTS request
MLME-ADDTS.indication
ADDTS
response timer
MLME-ADDTS.response
ADDTS response
MLME-ADDTS.confirm
414
Broadcast
& Multicast
Higher
level
SW
queues
AC VO
AC TXOPLimit
VI
[AC]
AC TXOPLimit
BKAIFSN[AC]
[AC]
AC TXOPLimit
BEAIFSNCW
[AC]
[AC]
[AC]
TSID = x
TSID = x
TSID = x
TSID = 0
TXOPLimit
[AC]
AIFSN
[AC]
CW
[AC]
AIFSN
[AC]
CW
[AC]
CW [AC]
CF-End
TX Descriptor
TX Descriptor
TX Descriptor
TX Descriptor
TX Descriptor
EDCA internal
contention and
collision algorithm
Low
level
SW
queue
TX Descriptor
TX Descriptor
TX Descriptor
TX Descriptor
HCCA
Scheduling
Algorithm
EDCA TXOP
HCF Scheduler
HW
Original 802.11
415
Admission
Control
Traffic Classifier
MLME
SAP
MA-UNITDATA
SAP
MLME
(QoS)
SAP
UMAC
Mgt
Entity
UMAC
Schedule
Generator
MAC
Packet Classifier
Broadcast
&
Multicast
LMAC
EDCA
TXOP Scheduler
HW
HCCA
PLME
SAP
PHY SAP
416
LAN far out weights the little loss in QoS, and there is a significant market to be tapped by the addition of
QoS to 802.11.
6. References
[1]
[2]
[3]
[4]
[5]
IEEE, Part 11, Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY)
Specifications, ANSI/IEEE Std. 802.11, 1999 Edition, (ISO/IEC 8802-11:1999(E)).
IEEE, Part 11, Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY)
Specifications: Medium, Access Control (MAC) Enhancements for Quality of Service (QoS),
ANSI/IEEE Std 802.11e, Draft 5.0, July 2003.
R. Braden, D. Clark, S. Shenker RFC1633 Integrated Services in the Internet Architecture: an
Overview, June 1994
Blake, D. Black, M. Carlson, E. Davies, Z. Wang, W. Weiss, RFC2475 An Architecture for
Differentiated Service, December 1998.
IEEE, Part 3, Medium Access Control (MAC) Bridges, ANSI/IEEE Std. 802.1D, 1998 Edition,
(ISO/IEC 15802-3:1998).
417
Amit Mukhopadhyay
Distinguished Member of Technical Staff
Bell Laboratories, Lucent Technologies
Zulfiquar Sayeed
Member of Technical Staff
Bell Laboratories, Lucent Technologies
Dong Sun
Member of Technical Staff
Bell Laboratories, Lucent Technologies
Carlos Urrutia-Valds
Member of Technical Staff
Bell Laboratories, Lucent Technologies
Abstract
The first version of Universal Mobile Telecommunication Services (UMTS) is being rolled out in much
of Europe during 20042005. A major carrier in the United States has already deployed third
generation 3G-1X EV-DO and others are going along either the code division multiple access (CDMA)
2000 or the UMTS path. Many wireline and wireless service providers, all across the globe, are
supplementing their services with 802.11wireless fidelity (Wi-Fi) access. The vision of International
Mobile Telecommunications (IMT)-2000 is finally becoming a realityend-user wireless service of up
to two megabits per second (Mbps) is indeed here.
However, providing the required throughput is only a part of the story. The real challenge is to provide
adequate quality of service (QoS) so that subscribers can effectively utilize this bandwidth. There are
many obstacles toward providing end-to-end QoS. First, QoS mechanisms in the standards bodies are not
equally mature in all the broadband wireless technologies. Second, even when some mechanisms are
standardized, their implementations can vary from vendor to vendor and the network operations personnel
in the service providers organization will need a lot of expertise in setting the right parameters for
419
different services. Last but not the least, end-to-end services may be carried over multiple service
providers networks, and the originating or terminating service provider may have no control over the
QoS implementations in those intermediate networks.
This document presents mechanisms for QoS implementations in UMTS and CDMA2000 networks and
points out some key characteristics. Also presented are some QoS items currently in the process of being
standardized in third-generation partnership project (3GPP/3GPP2). Further discussed is the QoS
framework in 802.11 networks and integration of Wi-Fi and 3G networks. Finally, the paper concludes
with thoughts on issues still unaddressed in standards organizations.
Introduction
During the days of the first-generation (e.g., AMPS, NMT, and DECT) and second-generation (e.g.,
GSM) CDMA One, and ANSI-136) wireless systems, voice in circuit mode was the main source of
traffic. QoS was quantified in terms of parameters similar to those in wireline, e.g., call set up delay,
round trip delay, echo, blocked calls, dropped calls etc and the main additional criteria was performance
during a hand-off. Voice QoS was measured primarily in terms of the International Telecommunication
Union (ITU) R rating [1, 2] or mean opinion score, [3] and frame error rate was the primary source of
dissatisfaction. Table 1 provides a simple overview of QoS requirements for voice.
VOIP Requirements
Parameters, e.g.,
Blocking
<0.1%
Bandwidth
64 kbps
Packet Delay
Jitter
<100 msec
Packet Loss
<0.1%
Echo Cancellation
As needed
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
Finally, appropriate service-level agreements (SLAs) will have to be instituted with partner operators to
provide end-to-end services. Figure 1 depicts the importance of having appropriate QoS mechanisms
installed in all networks involved in the traffic flow, irrespective of whether or not the wireless operator
has any control over that network.
421
HLR
Node B
Public Switched
T1/E1 Telephony Network
Iu-CS
RNC
Node B
GTP-U
SGSN
Legend:
GGSN
ATM Core
Node B
Iu-PS
Public Data
Network
IP
MSC
IP Core
Mobiles
Conversational
Streaming
Interactive
Background
Typical Characteristics
Symmetric bidirectional
Stringent delay and jitter
requirements
Mostly
unidirectional
Time relation
between info.
elements very
important
Asymmetric traffic
Sensitive to errors
Sample Applications
Audio/video
streaming
Web browsing
File transfer
422
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
UMTS
TE
MT
RNC
SGSN
GGSN
TE
End-to-End Service
User Traffic Classification
TE/MT Local
Bearer Service
External Bearer
Service
Encapsulated
into
GTP packets
Iu Bearer
Service
Fragment
into packets
Encapsulated
into
GTP packets
CN Bearer
Service
Fragment
into packets
Backbone
Bearer Service
(ATM, MPLS etc)
End-to-end QoS, which provides the QoS of bearer service across the whole UMTS (UTRAN and
CN) and external networks (e.g. Internet) based on the unified IP mechanism
UMTS bearer QoS, which provides the QoS of bearer service of radio channel and wired
transport network within the UMTS domain
423
Two entities play a very important role in the interaction between end-to-end bearer service and UMTS
bearer servicethe MT and the GGSN. The translation functions in these two entities at the two edges of
the UMTS network ensure that the QoS parameters from one bearer service are properly mapped to the
other. While the UMTS operator has control over its own UMTS network, it has to depend upon partner
networks to deliver end-to-end IPbearer QoS.
Once the IP bearer QoS parameters have been mapped to UMTS bearer QoS parameters, it becomes the
responsibility of the UMTS bearer to provide QoS within the UMTS domain. There are two fundamental
planes within the UMTS network to provide this QoS: the control plane (elements with signaling
mechanism supporting session set up and session modification) and the user plane (components carrying
user traffic). Each plane has its own set of mechanisms for ensuring QoS.
In the control plane, the various bearer service (BS) managers in MT, UTRAN, SGSN, and GGSN
perform the management operations for establishing, maintaining, and terminating the QoS service. In
particular, the UMTS BS manager is responsible for the control of the overall signaling exchange,
supported by the other functions for tasks like primitives conversion, admission control, etc. The control
plane functions related to QoS are as follows:
Admission or capability function, which controls the admission of a service in two aspects: (1) the
allocation of resources based on the available resources (within its own control) or other policies,
and (2) the capability of the service invocation, based on administrative issues
Subscription control function, which authorizes the usage of the service for a particular user
Translation function, which is placed at the edges of the UMTS and converts service primitives of
the UMTS BS to corresponding external primitives toward external networks and/or the traffic
engineering (TE)
Service manager function, which coordinates the functions of the control plane for establishing,
modifying, and maintaining the service for which it is responsible. It provides all the user plane
management functions with relevant QoS attributes. There are different instances of service
managers as subsequently explained.
Table 3 shows the different control plane QoS entities and the elements where they are located.
Function/ Element
MT
UTRAN
Translation Function
SGSN
Local
UMTS
Radio
Iu
x
x
CN
x
x
Ext
x
x
Physical
x
x
Subscription Control
BS
Manager
GGSN
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
The user plane QoS management functions maintain user data traffic within certain limits, per QoS
attributes. The main functions consist of the following:
Classification function, which decides on a particular service, among the established UMTS BS
for an MT, to accommodate a specific flow of packets, based on information of the packet header
or other traffic characteristics
Mapping function, which marks the packets in order to receive the intended QoS at each network node
Traffic-conditioning function, which performs policing to check that the traffic is conformant to
the negotiated QoS profile. If it is not conformant, the packets are either dropped or marked
appropriately for preferential rejection in case of congestion. A traffic shaper may be used to
bring the traffic stream into compliance with its traffic profile. Traffic conditioning is performed
in the MT for upstream traffic.
Resource-management function, which performs scheduling, along with bandwidth and queue
management, in order to distribute the available resources among the established services
appropriately
Table 4 shows different QoS functions in the user plane and their locations.
Function/ Element
MT
Mapping
UTRAN
SGSN
GGSN
Classification
Resource Management
Traffic Conditioning
x
x
425
Attribute
Description
Significance
Traffic Class
Maximum Bitrate
Guaranteed Bitrate
Delivery Order
Transfer Delay
Allocation/Retention Priority
426
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
administrative rights. The established PDP context controls the QoS and allocates the bearer resource to
GTP tunnel within the UMTS network.
Figure 4 illustrates a detailed UMTS BS QoS establishment process. When the UE, also referred to as
mobile station (MS) consisting of TE and MT) initiates a UMTS call, the UE gathers application
attributes and translates these attributes into appropriate UMTS QoS parameters, and then encapsulates
them in the PDP context activation request to send the activate PDP context request to SGSN.
UE
RAN
Establishes
Establishes aa signalling
signalling
connection
connection between
between the
the
MS
MS and
and the
the SGSN,
SGSN, and
and
allocates
allocates resources
resources on
on
SGSN
SGSN for
for GTP
GTP tunnel
tunnel
SGSN
GGSN
Perform
Perform admission
admission control
control
Map
Map QoS
QoS profile
profile to
to RAB
RAB QoS
QoS
Parameters
Parameters
Map
Map QoS
QoS profile
profile to
to DiffServ
DiffServ code
code
points
points for
for backbone
backbone QoS
QoS
Setup
Setup radio
radio channel
channel
Setup
Setup channel
channel coding
coding and
and
interleaver
interleaver
Allocate
Allocate
resources
resources in
in
GGSN
GGSN
Map
Map to
to
external
external bearer
bearer
QoS
QoS
(DiffServor
(DiffServor
RSVP)
RSVP)
Setup
In
In case
case the
the QoS
QoS attributes
attributes
have
have been
been changed,
changed, the
the
SGSN
SGSN may
may inform
inform the
the
GGSN
GGSN about
about the
the changed
changed
QoS
QoS attributes
attributes
radio resource control (RRC) messages. On completion of radio resource allocation, the RNC sends the
RAB assignment response to the SGSN.
In case the QoS attributes have been downgraded during the RAN setup, the SGSN may inform the
GGSN about the downgraded QoS attributes by sending an update PDP context request to the affected
GGSN. The GGSN confirms the new QoS attributes by sending an update PDP context response to the
SGSN. The SGSN sends Activate PDP Context Accept to the MS. At this point, the UMTS bearer setup
is complete with required QoS.
Use of both IETF differential services (DiffServ) and integrated services model (IntServ) models are
supported by UMTS. As an example, the DiffServ model [14] can be used in UMTS for Iu, CN, and
external network QoS. The DiffServ approach does not reserve resources for individual data flows or
users. Instead, DiffServ differentiates between different classes of traffic such as voice, video, or data.
Each packet includes information about its class of service. Routers process the class of service value to
provide appropriate QoS. It is important to note that traffic is not prioritized within a given service class.
Hence, the differentiated services approach cannot strictly promise quantifiable QoS for a single user or a
specific session.
All the routers in the network are required to support defined classes of service and associated QoS
requirements. When a UE packet is received by the RNC, it marks the packet with the appropriate class of
service before forwarding it to the SGSN. On receiving the packet from the RNC, the SGSN marks it with
the class of service and sends it to through the backbone network to the GGSN. Similarly, the GGSN
marks the packet with the appropriate class information before forwarding it to the external network.
UMTS Rel 5/Rel 6
UMTS Rel 5 introduces the IP multimedia core network sub-system (IMS) with its associated session
initiation protocol (SIP) signaling for the establishment and control of SIP-based IP multimedia sessions.
Figure 5 shows a simplified Rel 5 IMS architecture and its main components.
Call session control function (CSCF). This function can be of three different types: proxy,
interrogating, and serving and serves as the central controller for all sessions.
Break-out gateway control function (BGCF) and media gateway control function (MGCF). These
functions together with the media gateway (MGW) are used to interconnect to the public
switched telephone network (PSTN).
Media resource feature controller (MRFC) and its associated media resource function processor
(MRFP) are used for media duplication, codec modifications, tone generation, and the like.
Rel 5 also introduces a mechanism for QoS management at the application layer by the use of the service
based local policy (SBLP) decision function [6][7]. This is important for the newly introduced SIPbased
IP multimedia sessions supported by UMTS Rel 5 and beyond. This concept is an adaptation of the IETF
policy-based admission control framework [13]. The objective of this framework is to enhance admission
control policies, which today only take into account the users request for resources and the availability of
these resources at the time of the request, to include other considerations such as the identity of the user,
the time of request, and security policies.
In Rel 5, the policy decision function (PDF) is part of the PCSCF and communicates with the GGSN via
the also new Go interface [4]. In Rel 6, the PDF has been separated from the PCSCF and the interface
between the PDF and the PCSCF has been standardized as the Go interface [5]. However its basic
functionalities remain the same. Figure 5 shows the PDF and Go interface in Release 5.
428
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
HSS
SIP
UE
Terminal
Node B
RNC
SGSN
GGSN
Go
IMS components
OSA
AS
OSA
Bearer
AS
Signaling
OSA
Gateway
Legend:
SIP AS
SIP AS
P-CSCF PDF
I-CSCF
S-CSCF
BGCF
MRFP
MRFC
MGCF
Public Switched
Telephony Network
MGW
SS7
DNS
Request (REQ) message. Used by the GGSN to request policy information from the PDF
Decision (DEC) message. Used by the PDF to inform the GGSN of its decision to authorize or
deny a QoS request
Report state (RPT) message. Used by the GGSN to acknowledge a command from the PDF or to
report errors
Delete request state (DRQ) message. Used by the GGSN to inform the PDF that allocated
resources are not longer available or applicable and the corresponding policy state should be
deleted from the PDF (e.g., at session termination)
To appreciate the SBLP mechanism more fully, let us examine the interactions among the UE, PDF,
and GGSN during a successful session initiation. In this example, UE1 is trying to set up a session with
UE2. Only the originating networks procedures are shown as the terminating procedures are basically
the same.
Figure 6 shows UE1 in a visited network sending an initial INVITE message to UE2 (not shown). UE2
could be in a visited or in its home network, but for the purpose of this example, whether UE1 or UE2 are
in their home or a visited network is not relevant. Furthermore, for the sake of simplicity, only the steps
relevant to QoS discussions will be expanded in detail.
429
INVITE
INVITE with
with SDP
SDP offer
offer is
is received
received by
by P-CSCF.
P-CSCF.
183
183 Session
Session Progress
Progress message
message received
received from
from UE2
UE2
with
with SDP
SDP offer
offer response.
response.
Authorize
Authorize QoS
QoS Resources:
Resources:
Offer-response
Offer-response pair
pair contains
contains sufficient
sufficient information
information
regarding
regarding BW
BW requirements,
requirements, and
and media
media
characteristics.
characteristics.
The
The PDF
PDF authorizes
authorizes the
the necessary
necessary resources.
resources.
An
An Authorization
Authorization Token
Token is
is generated
generated by
by the
the PDF.
PDF.
and
and sent
sent on
on the
the 183
183 Session
Session Progress
Progress message
message (the
(the
P-Media-Authorization
P-Media-Authorization header)
header) to
to UE1
UE1
430
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
Go
Go Interface
Interface messages:
messages:
GGSN
GGSN requests
requests authorization
authorization for
for
context
context and
and receives
receives policy
policy
enforcement
enforcement information
information from
from PDF.
PDF.
UE
SGSN
GGSN
1. Mapping of
SDP parameters
into UMTS QoS
2. Activate PDP Req.
3. Create PDP Req.
4. COPS REQ
Upon
Upon receipt
receipt of
of authorization
authorization token:
token:
UE
UE requests
requests aa PDP
PDP context
context activation
activation
for
for the
the agreed
agreed media.
media.
Activate
Activate PDP
PDP context
context request
request
contains
contains binding
binding information:
information:
1.Authorization
1.Authorization Token
Token
2.Packet
2.Packet Flow
Flow Identifiers
Identifiers
5. Process
authorization
request
6. COPS DEC
7. Policy
Enforcement
8. COPS RPT
PDP
PDP context
context activated.
activated. No
No media
media is
is flowing
flowing yet
yet and
and gate
gate status
status at
at GGSN
GGSN is
is closed.
closed.
431
UE1
UE1 sends
sends UPDATE
UPDATE message
message to
to UE2
UE2
indicating
indicating successful
successful reservation
reservation of
of
resources.
resources.
UE2
UE2 responds
responds with
with 200
200 (OK)
(OK) message
message
when
it
has
successfully
when it has successfully reserved
reserved its
its
resources.
resources.
User
User 22 is
is being
being alerted.
alerted.
PRACK
ackowledges
PRACK ackowledges alerting
alerting indication
indication
and
and 200
200 OK
OK acks
acks the
the PRACK
PRACK message.
message.
(This
(This is
is the
the 3-way
3-way handshake
handshake for
for the
the 180
180
Message.)
Message.)
432
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
User
User answers;
answers; UE2
UE2 sends
sends 200
200 (OK)
(OK)
message
message for
for original
original INVITE
INVITE request.
request.
Approval of
QoS commit
P-CSCF
PDF
GGSN
1. 200 OK
2. PDF approves
the QoS Commit
3. DEC
4. GGSN opens the
gates
P-CSCF
P-CSCF receives
receives the
the 200
200 (OK)
(OK) message.
message.
This
This triggers
triggers the
the Approve
Approve QoS
QoS commit
commit
procedure..
procedure..
P-CSCF/PDF
P-CSCF/PDF approves
approves QoS
QoS resources
resources and
and
sends
sends DEC
DEC message.
message.
GGSN
GGSN opens
opens the
the gate.
gate.
UE1
UE1 responds
responds with
with ACK
ACK message
message
and
and starts
starts the
the media
media flow
flow for
for this
this session.
session.
5. RPT
6. 200 OK
433
TE
Rm
interface
MT
RLP
Session
BSC
PCF
Legend:
BTS
Air
Interface
Rm
PDSN
GRE
Session
Abis
A8/A9
I
N
T
E
R
N
E
T
A10/A11
TCP
UDP
TCP
UDP
IP
IP
PPP
GRE
Low-Level
Low-Level
Interface
Interface
TE
IP
IP
WAN
WAN
PPP
RLP
RLP
IS2000
MT
IS2000 L2/T1
BTS
L2/T1
IP
L2/T1
BSC
L2/T1 L2/T1
PCF
GRE
IP
WAN
L2/T1
PDSN
router
TE
434
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
RAN gets standardized and SIPbased terminals become more prevalent in the near future, QoS in the
RAN will need more rigorous approach. QoS mechanisms will be needed not only to support real time
services like VoIP, Push-To-Talk and videoconferencing, but also to deal with performance enhancing
attributes of CDMA networks like soft handoffs.
The responsibility of the PDSN is treating packets based on the subscribers QoS profile as well as the
QoS requested in the service option is explained in detail in [17]. The PDSN uses DiffServ code points
(DSCPs) in the expedited forwarding, assured forwarding and experimental/local classes both toward the
A10/A11 interface as well toward the external IP network.
It may be noted that at the present moment, only SO33 is standardized and only one service instance is
allowed (no auxiliary service instances) per point-to-point protocol (PPP) session, though there may be
several service flows (source-destination pairs) of traffic. However, QoS parameters for RLP sessions and
the RP connections are associated with each service instance. Thus, for all practical purposes, different
types of data packets within SO33 do not get any priority service within the RAN. This situation is going
to change as soon as multiple service instances are allowed in the RAN.
CDMA2000 Bearer Service
This service is being defined within the domain of the wireless service provider, i.e., between the mobile
station and the PDSN. As indicated above, QoS in this entire network is not standardized yet, though
several aspects have been addressed and yet others have been left to be acted upon during future
evolutions of the standards. This subsection captures a step-by-step invocation of QoS procedures in
different network entities as envisioned in the standards, though all the mechanisms may not be already in
place. First, to define a few concepts:
A packet data session is a period of time during which the mobile station maintains a PPP
connection to the PDSN. A mobile can have only one PPP connection with the PDSN, but it can
have multiple IP addresses.
A service instance is a data service established over a PPP connection. Each service instance is
associated with an A10-A11 connection and an RLP connection with specific QoS parameters
associated with them. Different data services with the same QoS requirements, for example Rich
Voice and push-to-talk, can be grouped under the same service instance. The mobile station and
the RAN identify each service instance with an identifier call SR-ID.
A flow is a pair of source-destination combination for a service instance. A service instance can
have multiple service flows. A three-way VoIP call is an example of two service flows (to the two
external parties) over the same service instance (VoIP). Similarly, Rich Voice and
videoconferencing between two parties can be two flows (different RTP ports) within the same
service instance (if the two services have same QoS requirements). Packet filters are used to
identify different flows within a service instance.
Traffic flow templates are used to map traffic coming from external IP networks toward mobiles
into proper service instances and to indicate if special flow treatment (e.g., header compression)
is to be applied.
A subscriber QoS profile is stored in the HAAA and retrieved by the PDSN at the time of
registration. It contains the allowed DiffServ markings attribute and the service options profile
attribute. The PDSN uses this information to mark packets from the mobile toward the external
network and to set up different A10-A11 connections.
435
The following QoS-related events take place when a mobile engages in a packet data session set up:
Assuming assured mode subscription, the mobile sends a QoS BLOB to the PDSN. It contains
QoS requirements for the air interface for the forward and reverse links.
PDSN checks QoS requirements against the QoS profile received from the HAAA during
registration and sets up GRE tunnels to the PCF with appropriate parameters. (The GRE tunnel is
likely to be extended to the BSC [A8/A9 interface] in the future and some QoS mechanisms are
to be implemented in the A-bis interface between the BTS and the BSC but that is not yet a part
of the standards). PDSN may impose admission control if 1) the number of A10-A11 connections
requested exceeds the number allowed by the subscribers QoS profile, 2) the requested service
option is not a part of the QoS profile, or 3) PDSN does not have enough resources.
The mobile marks the packets with DSCPs. Before sending it out to external networks, the PDSN
will check that they are in conformance with the subscriber QoS profile. If the HAAA does not
provide QoS profile, PDSN will mark packets with default values per policies of the operator. If
the mobile station markings are not allowed by the QoS profile, PDSN will mark the packets
according to service providers local policies. PDSN may also mark the packets if the mobile
does not have the capability of marking the packets.
In the forward direction (toward the mobile), PDSN will match flows with appropriate TFTs and
send them along appropriate GRE tunnels.
As already indicated, QoS mechanisms in A8/A9 and A-bis interfaces are yet to be worked out in the
standards. Until then, over-provisioning or proprietary mechanisms will be used to provide QoS to end users.
Details of implementing QoS in GRE tunnels and in RLP connections are a subject of discussions in the
standards organizations today.
QoS in 802.11x Networks
The previous sections dealt with 3GPP and 3GPP2 standards, where the multiple layers of the
communication stacks within network elements as well as the inter-relationship of elements have been
standardized. This section takes a brief look into the 802.11 networks where things are markedly
different. 802.11 defines only two elements, the wireless access point (AP) bringing traffic into the
network, the station (STA) or user equipment and their interaction with one another. The interaction is
limited to the two lowest layers: media access control (MAC) and physical (PHY). The interaction is
classified into two groups: infrastructure mode and ad-hoc mode. Only the infrastructure mode (one AP,
many STAs) is dealt with here rather than the ad-hoc mode (no APs and many STAs). The physical
channels behavior, or rather the qualities that are brought forth, is determined by the design of the
physical layer. For example, with CDMA one needs to use a rake receiver, and with TDMA an equalizer
is needed. The physical medium remains the same, but the signal processing changes simply because the
physical algorithms change. Hence, in place of looking at the variegated PHY in analyzing the interrelationship of network elements, this section only focuses on the MAC layer. Below the MAC there are
only technology specific behaviors, and this is beyond the scope this paper.
As a standard, 802.11 was first introduced in 1997 and featured both frequency-hopping and directsequence modulation techniques for use with the wireless LAN (WLAN) environment. 802.11 originally
defined only the PHY and MAC layer, and did not address any QoS issues. The latter extensions to the
standardsa, b, e, g, i, etc.came later and address improvements or specific features. 802.11a was the
436
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
second standard (after the original 802.11) but the third wave of products (after 802.11 and 802.11b),
while 802.11b was the third standard but the second wave of products. Both 802.11a and 802.11b
improved upon 802.11s original data rates. 802.11e is the QoS extension of the 802.11 networks and
specifically addresses MAC layer extensions of the 802.11 MAC layer.
In order to understand the 802.11e extensions to the MAC for the support of QoS, the MAC of the 802.11
devices is explained briefly. 802.11 MAC layer defines the following terminology for access mechanisms:
Interframe Spacing
Once the data to be transmitted by an STA has been sent (i.e. the frame has been transmitted), all the
STAs in the network wait a certain amount of time before attempting to transmit. This duration of the
waiting period is called an inter-frame space. There are four types of interframe spacings: short
interframe space (SIFS), point coordination function (PCF), interframe space (PIFS), distributed
coordination function (DCF) interframe spacing (DIFS) and the extended interframe spacing (EIFS).
Figure 11 shows the relative sizes of the spacings within which a rudimentary QoS concept has been
built. The duration of the wait is dictated by the priority of the transmission, thus there are different
spacings.
o
SIFS is used for the highest-priority transmissions, such as request to send or clear to send
(802.11 uses carrier sense multiple access/collision avoidance (CSMA/CA) mechanisms for
medium access) frames and positive acknowledgments for the automatic repeat request
mechanism. High-priority transmissions can begin once the SIFS has elapsed. Once these
high-priority transmissions begin, the medium becomes busy, so frames transmitted after the
SIFS has elapsed have priority over frames that can be transmitted only after longer intervals.
PIFS is used by the PCF during a 802.11 access mode called the contention-free operation.
During this period, the AP acts as a point coordinator, polling the STAs for upstream and
downstream traffic. Stations with data to transmit in the contention-free period can transmit
after the PIFS has elapsed and preempt any contention-based traffic.
DIFS is the minimum medium idle time for contention-based services. Stations may have
immediate access to the medium if it has been free for a period longer than the DIFS.
EIFS is used only when there is an error in frame transmission and is not shown in the figure.
DIFS
PIFS
Busy
Medi u m
SIFS
Contention Window
Backoff
Window
Next Frame
Slot Time
Back-Off Window
This is the time the STA has to wait following collision detection, before the next transmission
attempt. If the medium were not idle, stations would defer to each other and employ an orderly
exponential back off algorithm to avoid collisions. After the frame has become idle, the interframe
spacing is allowed to elapse. After that, the contention window begins which defines the time for
437
which the STA will try to access the medium. The window is broken into slots and the size of the slot
depends on the PHY.
Contention Window
This determines the instant of transmission. The back off counter is randomly selected from [0, CW],
where CW is the contention window. Contention window sizes are always one less than a power of
two (e.g., 31, 63, 127, and 255).
Persistence Factor
It is not a part of 802.11 but was introduced in 802.11e. This paper introduces it for the sake of
clarity. It is a measure of how quickly a STA will attempt a re-transmission, based on priorities as
explained later. For 802.11, each time the retry counter increases, the contention window moves to
the next greatest power of two. Here, the factor two can be called the persistence factor, as will
become clear in the context of 802.11e.
802.11e specifies new elements into the MAC structure to support true QoS. The QoS classes that are
supported in 802.11 are defined as traffic categories (TC), best effort (zero), background (one), spare
(two), excellent (three), control lead (four), video >100ms latency (five), voice <100ms latency (six),
network control (seven). Here TC zero has the lowest quality and TC seven is guaranteed the highest
quality. The facilitation of such guarantees of quality has to do with the concepts already introduced
in the previous paragraphs: spacing, back-off, contention window, and persistence factor.
Legacy
backoff
(DIFS)
(15)
(2)
TC(0)
TC(7)
backoff
(AIFS(0))
(CW(0))
(PF(0))
backoff
(AIFS(7))
(CW7))
(PF(7))
Figure 12: Traffic Categories and Parameters for 11e QoS Realization
802.11e defines new variable quantities for the IFSs: arbitration interframe spacings (AIFS), which is
now a function of the TC, the concept being that when the channel goes idle, the STA with the most
urgent transmission, that is, the one having the highest TC, will begin transmitting first. The AIFS is
designed to be greater than or equal to the DIFS of the original 802.11 MAC, so that the legacy
systems do not quite play the game with the same rules. They always wait the least time, because one
is never sure how important their transmissions may be. This contention window is a function of the
TC, and the higher the TC the longer the STA will try to contend for the air, because it has something
important to communicate. The persistence factor (PF) is a measure of an STAs perseverance. The
438
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
persistence factor in legacy systems is always two, which means if a collision is detected, the STA
selects the transmission slot from the window of size n units. In case of subsequent collisions, the
STA selects the subsequent attempts from contention window sizes of 2n units, 4n units, 8n units, and
so on. Now the PF becomes a function of the TC, which means that persisting harder is possible when
there is something important to be communicated. If PF is two, the STAs contentions window
doubles with every failure, but when PF equals three, the CW triples with every failure. Thus, the
probability of an STA with PF equal to two getting access to the medium is higher than an STA with
PF equal to three, since the actual slot for transmission is selected randomly from [0, CW].
With these mechanisms in place, the 802.11 WLAN is now capable of translating whatever higher layer
QoS requirements one may impose on it to the Layer 2 or MAC. Since the 802.11 standards body does
not specify anything above Layer 2, it becomes the wireless service providers job to integrate the higher
layers onto the 802.11 STAs so that they may satisfy some desired QoS differentiation across users, and
over the different hybrid networks that the wireless session protocol (WSP) may decide to implement,
WLAN+3GPP, WLAN+3GPP2, etc.
3GPP/3GPP2 and 802.11 Integration
In integrating WLAN with 3GPP or 3GPP2, there are two paths the WSP can take. As already seen, the
QoS mechanisms relationship with the network elements (GGSN, PDSN/SGSN, PCF) is rather
analogous between 3GPP and 3GPP2. In this light, this section looks at the exemplary integration of
3GPP and WLAN. There are two ways the integration may take place, publicly integrated and privately
integrated, as shown in Figure 13.
GGSN
802.11
Gateway+FA
Internet
802.11
Hot-Spot
SGSN
RNC
802.11
Access Points
Publicly Integrated
SGSN
IWF
802.11
Gatew
ay
NodeB
3G Wireless
Access
NodeB
802.11
Hot-Spot
802.11
Access Points
Privately Integrated
Figure 13: Two Options for Integration: Publicly Integrated and Privately Integrated
For privately integrated architecture, an interworking function (IWF) is introduced between the AP and
the GGSN, and the IWF effectively acts as a SGSN. The benefits here are seamless handoffs, no need to
deploy new elements in the 3GPP network, and that QoS can be guaranteed across both networks
irrespective of where the calls are originated or terminated, since the operator owns the entire network.
However, an issue arises in that 802.11 traffic is directly injected into 3G WSPs access or CN, possibly
violating security of the networks (see following subsection). The 802.11 gateway/IWF, which is to be
developed and defined, may be complex and expensive. The 3G WSP may need to own and deploy
802.11 access points which may add to their operating expenses.
For the publicly integrated architecture, the WLAN and the 3G network are connected through a public
internet. The benefits of this method include the complete separation of networks, session hand-off
through mobile IP between the networks, minimal modifications to the 3G network, and full exploitation
of low-cost 802.11 technology. However, the issue with such integration is that there is a need for mobileIP software, a foreign-agent, a home-agent, and QoS is not guaranteed any longer, as a third party internet
connection is involved.
439
The corresponding highlights in 802.11 enhanced security, also known as 802.11i, are as follows:
In short many issues plagued the 802.11 communities, including some that have plagued the 3G
community as well, and many these issues are still not resolved in the 2G networks, namely the one-way
authentication that is performed in the 2G networks nowadays. In fact, the presence of legacy equipment
in an upgraded network is usually the element that opens up the entire network to security faults. As such,
the issues plaguing the 802.11 networks will remain as long as legacy equipment is maintained, and the
same is true for GSM/UMTS or any 2G/3G network. The 802.11i standard has elevated the WLAN
security to that of 3GPP levels. The security aspects that have been added by 802.11i include 802.1X
framework added to 802.11 making the security structure very flexible possibly accommodating SIMbased or NAI-based authentication mechanisms. One particular aspect of the 3GPP standards body is the
re-use of the USIM to authenticate a device in both the UMTS and the WLAN environment, thus
facilitating a uniform security mechanism and, resultantly, easier convergence of technologies [15]. It is
important to implement comparable security mechanisms in both 802.11 and 3G networks for integrated
networks to provide uniform user experience.
This flexibility allows the WSPs to select a variety of extensible authentication techniques permitting
authentication to be done at differing levels: device, user, application, etc. Hence the 802.11i security
framework allows for flexibility at the protocol level and degree of authentication as well.
Conclusion
In this paper, QoS mechanisms in UMTS, CDMA2000 and 802.11 networks have been presented. The
key issues in the three different broadband wireless access technologies have been highlighted. UMTS
standards specify a comprehensive framework for providing QoS from UE to the core. With the
440
Mohamed El-Sayed, Amit Mukhopadhyay, Zulfiquar Sayeed, Dong Sun, and Carlos Urrutia-Valds
introduction of IMS in Rel 5, an additional layer of QoS management has been added with the SBLP and
the PDF/PEP functionalities. CDMA2000 networks are also evolving to include QoS mechanisms across
all appropriate interfaces in the network (e.g from the UE to the core network). 802.11 networks are also
adapting to allow operators to set the QoS levels for Wi-Fi networks and this is important for integrating
3G and 802.11 networks.
Common problems across all these technologies are as follows.
First, QoS for delay sensitive services like VoIP require small packet sizes which results in relatively
higher overhead to payload ratio. Different algorithms for overhead compression and stripping are hot
topics of discussions in both 3GPP and 3GPP2.
Second, strict delay requirements exist in packet-backbone networks for delay-sensitive applications,
which were less of an issue in circuit-based backbone networks. Multi-protocol label switching traffic
engineering (MPLS-TE) is being proposed as the solution to meet these requirements.
Third, there is a current vision of plugging IP base stations to generic IP networks that may carry all types
of traffic of diverse QoS requirements. BSC and RNCs, PCF and SGSNs and PDSN and GGSNs may be
connected to these networks. In order to meet stringent QoS requirements of not only real time services
but also signaling and network management traffic, sophisticated mechanisms like DiffServ in
conjunction with MPLS may have to be extended to the edge of the network.
Fourth, appropriate QoS must be ensured as people roam across different access technologies and
different service providers' networks. Different interworking capabilities would need to be implemented
across network boundaries to allow this seamless roaming [21].
There are other aspects of QoS that have not been covered in this paper such as network reliability,
customer service, subscriber provisioning, charging, network design, and operations [18]. All these
encompass different aspects of QoS as perceived by subscribers which would need to be addressed by the
operator. Finally, it is important to emphasize the economic impact of being able to differentiate services
across a network as it gives operators the ability to set different pricing models depending on the level of
QoS provided, which in turn can bring in additional revenue and customer satisfaction [19].
References
[1]
ITU G.107, The E-model, a computational model for use in transmission planning, March 2003.
[2]
[3]
ITU P.800, Methods for objective and subjective assessment of quality, August 1996.
[4]
3rd Generation Partnership Project, Network Architecture (Release 5), 3GPP TS 23.002, Dec.
2002.
[5]
3rd Generation Partnership Project, Network Architecture (Release 6), 3GPP TS 23.002, Dec.
2002.
[6]
3rd Generation Partnership Project, Quality of Service (QoS) Concept and Architecture (Release
5), 3GPP TS 23.107, Mar. 2003.
441
[7]
3rd Generation Partnership Project, End-to-End Quality of Service (QoS) Concept and
Architecture (Release 5), 3GPP TS 23.207, Mar. 2003
[8]
3rd Generation Partnership Project, IP Multimedia Subsystem (IMS) Stage 2, 3GPP TS 23.228,
Dec. 2002.
[9]
3rd Generation Partnership Project, Signaling Flows for the IP multimedia call control based on
SIP and SDP; Stage 3 (Release 5), 3GPP TS 24.228, Mar. 2004.
[10] 3rd Generation Partnership Project, Policy Control over Go Interface, 3GPP TS 29.207, Mar.
2003.
[11] D. Durham, J. Boyle, R. Cohen, S. Herzog, R. Rajan, and A. Sastry, The COPS (Common Open
Policy Service) Protocol, IETF RFC 2758, January 2000.
[12] M. Handley, V. Jacobson, SDP: Session Description Protocol, IETF 2327, April 1998.
[13] R. Yavatkar, D. Pendarakis, and R. Guerin, A Framework for Policy-based Admission Control,
IETC RFC 2753, January 2000.
[14] RFC 2475, An architecture for differentiated services, IETF, December 1998.
[15] 3GPP TS33.234 (Rel 6) Wireless Local Area Network (WLAN) Interworking Security
[16] 3GPP2 P.S0001-B 1, Wireless IP Network Standard, October 25, 2002.
[17] Telecommunications Industry Association, Data Service Options for Spread Spectrum Systems
Addendum 2: CDMA2000 High-Speed Packet Data Service Option, TIA/EIA-IS-707-A-2.12,
Aug. 2000
[18] ITU, E.800, Quality of Service, Network Management and Traffic Engineering, August 1994.
[19] SIP in 3G Wireless Networks: Service Models, Architecture, and Network Design, to appear in
Bell Labs Technical Journal, September 2004.
[20] ITU-T, X.805, Security Architecture for Systems Providing End-to-End Communications,
October 2003.
[21] A SIP Enabled AllIP Architecture for Converged Next Generation Networks, to appear in Bell
Labs Technical Journal, September 2004.
442
443
throughput and delay degradation. This may render unacceptable the performance of time-bounded, highrate applications such as audio/video streaming and voice over Internet protocol (VoIP).
Up to now, the 802.11 wireless networking standard [2] provides best-effort services, while it establishes
a minimum of QoS support under the point coordination function (PCF) channel access function. A
number of previous works [36] have focused on developing additional QoS mechanisms on the legacy
802.11 specification. Toward that aim, the IEEE Task Group E (TGE) was formed in July 1999, in order
to define a standard set of QoS enhancements to the medium access control (MAC). The TGE has already
defined two fundamental wireless channel access methods for contention and contention-free based QoS
access respectively, as enhancements to the basic access functions defined by the 802.11 standard.
Moreover, given the increased market interest on QoS support and based on these two enhanced QoS
methods, the Wireless Fidelity Alliance (WFA) has recently introduced the wireless multimedia
enhancements (WME) [7] and the wireless scheduled multimedia (WSM) [8] specifications, as
prestandard subsets of the 802.11e draft supplement, in order to avoid possible market fragmentation
caused by multiple, noninteroperable prestandard subsets of the draft 802.11e standard.
In this work, we intend to evaluate and compare the QoS performance of the two channel access methods
defined by the 802.11e draft standard for high-rate, time-critical applications. As it will be shown in the
following paragraphs, the contention-based and contention-free nature of these access methods introduces
significant differences on the QoS performance, mainly in terms of the achieved throughput and packet
delay, which represent the most essential parameters for servicing real-time applications over WLANs
(e.g., audio/video streaming and VoIP). The paper is organized as following: In Section 2, a brief
overview of the 802.11 specification is provided as a basis for describing the 802.11e MAC
enhancements for QoS support in Section 3. Section 4 outlines the simulation methodology and the
testing parameters and criteria used for evaluating the QoS performance. In Section 5, the results obtained
by the simulations are presented, leading to the conclusions summarized in Section 6.
2. Overview of 802.11 Technology
The 802.11 standard define two medium access methods: the mandatory distribution coordination
function (DCF) and the optional point coordination function (PCF). The DCF operates during the
contention period (CP) as a listen-before-talk mechanism that uses a carrier sense mechanism access with
collision avoidance (CSMA/CA) for accessing the wireless medium.
Under DCF rules, the decision for accessing the medium is individually taken by the wireless stations (STAs).
More specifically, each STA senses the medium for at least a DCF interframe space (DIFS). If it is idle, the
STA additionally waits for a randomly selected time called back-off time, by decreasing a value B by one at
every time slot interval. The back-off value B is randomly chosen in the interval [0, CW), where CW is the
contention window with values in the range CWminCWCWmax, where CWmin and CWmax are constants
defined by the protocol. If B=0, a frame transmission is initiated. Upon a successful transmission, a new backoff value is selected and the contention window is set to CWmin, otherwise the CW value is doubled up to the
CWmax value. On the other hand, if the medium gets busy while a STA is decreasing the backoff timer, the
backoff procedure is paused and is resumed after the medium is sensed to be idle for DIFS.
Because more than one STA may concurrently gain access to the medium, collisions may occur. Upon a
collision detection obtained by an acknowledgment (ACK) scheme, the CW value is doubled up to the
maximum value CWmax and a new backoff value is selected.
While DCF is used for accessing the wireless medium during the CP, PCF is used within the contentionfree period (CFP). During the CFP, data transmissions are coordinated by the point coordinator (PC),
located into the access point (AP), which has higher access priority, as it detects the medium status for
444
Andreas Floros, Ph.D., Theodore Karoubalis, Ph.D., and Stylianos Koutroubinas, Ph.D.
PCF interframe space (PIFS), with PIFS<DIFS. The PC polls the STAs, until the CFP expires. Upon a
poll reception, a STA may transmit a single pending frame.
Although a number of PCF polling schemes have been proposed for application-specific bandwidthmanagement optimizations [9, 10], PCF raises a number of problems, such as unpredictable beacon
delays and unknown transmission durations [11, 12], which render it insufficient for practical QoS
implementations. As was previously mentioned, in order to overcome this problem, the IEEE802.11 TGE
is currently developing QoS enhancements for 802.11 wireless networks.
3. 802.11e QoS Mechanisms
According to the latest 802.11e draft specification [13], the 802.11e MAC layer includes both DCF and
PCF channel access functions introduced in the 802.11 standard, as well as the hybrid coordination
function (HCF) for QoS support, which consists of two access functions:
The hybrid controlled channel access (HCCA) for parameterized, controlled access
The enhanced distributed channel access (EDCA), for differentiated, distributed access to the
wireless medium using different levels of priorities
Both access methods result into specific time-length transmission opportunities (TXOPs) used by the
QoS stations (QSTAs) for data transmissions. A brief overview of both methods is provided in the
following paragraphs.
3.1. EDCA Overview
EDCA contention-based channel access function achieves traffic differentiation by mapping eight user
priorities to four access categories (ACs), which can be considered as independent backoff DCFlike
entities. Each backoff entity contends for medium access using a specific set of EDCA parameters (e.g.,
interframe space AIFS, contention window CW, maximum transmission length TXOP limit) under
the legacy DCF rules. The EDCA parameters are defined and announced by the hybrid coordinator (HC).
For AC differentiation, it is very important that the same values are used by all backoff entities that
belong to a specific AC.
The default EDCA parameters values used by QSTAs as defined in [13] appear in Table 1. The four
discrete ACs are labeled AC_VO (Voice), AC_VI (Video), AC_BK (Background), and AC_BE (Best
effort). An EDCA TXOP is granted to an AC, provided that it detected the medium to be idle for
arbitration interframe space AIFS[AC] = AIFSN[AC] x aSlotTime+SIFS (where aSlotTime and SIFS are
time intervals defined in [2]) and that its backoff counter B[AC] is zero. The backoff counter value
B[AC] is randomly chosen following the normal DCF rules described in Section 2, using the modified
CWmin and CWmax values also shown in Table 1.
TXOPLimit
AC
CWmin
CWmax
AIFSN
802.11b
802.11a/g
AC_BK
31
1023
AC_BE
31
1023
AC_VI
15
31
6.016 ms
3.008 ms
AC_VO
15
3.264 ms
1.504 ms
445
As TXOP continuation is optionally allowed under EDCA, more than one packet may be transmitted
within a single TXOP. The maximum allowed TXOP duration, however, is restricted by the EDCA
parameter TXOPLimit[AC]. A collision between two or more ACs belonging to different QSTAs is
treated as in legacy DCF. On the other hand, in the case that two or more ACs within the same QSTA
gain access to the medium, an internal collision occurs, and the TXOP is granted to the highest
priority AC.
According to [13], the EDCAbased transmissions may be subject to certain channel access
restrictions in the form of an admission control mechanism, which provides guarantees on the amount
of time an admitted traffic will access the wireless medium. These requirements are defined by the
traffic stream upon admission in a traffic specification (TSPEC) element. Generally, the admission
control mechanism is applied on a) the QAP side, providing guarantees on the amount of time an
admitted traffic will access the wireless medium and b) the QSTA side, for monitoring whether the
AC transmissions have exceeded the admitted time. Although the minimum functionality of the
EDCA admission control is analytically defined in [13], the employment of advanced, adaptive
admission schemes [14] can preserve the allocated EDCA access guarantees even under
unpredictable, time varying channel conditions.
3.2. HCCA Overview
HCCA may be initiated by the HC during both the contention free (CFP) and contention period (CP) in
order to meet the QoS requirements of particular traffic streams (TS) described in detail by appropriate
TSPEC elements. Using HCCA, TXOP allocation is performed by the HC through a polling mechanism.
More specifically, during the CFP, the HC allocates TXOPs to pollable QSTAs using specific QoS CF
poll frames or immediately transmits downlink data packets, according to a service schedule. The QSTAs
should make no attempt to access the medium unless they receive a QoS CFpoll frame, hence no
collisions occur within the CFP. On the other hand, during the CP, if the HC detects the medium to be
idle for PIFS, it may send a QoS CFpoll frame to a pollable QSTA without backing off. This ensures
that the HC has higher priority than EDCA and legacy DCF access. In both cases (CFP and CP), during a
polling TXOP, a polled QSTA may initiate multiple packet transmissions, separated by SIFS, provided
that the total transmission time does not exceed the TXOPLimit announced with the corresponding CF
poll frame.
The service schedule (e.g., the polling time instances and the maximum allowed transmission lengths) is
calculated by the HC, taking into account the time requirements defined by the TSPEC values sent to the
HC from the requesting QSTAs. Upon a TS service request, the HC must decide whether this stream can
be serviced under the QoS requirements described in the TSPEC element. This decision is taken by the
admission control unit (ACU), and it is announced to the requesting QSTA through an HC response. If
the admission is successful, the scheduler is responsible for granting channel access to this TS based on
the negotiated TSPEC parameters. This guaranteed channel access is offered in terms of TXOPs that
satisfy the service schedule.
An HCCA service scheduler reference design (termed as simple scheduler) is defined in [13]. The simple
scheduler uses the mandatory set of TSPEC parameters (mean data rate, nominal MAC service data unit
[MSDU] size and maximum service interval) to allocate TXOPs to the i-th QSTA with time-length equal to
SIxpi Li
M
TXOPi = max
+ O, + O
L R
Ri
i i
where p is the TSPEC mean data rate in bps, L and M the nominal and maximum allowable MSDU size
(in bits) respectively and R the physical transmission rate negotiated in the TSPEC. The overhead value O
446
(1)
Andreas Floros, Ph.D., Theodore Karoubalis, Ph.D., and Stylianos Koutroubinas, Ph.D.
(in seconds) represents the overhead introduced by the MAC layer including interframe spaces,
acknowledgments (ACKs) and CFpolls and should be calculated for every TS. The parameter SI
represents the scheduled service interval calculated as the first sub multiple of the beacon interval that is
less than the minimum of all maximum service intervals for all admitted streams. Using the simple
scheduler, the TXOP starting time is constant within each SI, as the polling operation is performed in a
serial manner. An example is illustrated in Figure 1, where SI=50ms and 2 polled STAs with 1 TS each
are considered.
TXOP1 TXOP2
SI=50ms
TXOP1 TXOP2
SI=50ms
447
TSPEC Parameter
Value
Traffic Type
TSID
Direction
Access Policy
TSInfo
Aggregation
APSD
User Priority
TSInfo Ack Policy
Schedule
Nominal MSDU Size
Maximum MSDU Size
Minimum Service Interval
Maximum Service Interval
Inactivity Interval
Suspension Interval
Service Start Time
Minimum Data Rate
Mean Data Rate
Peak Data Rate
Maximum Burst Size
Delay Bound
Minimum PHY Rate
Surplus Bandwidth Allowance
Medium Time
Periodic
Stream Specific
Uplink
HCCA or EDCA
Node Dependent
No APSD
5
Normal
0 (reserved)
1364 octets
same as Nominal MSDU size
0 msec
16 msec
1 minute
same as Inactivity Interval
0 (unspecified)
0 (unspecified)
6 Mbps
0 (unspecified)
0 (unspecified)
64 msec
same as PHY rate employed
1.2
0 (reserved)
6
1
1
6
12
2
1
12
24
4
1
24
36
6
1
36
48
8
1
48
54
9
1
54
Table 3: Number of QSTAs and SDTV Traffic Streams versus Transmission PHY Rate
For EDCA simulations, the EDCA parameters appeared in Table 1 were used, while TXOP continuation
was also separately considered. EDCA parameter changes were not allowed. The EDCA admission
control algorithm defined by the 802.11 draft standard was employed, with the exception that the TSPEC
service requests were always accepted. On the other hand, for HCCA testing, the simple scheduler
described in Section 3.2 and the admission control unit reference design introduced in [13] were
employed. It must be also noted that during HCCA simulation, no EDCA data transmissions were
allowed, as the TSPEC access-policy parameter value was set to HCCA or EDCA.
4.1. Test Criteria and Measurements
The QoS performance of the 802.11e channel access methods can be evaluated by the end-to-end
throughput and packet delay measurement. In this work, the mean total throughput was measured for all
uplink traffic streams activated within the QoS BSS. Moreover, the packet delay measurement was also
obtained as the time interval elapsed between the insertion of a data frame to a transmission queue and the
time instance of its successful reception by the remote wireless receiver (in our case the QAP).
Although the throughput and packet delay measurements represent the most essential QoS parameters, a
number of additional measurements were obtained in order to derive a clear assessment of the overall
QoS performance achieved by the 802.11e QoS access methods. In the case of EDCA simulations, the
(%) medium collisions ratio was obtained as the ratio of total number of the collisions occurred to the
448
Andreas Floros, Ph.D., Theodore Karoubalis, Ph.D., and Stylianos Koutroubinas, Ph.D.
total number of the EDCA TXOPs obtained by the QSTAs during a simulation. Moreover, as EDCA
provides no strict guarantees for service times, the number of the TSPEC maximum service interval
(MaxSI) violations was measured. A MaxSI violation occurs when the time interval between two
consecutive EDCA TXOPs won by a QSTA for a specific TS transmission is greater than the
corresponding TSPEC maximum service interval parameter. The (%) MaxSI violation ratio was also
calculated as the ratio of the MaxSI violations to the total number of TXOPs won by the QSTAs.
On the other hand, in the HCCA access case, the simple scheduler derives a service interval that is always
less than the minimum of maximum service interval values for all polled TSs. Hence, provided that
concurrent EDCA data access is not allowed, MaxSI violations cannot occur during CFP. However, when
an HCCA TXOP is granted to a QSTA, under certain conditions (e.g., empty transmission queues) a
portion of the total TXOP granted length might be not used for transmissions, resulting into TXOP losses.
Normally, this unused TXOP length is returned back to the QAP, but using the simple scheduler prohibits
any reallocation of the returned bandwidth, due to the strict scheduled service timing. Thus, in the case of
the simple scheduler employment, the originally allocated bandwidth for data transmissions (calculated as
the summation of the achieved throughput plus the percentage of the bandwidth lost by no-transmissions
within HCCA TXOPs) is usually higher than the actually measured throughput value.
5. Results
5.1. EDCA Results
In Table 4, the EDCArelated measurements obtained without TXOP continuation are shown as a
function of the transmission PHY rate. From this table it is clear that the following is true:
The achieved throughput is much lower than the theoretical one in all test cases and this
difference is increased with the PHY rate employed. Under nonideal wireless medium conditions,
it is obvious that the throughput will be lower, due to the bandwidth allocation for packet
retransmissions.
The mean packet delay increases with the PHY rate, due to the increment of the number of
QSTAs within the BSS. It is significant to note that for high PHY rates (>36Mbps) the presence
of eight or more contending QSTAs renders then mean packet delay value greater than the
TSPEC maximum service interval value (which equals to 16 msec). No MaxSI violations occur,
however, as each QSTA wins EDCA TXOPs before the maximum service interval expires.
The number of the won TXOPs is increased with the PHY rate, due to the increment of the
number of the contending QSTAs and the total traffic load appeared into the WLAN. Under
strong contention, however, additional medium collisions are raised, and the corresponding (%)
medium collision ratio is also increased.
449
PHY Rate
6 Mbps
Mean throughput (Mbps)
3,5
Mean packet delay (ms)
10,07
Total number of TXOPs
15182
Medium collisions
5
(%) Medium collisions
0,03%
MaxSI violations
0
Mean MaxSI violation length (ms)
0
(%) MaxSI violations
0
12 Mbps
6,4
11,19
27793
91
0,32%
0
0
0
24 Mbps
10,9
13,16
47555
524
1,10%
0
0
0
36 Mbps
14,3
15,12
62325
2161
3,46%
0
0
0
48 Mbps
16,9
17,16
73388
3420
4,66%
0
0
0
54 Mbps
17,9
18,21
77835
4413
5,66%
0
0
0
12 Mbps
9,4
5,54
41103
14
0,04%
41
0,173
0,1%
24 Mbps
18,9
4,74
78945
97
0,12%
1092
1,957
1,38%
450
In the case that PHY = 6 Mbps, the SDTV stream is not admitted by the admission control unit,
hence no traffic is serviced, as the traffic speed is equal to the theoretical offered bandwidth and
no service schedule can be calculated. The same is observed for higher PHY rate values, as in
every case one SDTV stream in always not admitted.
The measured throughput is lower than the theoretical one. Taking into account that in all cases,
however, an SDTV TS is not admitted and that significant TXOP losses are raised due to the
simple scheduler employment, the originally allocated bandwidth for data transmissions is higher
Andreas Floros, Ph.D., Theodore Karoubalis, Ph.D., and Stylianos Koutroubinas, Ph.D.
than the measured mean throughput value. It is expected that the development of enhanced
HCCA scheduling algorithms will reduce the TXOP losses, providing optimized and efficient
bandwidth management and utilization.
The mean packet delay is also low, presenting a slight increment with the PHY rate, due to the
presence of more contending QSTAs within the BSS.
Under HCCA, no MaxSI violations occurred, as all the polling instances pre-calculation is based
on the TSPECs maximum service interval value.
PHY Rate
6 Mbps
Measured throughput (Mbps)
12 Mbps
5,99
3,12
1
0%
5,99
24 Mbps
13,2
4,48
3
19,7%
15.8
36 Mbps
22,798
4,45
5
19,6%
27,26
48 Mbps
27,598
5,55
7
29,2%
35.65
54 Mbps
28,798
5,47
8
23,8%
35.65
40
35
30
25
20
15
HCCA
10
5
0
6Mbps
12Mbps
24Mbps
36Mbps
48Mbps
54Mbps
451
20
18
16
14
EDCA (no continuation)
12
10
HCCA
6
4
2
0
6Mbps
12Mbps
24Mbps
36Mbps
48Mbps
54Mbps
452
Andreas Floros, Ph.D., Theodore Karoubalis, Ph.D., and Stylianos Koutroubinas, Ph.D.
Using a sequence of simulations, it was found that EDCA is inconvenient for real-time transmission of
high-rate data streams, as it introduces large packet delay values, especially under heavy traffic
congestion, while the achieved throughput is also low. In particular, under heavy traffic load, the
contention nature of EDCA cannot provide certain throughput and latency guarantees. The development
of optimized admission control algorithms may partially overcome this problem, although it is expected
that such algorithms will significantly reduce the amount of available bandwidth that will be allocated to
the admission control unit for traffic service, in order to avoid possible network congestion that may lead
to QoS guarantees violation. It is obvious that such an approach will directly affect the number of the
traffic streams that will be able to be admitted for service.
In order to avoid limited admission channel capacity, EDCA TXOP continuation is defined in the 802.11e
specification as an optional protocol feature. TXOP continuation overcomes the effects induced by the
EDCA high contention by allowing multiple frame transmissions within a single EDCA TXOP. As the
consecutive frame transmissions are performed within the same TXOP and are separated only by SIFS
time, the contention is lower than in the case of simple EDCA, while the effective bandwidth allocated for
transmissions and the corresponding admission capacity is increased. Hence TXOP continuation achieves
high throughput values and low packet delay values, while in the presence of an admission control
mechanism, and more traffic load can be efficiently serviced. The expense for that is the occurrence of
maximum service interval violations, which depend on the maximum allowed TXOP length defined by
the QAP.
On the other hand, the bandwidth reservation scheme applied by the HCCA method, significantly
improves the overall QoS performance, as it provides high bandwidth allocation with very low packet
delays (lower than in the case of EDCA with TXOP continuation). The service scheduler type is a
significant parameter that affects the QoS performance as, due to the strict nature of bandwidth
reservation, may result into significant TXOP losses (that is bandwidth not used for data transmissions)
and overall throughput degradation. The development of optimized, enhanced schedulers can dramatically
improve the HCCA QoS performance and allow real-time streaming of bandwidth-demanding
applications over the wireless medium. Moreover, the strict TXOPallocation nature of HCCA provides
better-bandwidth management capabilities if advanced, optional protocol features are employed, such as
the direct link protocol for direct QoS link establishment.
7. References
[1]
[2]
[3]
D-J. Deng and R-S. Chang, A priority scheme for IEEE 802.11 DCF access method, IEICE
Transactions on Communications, Vol. E82-B, No. 1, January 1999.
[4]
N. H. Vaidya, P. Bahl, and S. Gupta, Distributed fair scheduling in a wireless LAN, Sixth Annual
International Conference on Mobile Computing and Networking, Boston, August 2000.
453
[5]
[6]
[7]
Wi-Fi Alliance Technical Committee, Quality of Service (QoS) Task Group, Wireless Multimedia
Enhancements (WME), March 2004.
[8]
Wi-Fi Alliance Technical Committee, Quality of Service (QoS) Task Group, Wi-Fi Scheduled
Multimedia (WSM) specification, Version 0.09, June 2004.
[9]
E. Ziouva and T. Antonakopoulos, A Dynamically Adaptable Polling Scheme for Voice Support in
IEEE 802.11 Networks, Computer Communications, Vol. 26, No. 2, February 2003, pp. 129142.
[10] C. Coutras, S. Gupta, and N. B. Shroff, Scheduling of Real-Time Traffic in IEEE 802.11 Wireless
LAN, Wireless Networks, Vol. 6, No. 6, December 2000, pp. 457466.
[11] S. Mangold, S. Choi, G. R. Hiertz, O. Klein, and B. Walke, Analysis of IEEE 802.11e for QoS
Support in Wireless LANs, IEEE Wireless Communications, Vol. 10, December 2003, pp. 212.
[12] S. Mangold, S. Choi, P. May, O. Klein, G. Hiertz, and L. Stibor, IEEE 802.11e Wireless LAN for
Quality of Service, Proc. European Wireless 02, Florence, February 2002.
[13] IEEE802.11 WG, IEEE802.11e/D8.0, Draft Amendment to Standard for Information Technology
Telecommunications and Information Exchange between Systems LAN/MAN Specific
Requirements Part 11: Wireless Medium Access Control (MAC) and Physical Layer (PHY)
Specifications: Medium Access Control (MAC) Quality of Service (QoS) Enhancements,
February 2004.
[14] A. Floros and T. Karoubalis, CARMEN: Mechanisms for enhancing performance and bandwidth
management of wireless contention based quality of service, U.S patent application, ATMEL
Hellas S.A., Multimedia & Communications Group, July 2004.
[15] A. Floros and T. Karoubalis, Delivering high-quality audio over WLANs, Audio Engineering
Society 116th Convention, Berlin, May 811, preprint 5996.
454
T. Jensen, Ph.D.
Telenor R&D, Norway
H. Kjnsberg, Ph.D.
Telenor R&D, Norway
Abstract
Considering the growing complexity of services and network equipment to manage, it becomes vital to
have an effective framework for dealing with key performance indicators (KPIs). This paper presents
such a KPI framework, with main focus on end-to-end quality of service (QoS). A general framework
should provide the opportunity to compare the performance of different services and operations across a
number of technologies and markets. Moreover, the KPIs should trigger actions when goals are not met.
The framework must cover the needs related to existing services as well as end user services of the future.
The paper discusses the different groups of receivers of KPIs, principles, and issues related to measuring
network and service parameters and the principles of aggregating measured parameters to produce KPIs
for the various receiver groups.
1. Introduction
Monitoring and resolving end-to-end performance in wireless telecommunications networks is a task of
increasing importance. The rapid growth in the number of subscribers, the introduction of a range of new
and challenging services and applications, and the increasing number of interacting technologies are all
elements that increase the complexity of the system and the need to monitor and follow the evolution of
its performance. The end users degree of satisfaction is Alpha and Omega for providers, naturally also
affecting equipment vendors and others involved. Considering increasing competition it is more important
than ever to be able to measure and control the service quality as perceived by the end users. This point of
view is reflected in works like for instance [i], [ii].
A main objective of this paper is to describe KPIs for end-to-end quality of wireless services. The principles of
the KPI framework, however, can be applied for other services and network operations as well.
Motivations for a KPI framework are to follow up performance effectively, allow comparisons across markets
and technologies, check that defined target values are met, and last but not the least trigger actions whenever
455
goals are not met. In order to devise a KPI framework, one must investigate which types of KPIs are needed,
and present ideas and principles for how to produce these KPIs from measurements in the network as well as
how to build a hierarchy of KPIs efficiently, given a set of KPI receivers.
To limit the scope of this paper only technical indicators are treated. That is, other aspects, such as those
related to finance, reputation, and so forth, are not discussed in any detail. Moreover, the technical issues
related to the network are the main focus, leaving out for instance the end users experiences of the
providers customer-care department. That is not to say that such aspects are unimportant for the total
service quality as perceived by the end user but is merely a limiting of scope.
The paper is organized as follows: The next section presents the broad idea of a common measurement
platform serving the needs of all measurement-based operations in a modern communication network.
This is the main idea on which the KPI framework presented here is based. The subsequent section
presents the principles of this framework, followed by three sections discussing its main elements: KPI
receivers, performance measurements, and parameter aggregation. Finally, a short section of concluding
remarks is provided.
2. Common Measurement Platform
Many of the activities involved in the daily operation and future planning of a system require data on
which to base decisions. In order to produce these data, measurements must normally be performed.
System performance, accounting, and fault and security management are all areas within management of
modern communication networks that require measurements of some sort (measurement tasks). Also
business and marketing functions need network and service measurement data to make informed
decisions on future strategies.
In most of the current operations so far, the data needed by each of the measurement tasks are collected
by separate measurement arrangements, which side-by-side collect the same basis data. Historically, the
use of a range of tailor-made arrangements may have been legitimate, but a continued effort in this
direction is not the most beneficial.
In contrast to parallel arrangements, one single measurement platform could perform all the
measurements, and gather the results. Afterwards the collected measurement data can be post-processed
to fit each of the different measurement needs. Potential cost-efficiency gains, given that the measurement
data required from the different measurement tasks (partly) overlap, include the following:
The total amount of measurement observations needed for obtaining the desired level of
confidence (the data must be collected, processed, and transported to a data repository)
In the years to come, more detailed measurements will be needed by several measurement tasks. QoS and
service level agreements (SLAs) must be verified, and the focus on cost efficiency will put stronger demands
on the task of following network utilization. More advanced mechanisms will be needed to detect attacks on
services and infrastructure, including for instance real-time analysis of traffic patterns and packet contents. As
mobile services are growing into a pillar of society and the dependability requirements thereby increase,
measurement-based techniques for detecting faults even before they happen might be very interesting. We can
also foresee a tighter integration between management areas like for instance performance management, SLA
verification, and accounting. All in all, an increased need for detailed measurement data from the different
measurement tasks as well as a tighter integration between the different measurement tasks is foreseen. The
456
scaling issues with many standalone measurement systems, as compared to one single measurement platform,
will then become more apparent.
3. General Principles of KPI Framework
The framework should allow KPIs for service performance to be followed over time as well as in (nearly)
real time. The framework also should allow the observed KPI values (short term and long term) to be
compared to defined target values and thresholds, and if necessary trigger appropriate actions. Potentially
different target and threshold values can be assigned for the same service in different markets. Examples
of markets are customer groups (residential, enterprises) and operations in different geographical regions.
The KPI framework could also be used if different providers wanted to compare the technical service
quality of their operations. This should be allowed independently of technology used and dimensioning.
In such cases, the target values for individual operations could be normalized to 100 to reflect the
different quality acceptance levels in the different markets. Because a normalized target does not reflect
the actual performance, the original target values should also be kept. Combined with economic data like
operative expenditures (OPEX) and capital expenditures (CAPEX), the KPI framework could support
figures like service quality per expenditures as a measure of system and operations cost effectiveness.
Such benchmarking is an increasing task in todays wireless market.
The KPI framework roughly consists of three separate parts:
Defining which KPIs that are relevant to follow, and their target values, for various KPI receiver
groups
Defining how to produce a set of basis end-to-end performance indicators, either by using active
end-system measurements or passive logging in the network
Defining how the set of KPIs can be derived from the basis performance indicators (parameter
aggregation)
Figure 1 gives an overview of these three steps. A number of different people within a providers
organization are interested in KPIs for performance and service quality but with different needs regarding
Customer Care
Operations &
Maintenance
Business
Management
Service
Management
User groups
needs
for KPIs
Marketing
KPI F.1
KPI E.1
KPI D.1
KPI D.2
Network
Planning
Aggregation
into KPIs
KPI C.1
KPI B.1
KPI A.1
KPI A.2
- Delay
- Loss
- Utilisation
- Throughput
- Bit-error rate
-
Network
and service
measurements
level of detail and time scale. For each group of KPI receivers, it must be defined which KPIs they want
to follow, including granularity and time scale. For example, the top-level business-management group
may need only high-level information (KPI C.1), while the operations and maintenance group may need
very detailed information (KPIs A.1, A.2, and A.3). Each of the KPIs must have an associated target
value as well as threshold values that may trigger suitable actions if the observed values are below or
above the thresholds. Having more than one threshold value allows triggering of actions when service
levels are too high as well as too low. Potential actions include looking into the cost efficiency of the
network operation and possibly shift resources to other areas. The threshold values may differ for the
different services and potentially also for the different markets.
Figure 1 also shows a hierarchy that may aggregate basis performance indicators into more high-level
KPIs. A fundamental motivation for the aggregation procedures is to allow for deriving different KPI
results from the same set of underlying parameter measurements. The interested KPI receivers may
extract exactly the KPIs they need at one or more levels of the aggregation hierarchy. Ideally, one single
KPI at the top should give all relevant information on all the services and system quality.
Performance measurements in communication networks have often been limited to the perspective of
network elements, servers, or individual network segments. End-to-end performance and service quality
parameters are more rarely measured, normally using active measurements from clients. Often, this is not
a perfect method since the results in some cases are only representative for the probe traffic [iii]. Hence, a
main idea of the framework is that measurements in the network can be used to infer the end-to-end
service quality expected by an end user. These may be complemented by active tests to the extent found
reasonable. Deriving the relation between a series of network measurements, however, possibly combined
with active tests, and end-to-end service performance is not an easy task. Still, the potential economical
and technical benefits are greatalso adding to the motivation for devising a KPI framework.
4. KPI Receivers
Technical KPIs are of interest for a number of different actors: network operators, service and content
providers, end users, vendors, researchers, and regulators. These actors have different perspectives and
often need measurements of various types, levels of detail, and on various time scales. The network
operator is often one of the most demanding actors, because there is in general great variety in the
requests from the different departments and areas within a single company.
It is convenient to introduce the concept of KPI receiver groups, as in the following where the needs and
purposes of each group are described. The groups listed here are similar to those mentioned by [i] and [ii].
Top-Level Business Management
This receiver groups main responsibility is to guarantee the financial health of a service operation [i]. Report
examples listed in [ii] include reports on traffic profiles and session lengths, giving information on user
behavior and trends; the amount of usage of services, giving valuable input to further development and
deployment; SLA monitoring and evaluation for selecting peering networks; and reports on resource
availability, coverage, and utilization to support decisions on possible network expansions. These reports may
by some be viewed as examples of top-level KPIs but are in our framework rather considered as examples of
underlying data used to generate the latter. If they serve the role of underlying data, there should be drill-down
options connected to the (few) top-level KPIs. Common for all KPIs regularly reported to the top-level
business-management team is that the time scale should be long, say weeks or even months.
Marketing
This receiver group is responsible for pricing strategies, it should predict the effects of possible changes
in service portfolio, identify target groups for marketing campaigns, and also participate in predicting the
458
profitability of future investments. The KPI reports needed for the marketing department are often of the
same kind as the (detailed) reports for the top-level business management but the actions triggered are
different [ii]. For instance, reports on user behavior and service usage are important in connection to
pricing strategies, whereas SLA and system utilization reports can be used to identify campaign target
groups. The marketing department needs KPIs on a long time scale to evaluate trends but also shorter
time scales, say daily reports, can be useful.
Network Planning
This receiver group is responsible for ensuring that the network fulfils defined QoS requirements, given
the growth of user base and predicted usage patterns (taking into account changes in pricing strategies and
new services) [ii]. It needs reports that support long-term trend analysis, but there is often also a need for
reports on a smaller time scale, say daily. Examples of useful KPI reports are traffic distribution and
volumes, taking into account geographical segmentation as well as segmentation into different service
classes: alarm and network failure statistics, handover statistics, and availability reports.
Operations and Maintenance
This receiver groups main goal is to prevent network failures and to minimize the mean time to repair
failures, given that they occur. Hence update of KPI reports are needed on small time scales, on the order
of seconds and minutes. Examples of KPIs are fraction of dropped or blocked sessions, number of
handover failures, congestion metrics, the amount of lost data packets, delay reports, and interference
problems. KPI reports on SLA monitoring may also be of interest in order to identify problem areas
outside the providers own network.
Service Management
Each service manager is responsible for the performance of a specific service. The requested KPIs may
be similar to the KPIs of the operations and maintenance department but should be specific to the
service in question. Report updates need not be on the smallest time scales; hourly or daily reports are
commonly sufficient.
Customer Care
Customer care should be aware of network problems as well as service delivery problems to give answers
to customer complaints right away. Lists of cells or areas with the largest number of dropped sessions,
handover failures, interference, congestion and blocking will help in satisfying the end user, as will lists
of failing services. The reports need to be updated fairly often, at least on an hourly basis.
For a newly built system the total set of KPI reports will differ significantly from the set required for a
mature operation, as will the time scales for KPI updates. This is discussed in more detail in [ii].
Introduction of new services will also affect the update time scales and choice of KPI reports.
5. Performance Measurements
The measurement platform should collect and process performance data from a number of different
network, server and service/application components. Design of a measurement platform should include an
evaluation of which parameters that need to be measured, at what protocol layer the measurements should
be performed, the number and location of the measurement loggings points, and which measurement
method(s) to use. Particular care must be taken to ensure that the choice of observation periods and levels of
detail and accuracy in the measurements does not place unreasonable demands on the measurement
infrastructure with regard to processing, storage, and transfer capacity. To reduce the amount of
measurement data one may accumulate statistics, filter which packets/flows to log, and compress
measurement data. Finally, decisions must be made regarding how measurement results should be collected
and transferred to a data repository and how the measurement data should be handled and post-processed.
459
Speed
(successful
performance)
Accuracy
(incorrect
performance)
Dependability
(nonperformance)
Access
User
information
transfer
Disengagement
Table 1: ITU-T I.350 3x3 Matrix for Defining Service-Performance Parameters
5.2 Measurement Layer
Measurements may be carried out at several protocol layers. For each measurement layer, one or more
primary parameters can be defined for each combination of function/parameter type in the 3x3 matrix in
Table 1.
Application-layer measurements provide a detailed and realistic picture of the end-to-end performance of
an application or a service, and it is possible to find how the individual transactions of the application
contribute to the overall performance. File-transfer completion time, call set-up delay, and domain name
system (DNS) lookup time are relevant examples.
Network-layer measurements on the other hand may give detailed hop-by-hop information on typical
network-layer performance parameters like bandwidth, packet delay, delay jitter, and packet loss. While
network-layer measurements may indicate a maximum limit on the expected application performance,
performance bottlenecks at higher layers or in the interaction between the layers will cause application
performance to degrade further.
460
mobile communications (GSM) messaging services, seem less mature and rich compared to voice (and
video). Nevertheless, some experiments have been conducted, as for example those reported to the open
mobile alliance (OMA) on the push-to-talk service [xii].
Devising aggregation schemes means defining KPIs and specifying how they are calculated from the
basic measurement parameters. It is important that the defined KPIs are intuitive, reflect end users views,
and provide a sound basis for making decisions. Crucial points are that KPIs must allow for capturing
degraded service operation and locating causes when the former occur. Moreover, the statistical
characteristics on which KPIs are based must be taken into account. Obvious choices for statistical
characteristics are mean values, variance, and percentiles. Commonly, mean values and percentiles are
selected. The mean values will in many cases be straightforward to aggregate, whereas working on
percentiles may be more complicated.
From the outset, three classes of parameters could be recognized: i) the availability of the service, ii) the
speed related to the correct completion of a service, and iii) the accuracy of the service execution (e.g., in
cases where incorrect results are obtained or incorrect receivers are reached), see Table 1. Parameters
related to speed can be for instance end-to-end delay, throughput, and time to complete a service. Delay
contributions are normally additive, while availability and accuracy contributions generally are
multiplicative. Hence, it is expected that an approach for availability would be similar to one for
estimating accuracy.
It can be shown that basing KPI results on collected observations from separate monitoring apparatus,
assuming the observations are mutually independent, gives conservative results regarding availability and
accuracy parameters. That is, for availability and accuracy an approach based on independence between
all individual components contributing to the realization of the service, allows for estimates that are on
the safe side with respect to users experiences. Such a conservative approach may be sufficient for some
purposes, whereas for other cases it will be necessary also to carry out a dependency analysis of the
different components and thereby arrive at a more realistic value.
The estimates of expected delays should in general not be biased, neither toward a conservative approach
or otherwise but should rather be as close to the true values as possible. The same argumentation as
above, however, yields that conservative estimates, giving some upper bounds on delay, may be feasible
in some scenarios. Moreover, a balance must be found between the desired level of accuracy and the cost
associated with obtaining it. Regarding KPIs on delay percentiles, dependencies between systems
contributing to the overall service execution should be taken into account.
Figure 1 illustrates how different KPIs can result from the same set of underlying parameter
measurements. As previously described, it is important to be aware of the pitfalls related to dependencies
between the different measured values. If independence is assumed, values can be obtained in a more
straightforward manner, but the introduced incorrectness may be intolerable.
Focusing more on the usage of KPIs, Figure 2 shows a schematic illustration of the relationship between
the observed disturbance given by measured parameters or derived KPIs, and the end users degree of
satisfaction. The idea is that a disturbance below a threshold, say a, has no practical impact on a user.
Similarly, a disturbance above another threshold, say b, is not specifically recognized either, as a user
would most likely abandon the service offering. It still remains to describe the specific curve forms that
apply between the two thresholds, and the shape of this part of the curve will differ for the different
service types. An example is provided by ITUs E-model where curves for voice services are specified.
Curves like the one shown in Figure 2 will be helpful for identifying appropriate target and threshold
values, i.e., a and b, for the KPIs.
462
degree of
satisfaction
disturbance
Figure 2: Schematic Illustration of Relation between Parameter (Disturbance) and the End Users
Degree of SatisfactionSpecific Form of Curve Depends on Service
In addition to aggregating measured parameters into KPIs for individual services, other aggregation levels
are relevant. These levels correspond to the KPI receiver types as described in section 4 and illustrated in
Figure 1. For instance, the top level management would also ask for KPIs reflecting the overall operation,
not only those related to individual services and networks. Achieving single KPIs reflecting the
impression of a specific user type or the operation in a certain market segment can be possible by
performing aggregation methods like for instance placing weights or more general functions on network
and service KPIs.
Once threshold value(s) for a given KPI is (are) identified, for instance by using curves like the one
shown in Figure 2, the KPI receivers can follow the KPI over time and let the threshold values be used
for triggering actions. The actions goal is to keep the KPI value within the specified threshold levels and
preferably close to the target level.
Figure illustrates this procedure, the (normalized) KPI level being used as an action trigger.
KPI level
High threshold value
100
Target value
acceptance
range
time/periods
Figure 3: Following KPI Results Evolving over Time Allows for Triggering Actions, Possibly also Based
on Prediction and Trends Analyses
463
Concluding Remarks
This paper presents main ideas and principles for a general framework for KPIs for end-to-end quality of
service in a mobile wireless network. Mainly technical indicators have been discussed, although the KPI
framework is applicable to a much wider extent.
Types of needed KPIs are identified, and how to produce these KPIs from measurements is discussed, as
well as how to build a hierarchy of KPIs efficiently, given a set of KPI receivers. A main idea is that
instead of using active tests only, measurements in the network can be used to infer the end-to-end service
quality that can be expected by an end user. The same measurement platform should ideally be used for
all measurement-based operations in the network. This will contribute to a reduction of the overall cost
and work related to monitoring service performance. The platform must be designed such that it takes into
account the different groups interested in performance reports within the organisation.
References
i
ii
Arif Pathan. The Functional Areas of Network Management. (May 31, 2004) [on-line] URL:
http://trade.haml.fi/~lseppana/courses/projects/FCAPS.htm
iii
Poul Heegaard. Active tests for SLA validation Correctness, precision and intrusiveness, Fornebu, Telenor
Research and Development, 2004. Scientific Report N30/2002.
iv
Telecommunication Management; Performance Management (PM); Concepts and Requirements (Release 4).
Sophia Antipolis, France, June 2003. 3GPP TS 32.401 v4.3.0. (May 30, 2004) .[on-line] URL:
http://www.3gpp.org/ftp/Specs/archive/32_series/32.401/32401-430.zip
Internet Engineering Task Force. IP Performance Metrics (ippm). (July 07, 2004). [on-line] URL:
http://www.ietf.org/html.charters/ippm-charter.html
vi
ITU. Internet protocol data communication service IP packet transfer and availability performance parameters.
2000. ITU-T Y.1540 (former I.380).
vii
ITU. Internet protocol data communication service Network performance objectives for IPbased services.
2002. ITU-T Y.1541.
viii
ITU. General aspects of quality of service and network performance in digital networks, including ISDNs. 1993.
ITU-T I.350.
ix
Performance Measurement Tools Taxonomy. Cooperative Association for Internet Data Analysis (CAIDA). (July
08, 2004). [on-line] URL: http://www.caida.org/tools/taxonomy/performance.xml
Network Monitoring Tools. Stanford Linear Accelerator Center (SLAC). (July 08, 2004). [on-line] URL:
http://www.slac.stanford.edu/xorg/nmtf/nmtf-tools.html
xi
ITU Recommendation G.109: Definition of categories of speech transmission quality. Sept. 1999.
xii
Open Mobile Alliance, Push to Talk over Cellular Working Group. (July 07, 2004). [on-line] URL:
http://www.openmobilealliance.org/tech/wg_committees/poc.html.
464
Final-Year M.Tech. in CSE. E-mail seema_kauser@yahoo.com with any questions concerning this paper.
465
802.11 Authentication
Non cryptographic
Does not use RC4
A station is allowed to
join a network without
any identity
Cryptographic
Uses RC4
A station is allowed to
join a network, if it
proves WEP key is
shared. (Fundamental
security is based on
knowledge of secret key)
466
Seema Kauser
Security Issue or
Vulnerability
Cryptographic keys are
shared.
Initialization vectors
(IVs) are short
Cryptographic keys
cannot be updated and
changed frequently.
Authentication not
enabled; only simple
SSID identification
occurs.
Device authentication
is simple shared-key
channel-response.
Remarks
Keys that are shared
can compromise a
system. As the number
of people sharing the
key grows, the security
risks also grow.
24-bit IVs cause the
generated key stream
to repeat. Repetition
allows easy decryption
of data for a
moderately
sophisticated
adversary.
Cryptographic keys
should be changed
often to prevent brute
force attack.
Identity-based systems
are highly vulnerable
particularly in a
wireless system
because signals can be
more easily
intercepted.
One-way challengeresponse
authentication is
subject to man-in themiddle attacks.
Mutual authentication
is required to provide
verification that users
and the network are
legitimate.
The client needs to
authenticate the AP to
ensure that it is
legitimate
Mutuality
It must provide mutual authentication, that is, the authenticator must authenticate the user and the
user must be able to authenticate the authenticator as well. Mutual authentication is particularly
important over wireless networks because of the ease with which an attacker can set up a rogue
access point.
467
Self-Protecting
It must protect itself from eavesdropping since the physical medium is not secure. The
authentication must proceed in such a way that eavesdroppers cannot learn anything useful that
would allow them to impersonate the user later.
468
Seema Kauser
EAP client connects to the network and tries to access information on the network.
Communication is performed using the EAPOL protocol.
Authenticator responds to the client by asking for its identity using EAPOL, client responds to the
authenticator with its identity information using EAPOL.
Authenticator forwards the clients identity information to the authentication server using the
necessary protocol agreed upon by the authentication server and supplicant.
469
Authentication server responds with a challenge to the authenticator, which will be forwarded to
the supplicant. This message is transmitted over the remote authentication dial-in user service
(RADIUS) protocol back to authenticator.
Client examines the challenge and determines if it can support the requested EAP authentication
protocol. If it cannot, then it issues a negative acknowledgment (NAK) [request and tries to
negotiate an alternative authentication method. If the client supports the authentication type, then
it responds with its credential information using EAPOL.
Authenticator relays the clients credentials to the authentication server using the RADIUS protocol.
If the clients credentials are valid, then the authentication server authenticates and authorizes the
client. Otherwise, the client is rejected and the appropriate RADIUS access-accept or accessreject message is sent back to the authenticator using the RADIUS protocol. Authenticator
forwards the access-accept or access-reject to the supplicant.
Wired
Connection
Authentication
Server
(Ex: RADIUS
Server)
Wireless or wired
medium
Authenticator
(Wireless Access
point)
Supplicant
(Wireless station)
802.1x/EAP
request identity
802.1x/EAP
response
RADIUS access
request/identity
470
Seema Kauser
use any public key infrastructure (PKI) certificates to validate the client or provide strong encryption to
protect the authentication messages between the client and the authentication server. This makes the
EAPMD5 authentication protocol susceptible to session hijacking and man-in-the-middle attacks. EAP
MD5 is best suited for EAP message exchanges in wired networks where the EAP client is directly
connected to the authenticator and the chances of eavesdropping or message interception is very low. For
wireless 802.1X authentication, stronger EAP authentication protocols are used.
EAPTLS (RFC 2716)
EAPTLS provides strong security by requiring both client and authentication server to be identified and
validated through the use of PKI certificates. EAPTLS provides mutual authentication between the client
and the authentication server and is very secure. EAP messages are protected from eavesdropping by a
TLS tunnel between the client and the authentication server. The major drawback of EAPTLS is the
requirement for PKI certificates on both the clients and the authentication servers, making roll out and
maintenance much more complex. EAPTLS is best suited for installations with existing PKI certificate
infrastructures. Wireless 802.1X authentication schemes will typically support EAPTLS to protect the
EAP message exchange.
EAPTTLS (Internet Draft)
EAPTTLS is an extension of EAPTLS and provides the benefits of strong encryption without the
complexity of mutual certificates on both the client and authentication server. Like TLS, EAPTTLS
supports mutual authentication but only requires the authentication server to be validated to the client
through a certificate exchange. EAPTTLS allows the client to authenticate to the authentication server
using usernames and passwords and only requires a certificate for the authentication servers. EAPTTLS
simplifies roll out and maintenance and retains strong security and authentication. A TLS tunnel can be
used to protect EAP messages and existing user credential services. Backward compatibility for other
authentication protocols such as PAP, CHAP, MSCHAP, and MSCHAPV2 are also provided by
EAPTTLS. EAPTTLS is not considered foolproof and can be fooled into sending identity credentials if
TLS tunnels are not used. EAPTTLS is best suited for installations that require strong authentication
without the use of mutual certificates. Wireless 802.1X authentication schemes will typically support
EAPTTLS.
PEAP (Internet Draft)
PEAP is similar to EAPTTLS in terms of mutual authentication functionality and is currently being
proposed by RSA Security, Cisco, and Microsoft as an alternative to EAPTTLS. PEAP addresses the
weaknesses of EAP by doing the following:
PEAP allows other EAP authentication protocols to be used and secures the transmission with a TLS
encrypted tunnel. It relies on the mature TLS keying method for its key creation and exchange. The PEAP
client authenticates directly with the backend authentication server and the authenticator acts as a passthrough device, which does not need to understand the specific EAP authentication protocols. Unlike
EAPTTLS, PEAP does not natively support username and password authentication against an existing
user database. Vendors are answering this need by creating features to allow this. PEAP is best suited for
installations that require strong authentication without the use of mutual certificates
471
Cisco LEAP
Ciscos LEAP was developed in November 2000 to address the security issues of wireless networks.
LEAP is a form of EAP that requires mutual authentication between the client and the authenticator. The
client first authenticates itself to the authenticator, and then the authenticator authenticates itself to the
client. If both authenticate successfully, a network connection is granted. Unlike EAPTLS, LEAP is
based on username and password schemes and not PKI certificates, simplifying roll out and maintenance.
The drawback is that it is proprietary to Cisco and has not been widely adopted by other networking
vendors. LEAP is best suited for wireless implementations that support Cisco APs and LEAPcompliant
wireless network interface cards (NICs).
As the wireless security requirements push the development of 802.1X and secure data transport, newer
EAP authentication protocols will be developed to answer the security issues. With IEEE 802.1X and the
EAP standard, these new EAP security protocols should continue to work with the existing ones.
4 . Conclusion
Securing a wireless network provides tremendous cost savings, productivity benefits, and a competitive
market advantage. Choosing the highest level of security available is a good investment, because security
breaches can be a significant expense. Most attacks go unnoticed, and enterprises can be vulnerable to
damages. Security breaches such as stolen information, corrupt data, and network downtime can be
expensive. They can also result in consequential damages.
Although EAP supports a bunch of authentication methods, only four are commonly used. These four are
summarized in Table 2.
Protocol
EAPMD5
EAPTLS
EAPTTLS
PMK Created?
NO
YES
YES
Deployment Difficulty
Easy
Hard
Easy
Wireless Security
POOR
STRONG
GOOD
472
Institute for Electrical and Electronics Engineers, IEEE 802.11, Wireless LAN Medium Access
Control (MAC) and Physical Layer (PHY) Specifications, 1999.
IEEE 802.1X, Standard for Port Based Network Access Control, 2001.
IEEE 802.11i/Draft 7.0, Wireless LAN Medium Access Control (MAC) and Physical Layer
(PHY) Specifications: Specification for Enhanced Security, October 2003.
Blunk, L. and J. Vollbrecht, PPP Extensible Authentication Protocol (EAP), RFC 2284, March
1998.
Dierks, T. and C. Allen, The TLS Protocol Version 1.0, RFC 2246, November 1998.
Aboba, B. and D. Simon, PPP EAP TLS Authentication Protocol, RFC 2716, October 1999.
Paul Fank and Simon Balke-Wilson, EAP Tunneled TLS Authentication Protocol, Draft-ietfpppext-eap-ttls-01.txt, February 2002.
A Novel Broadband/Wireless
Routing Algorithm
Deductive Probe Routing with MultiQoS Constraints
Yuxing Tian
Senior Engineer, GTSS
Motorola Inc.
473
A Novel Broadband/Wireless Routing Algorithm Deductive Probe Routing with MultiQoS Constraints
been developed and accepted by service providers to meet the demand of real-time multimedia
communications applications. Among them wireless fidelity (WiFi) (IEEE 802.11), Worldwide
Interoperability for Microwave Access, Inc. (WiMAX) (IEEE 802.16), and Mobile-Fi (IEEE 802.20) are
representative wireless access technologies that offer portability, mobility, coverage, and broad
bandwidth for a new generation of multimedia applications. Under the competition to provide mobility,
global coverage, and decent bandwidth for multimedia applications, wireless phone companies are also
upgrading their infrastructures to the third-generation (3G) standards. Combination of wireless access
network technology and the growth in the capacity of the Internet backbone has enabled multimedia
communications applications to various kinds of mobile access devices, including but not limited to
notebook computers, personal digital assistants (PDAs), camera phones, and 3G handsets that support
business applications. This development in turn contributes more growth stimulation back to the
communications technology evolution for the next generation of applications. At the same time, emerging
standards are being published to provide greater mobility and broader bandwidth access for the Internet,
while some existing technologies begin to integrate into each other, like seamless handover technique
between a WiFi voice over IP (VoIP) session and wireless telephone connections (see Figure 1).
474
In this paper, we propose a novel approach to QoS routing that properly treats different QoS parameters to
find the best available path from a source to a destination, which has the required network resources to meet
the QoS constraints of a new connection request. We will also provide a performance analysis on its
computational complexity and cost. The rest of the paper is organized as follows. We start with a discussion
on key issues in QoS routing. Then we present our new QoS routing algorithm, in the context of one wellknown QoS parameter, followed by a discussion on the capability of this new routing algorithm working
with other existing routing algorithms. Finally, we conclude the paper with a summary.
2. Key Issues in QoS Routing
QoS constraints provide a mechanism for different communications applications to quantitatively express
their demand on network resources to achieve desired user experiences. All established QoS parameters
may be divided into two categories. One category is comprised of parameters describing individual link
states that only have local impact, and the other is comprised of parameters describing end-to-end path
state information. In this paper, we adopt the terminology of ATM Forum private network node interface
(PNNI) specification. We use the term attribute to describe a link state parameter that is considered
individually to determine whether a given link is acceptable and/or desirable for carrying a given
connection [4]. The term metric is used to describe a link parameter that requires the values of the
parameters for all links along a given path to be combined to determine whether the path is acceptable
and/or desirable for carrying a given connection [4]. For example, popular QoS attributes are bandwidth,
error rates, and the popular QoS metrics are delay, jitter, and cell loss ratio, etc. Bandwidth is the
principal resource a network provides to all user applications, i.e., the size of the pipe to transport data
from their sources to the destinations. Interestingly, the majority of multiQoS requirements of
multimedia connection requests usually have both QoS attributes and metrics tied together, like
bandwidth and delay.
Routing algorithm is the key part of the connection admission control to determine the availability of
sufficient network resources to support new connection requests while maintaining the service commitment
to the established connections. Routing with multiple QoS constraints are essential to the success of realtime multimedia applications, due to the stringent requirements on connection bandwidth, jitter, delay, and
error rates, etc. For wireless access networks, error rates are mostly relaxed because of the interference and
vulnerability of the signals when passing through the air. Unfortunately as it is shown in [5], finding a
feasible path with two independent path constraints is NPComplete, i.e., the type of problems cannot be
solved quickly in practice without committing extreme amount of computation resources.
As the size of network grows, aggregation techniques have been employed to address the scalability issue.
Because of the irregularity of the network topologies, errors are inevitable when QoS attributes are
aggregated together. At the meantime, maintaining link status information on the aggregated nodes is
expensive in terms of both computation and control signal exchange traffic in the network. Inaccuracy is
easily introduced with regard to the dynamic nature of the communications network when the traffic
pattern and distribution change faster than the update cycles. Link status data are usually collected and
updated on QoS routers throughout the network. To make things worse when network traffic load is high,
such collection and update traffic compete with data traffic for the limited network resources.
3. Deductive Probe (DP) Routing Algorithm
Today, most QoS routing algorithms treat all QoS attribute and metric parameters mechanically, which is
not very efficient. We have integrated the metric type of QoS requirements into the computation logic for
finding optimal point-to-point path and drastically simplified the complexity of QoS routing algorithm.
Because of its transmutation from QoS metric to routing logic, we name our routing algorithm deductive
probe (DP) routing with multiQoS constraints. It is applicable to any entry point to the Internet, be it a
475
A Novel Broadband/Wireless Routing Algorithm Deductive Probe Routing with MultiQoS Constraints
cell phone network base station, a WiFi hotspot hub, or a WiMAX or Mobile-Fi transceiver tower. For
QoS attributes, like link bandwidth and link error rates, we still pass them in as arguments for QoS
requirements to be satisfied by the routing algorithm. The DP routing algorithm utilizes node coloring
technique as used in breadth-first search techniques and selectively floods the probe packets to
downstream links as specified in its reachability table. The reachability table is locally maintained by
each network node, which consists of addresses and address prefixes describing the destinations to which
connections may be routed. Even though generic flooding technique is expensive and bandwidth
consuming, our DP algorithm does not suffer this drawback as described later. Theoretically, the worstcase computation complexity is O(|E|), where E is the set of links in the network.
Another advantage of the DP algorithm is that no link status information needs to be maintained. It can
always find the shortest route among all paths that meet the QoS requirements and leads to optimized
network utilization without imposing traffic consumption on its resources. Even though our QoS routing
algorithm does not scale as well as some existing routing algorithm, e.g., asynchronous transfer mode
(ATM) PNNI architecture [4], our algorithm outperforms them in more critical times when networks
loads are high. As mentioned earlier, there is no erroneous network state aggregation and expensive
maintenance work for our routing algorithm. The worst-case computation complexity of our routing
algorithm is the same as its best case, i.e., to traverse the exact path to be established. In another word, it
is always known to be under the requested maximum end-to-end delay of the applications. By setting the
probe packets at the same priority level as the applications and their discard flag to be true, our routing
algorithm imposes virtually no load on network resources especially during overload conditions.
The DP routing algorithm has two major components. One is invoked by routing nodes to probe
downstream for eligible paths, and the other reports probe status back to the upstream node. The version
presented next has transformed the most popular QoS metric parameter, the maximum end-to-end delay, to
our generic DP routing algorithm logic. Both pseudocode and flowcharts are given for each component.
3.1 DP Routing Downstream
Here we use the label linkQoS to represent any QoS attributes, be a specific link state constraint or a set
of some. For example, it could be link bandwidth or error rates or both. Our DP algorithm does not
perform an exhaustive search. Instead, we have simplified the coloring scheme to use one flag rather than
two colors as generic breadth-first-search algorithms do for finding shortest paths (see Figure 2).
DP_ROUTING_DOWNSTREAM (source, destination, timeout0, linkQoS)
1. if source.IPaddress = self.IPaddress
2.
then source.ID = self.IPaddress + self.ApplicationPID
3. i = source.ID + destination.IPaddress + destination.Socket
4. if routing_task[i].Colored = True
5.
then STOP()
6.
else routing_task[i].Colored True
7. aliveFlag False
8. for j 1 to |D[]: downstream links reaching destination|
9.
if D[j].linkQoS() linkQoS
10.
timeout1 timeout0 DELAY(self)
11.
then if timeout1 > 0
12.
then D[i].SEND(source, destination, timeout1, linkQoS)
13.
aliveFlag True
14. if aliveFlag = False
15.
then k GETUPLINKID(routing_task[i])
16.
S[k].SEND(source, destination, NACK)
476
A Novel Broadband/Wireless Routing Algorithm Deductive Probe Routing with MultiQoS Constraints
indicating its passage through the node so that any later arrivals of other subprocesses for the same
connection request will not perform any redundant work. This is due to the nature of the flooding
mechanism; different probe subprocesses may come across the same routing node via separate paths and
at different time.
The rest of the algorithm explores for qualifying route over all downstream links leading to the
destination, by looking up a reachability table. For each link, subprocess examines whether its link state is
able to accommodate the requested QoS attribute requirements. If the requirements can be satisfied, it
calculates the new timeout value by subtracting the delay at this node, such as queuing delay and the link
transmission delay. If the delay metric is not violated, it sends probe packet to the downstream link, and
marks the probe subprocess as alive. After all the downstream links have been examined, the DP
subprocess checks whether the search is still alive. If not, it sends a negative acknowledgement packet
(NACK) back upstream node to indicate incapable of finding any route that could support the QoS
requirements of the new connection.
The most expensive part of the algorithm is the block of the second half, which has a theoretical worstcase computation complexity of O(|E|), where E is the set of the network links. But, as we know in
reality, the number of downstream links at each node is so limited that the effective computation
complexity can be considered O(1) in practice, and this makes the whole DP Routing has a
computational complexity of O(|V|), where V is the set of all network nodes, that is usually far less than
the number of links in the network.
3.2 DP Routing Upstream
This portion of the algorithm is under the same context as that described for the downstream algorithm
(see Figure 3).
DP_ROUTING_UPSTREAM (source, destination, status)
1. if source.IPaddress = self.IPaddress
2.
then if status = ACK
3.
then source.ADMITTED()
4.
STOP()
5.
if self.TIMEOUT() = True
6.
source.REJECTED()
7.
else i = source.ID + destination.ID
8.
j GETUPLINKID(routing_task[i])
9.
if routing_task[i].TIMEOUT() = True
10.
then STOP()
11.
else if status = ACK
12.
then S[j].SEND(source, destination, ACK)
13.
else S[j].SEND(source, destination, NACK)
478
479
A Novel Broadband/Wireless Routing Algorithm Deductive Probe Routing with MultiQoS Constraints
[2]
W.C. Hardy, QoS Measurement and Evaluation of Telecommunications Quality of Service (John
Wiley & Sons, Ltd., 2001).
[3]
ATM Forum technical committee, Traffic Management Specification, Version 4.1 (AF-TM0121.000), March 1999.
[4]
ATM Forum technical committee, Private NetworkNetwork Interface Specification, Version 1.0
(PNNI 1.0), March 1996.
[5]
M. Garey and D. Johnson, Computers and Intractability: A Guide to the Theory of NPCompleteness (New York: W.H. Freeman and Co., 1979).
[6]
Cristina Aurrecoechea, A.T. Campbell, and L. Hauw, A survey of QoS architectures, Multimedia
Systems, Vol. 6, 1998, pp. 138151.
480
[7]
S. Keshav, Report on the Workshop on Quality of Service Issues in High-Speed Networks, ACM
Computer Communications, Rev 22 (1), 1993, pp. 615.
[8]
[9]
[10] J.D. Gibson, Multimedia Communications: Directions and Innovations (Academic Press, 2001).
[11] T.H. Cormen, C.E. Leiserson, R.L. Rivest, and C. Stein, Introduction to Algorithms, Second Edition
(The MIT Press, 2001).
481
Acronym Guide
2B1Q
2G
3DES
3G
3GPP
3R
4B3T
4F/BDPR
4F/BSPR
4G
AAA
AAL[x]
ABC
ABR
AC
ACD
ACF
ACH
ACL
ACLEP
ACM
ACR
ADM
ADPCM
ADS
ADSI
ADSL
AES
AFE
AGW
AIM
AIN
ALI
AM
AMA
AMI
AMPS
AN
ANI
ANM
ANSI
ANX
AOL
AON
AP
APC
API
APON
APS
ARCNET
ARI
ARM
ARMS
ARP
ARPANET
ARPU
A-Rx
AS
ASAM
ASC
ASCII
ASE
ASIC
ASIP
ASON
ASP
ASR
ASSP
ASTN
ATC
ATIS
ATM
ATMF
ATP
ATUC
ATUR
A-Tx
AUI
AVI
AWG
AYUTOS
B2B
B2C
BBS
BCSM
BDCS
BDPR
BE
BER
BERT
BGP
BH
BHCA
BI
BICC
BID
484
BIP
BISDN
BLEC
BLES
BLSR
BML
BOC
BOF
BOND
BOSS
BPON
BPSK
BRAS
BRI
BSA
BSPR
BSS
BTS
BVR
BW
CA
CAC
CAD
CAGR
CALEA
CAM
CAMEL
CAP
CAPEX
CAR
CARE
CAS
CAT
CATV
C-band
CBDS
CBR
CBT
CC
CCB
CCF
CCI
CCITT
CCK
CCR
CCS
CD
cDCF
CDD
CDDI
CDMA
CDMP
CDMS
CDN
CDPD
CDR
CDROM
CDWDM
CE
CEI
CEO
CER
CERT
CES
CES
CESID
CEV
CFB/NA
CFO
CFP
CGI
CHCS
CHN
CHTML
CIC
CID
CIM
CIMD2
CIO
CIP
CIR
CIT
CLASS
CLE
CLEC
CLI
CLID
CLLI
CLR
CM
CM&B
CMIP
CMISE
CMOS
CMRS
CMTS
CNAM
CNAP
CNS
CO
486
CODEC
COI
COO
COPS
CORBA
CORE
CoS
COT
COTS
COW
CP
CPAS
CPC
CPE
CPI
CPL
CPLD
CPN
CPU
CR
CRC
CRIS
CRLDP
CRM
CRS
CRTP
CRV
CS
CS[x]
CSA
CSCE
CSCF
CSCW
CSE
CSF
CSIWF
CSM
CSMA/CA
CSMA/CD
CSN
CSP
CSR
CSU
CSV
CT
CT2
CTI
CTIA
CTO
CUG
CWD
coder-decoder OR compression/decompression
community of interest
chief operations officer
common open policy service
common object request broker architecture
council of registrars
class of service
central office terminal
commercial off-the-shelf
cell site on wheels
connection point
cellular priority access service
calling-party category (also calling-party control OR calling-party connected)
customer-premises equipment
continual process improvement
call processing language
complex programmable logic device
calling-party number
central processing unit
constraint-based routing
cyclic redundancy check OR cyclic redundancy code
customer records information system
constraint-based routedlabel distribution protocol
customer-relationship management
capacity, reach, and speed
compressed real-time transport protocol
call reference value
client signal
capability set [x]
carrier serving area
converged service-creation and execution
call-state control function
computer supported collaborative work
CAMEL service environment
critical success factors
control signal interworking function
customer-service manager
carrier sense multiple access with collision avoidance
carrier sense multiple access with collision detection
circuit-switched network
communications service provider OR content service provider
customer-service representative
channel service unit
circuit-switched voice
computer telephony
cordless telephony generation 2
computer telephony integration
Cellular Telecommunications & Internet Association
chief technology officer
closed user group
centralized wavelength distribution
487
CWDM
CWIX
DAC
DACS
DAM
DAMA
DAML
DARPA
DAVIC
DB
dB
DBMS
dBrn
DBS
DC
DCC
DCF
DCLEC
DCM
DCN
DCOM
DCS
DCT
DDN
DDS
DECT
demarc
DEMS
DES
DFB
DFC
DGD
DGFF
DHCP
DiffServ
DIN
DIS
DITF
DLC
DLCI
DLE
DLEC
DLR
DM
DMD
DMS
DMT
DN
DNS
DOC
DOCSIS
488
DOD
DOJ
DoS
DOS
DOSA
DOT
DOW
DP
DPC
DPE
DPT
DQoS
D-Rx
DS[x]
DSAA
DSC
DSCP
DSF
DSL
DSLAM
DSLAS
DSP
DSS
DSSS
DSU
DTH
DTMF
DTV
D-Tx
DVB
DVC
DVD
DVMRP
DVoD
DVR
DWDM
DXC
E911
EAI
EAP
EBITDA
EC
ECD
ECRM
ECTF
EDA
EDF
EDFA
EDGE
EDI
EDSX
Department of Defense
Department of Justice
denial of service
disk operating system
distributed open signaling architecture
Department of Transportation
day of week
detection point
destination point code
distributed processing environment
dial pulse terminate
dynamic quality of service
digital receiver
digital signal [level x]
DECT standard authentication algorithm
DECT standard cipher
DiffServ code point
dispersion-shifted fiber
digital subscriber line [also xDSL]
digital subscriber line access multiplexer
DSLATM switch
digital signal processor OR digital service provider
decision support system
direct sequence spread spectrum
data service unit OR digital service unit
direct-to-home
dual-tone multifrequency
digital television
digital transceiver
digital video broadcast
dynamic virtual circuit
digital video disc
distance vector multicast routing protocol
digital video on demand
digital video recording
dense wavelength division multiplexing
digital cross-connect
enhanced 911
enterprise application integration
extensible authentication protocol
earnings before interest, taxes, depreciation, and amortization
electronic commerce
echo-cancelled full-duplex
echo canceller resource module
Enterprise Computer Telephony Forum
electronic design automation
electronic distribution frame OR erbium-doped fiber
erbium-doped fiber amplifier
enhanced data rates for GSM evolution
electronic data interchange
electronic digital signal cross-connect
489
EFM
EFT
EJB
ELAN
ELEC
EM
EMI
EML
EMS
ENNI
ENUM
EO
EO
EoA
EOC
EoVDSL
EPD
EPON
EPROM
ERP
ESCON
ESS
ETC
EtherLEC
ETL
eTOM
ETSI
EU
EURESCOM
EXC
FAB
FAQ
FBG
FCAPS
FCC
FCI
FCIF
FDA
FDD
FDDI
FDF
FDM
FDMA
FDS1
FE
FEC
FEPS
FEXT
FHSS
FICON
FITL
490
FM
FOC
FOT
FOTS
FP
FPB
FPGA
FPLMTS
FPP
FR
FRAD
FSAN
FSC
FSN
FT
FT1
FTC
FTE
FTP
FTP3
FTTB
FTTC
FTTCab
FTTEx
FTTH
FTTN
FTTS
FTTx
FWM
FX
G3
G4
GA
Gb
GbE
GBIC
Gbps
GCRA
GDIN
GDMO
GE
GEO
GETS
GFF
GFR
Ghz
GIF
GIS
GKMP
GMII
GMLC
GMPCS
GMPLS
GNP
GOCC
GPIB
GPRS
GPS
GR
GRASP
GSA
GSM
GSMP
GSR
GTT
GUI
GVD
GW
HCC
HD
HDLC
HDML
HDSL
HDT
HDTV
HDVMRP
HEC
HEPA
HFC
HIDS
HLR
HN
HOM
HomePNA
HomeRF
HPC
HPO
HQ
HSCSD
HSD
HSIA
HSP
HTML
HTTP
HVAC
HW
IAD
IAM
IAS
IAST
IAT
IBC
492
IC
ICD
ICDR
ICL
ICMP
ICP
ICS
ICW
IDC
IDE
IDES
IDF
IDL
IDLC
IDS
IDSL
IEC
IEEE
I-ERP
IETF
IFITL
IFMA
IFMP
IGMP
IGP
IGRP
IGSP
IHL
IIOP
IIS
IKE
ILA
ILEC
ILMI
IM
IMA
IMAP
IMRP
IMSI
IMT
IMTC
IN
INAP AU
INAP
INE
InfoCom
INM
INMD
INNI
INT
InterNIC
integrated circuit
Internet call diversion
Internet call detail record
intercell linking
Internet control message protocol
integrated communications provider OR intelligent communications platform
integrated communications system
Internet call waiting
Internet data center OR International Data Corporation
integrated development environment
Internet data exchange system
intermediate distribution frame
interface definition language
integrated digital loop carrier
intrusion detection system
integrated services digital network DSL
International Electrotechnical Commission OR International Engineering Consortium
Institute of Electrical and Electronics Engineers
integrated enterprise resource planning
Internet Engineering Task Force
integrated [services over] fiber-in-the-loop
International Facility Managers Association
Ipsilon flow management protocol
Internet group management protocol
interior gateway protocol
interior gateway routing protocol
independent gateway service provider
Internet header length
Internet interORB protocol
Internet Information Server
Internet key exchange
in-line amplifier
incumbent local-exchange carrier
interim link management interface
instant messaging
inverse multiplexing over ATM
Internet message access protocol
Internet multicast routing protocol
International Mobile Subscriber Identification
intermachine trunk OR International Mobile Telecommunications
International Multimedia Teleconferencing Consortium
intelligent network
INAP adaptation unit
intelligent network application part
intelligent network element
information communication
integrated network management
in-service, nonintrusive measurement device
internal network-to-network interface
[point-to-point] interrupt
Internet Network Information Center
493
IntServ
IOF
IOS
IP
IPBX
IPcoms
IPDC
IPDR
IPe
IPG
IPO
IPoA
IPQoS
IPSec
IPTel
IPv6
IPX
IR
IRU
IS
ISIS
ISA
ISAPI
ISC
ISDF
ISDN
ISDNBA
ISDNPRA
ISEP
ISM
ISO
ISOS
ISP
ISUP
ISV
IT
ITSP
ITTP
ITU
ITUT
ITV
IVR
IVRU
IWF
IWG
IWU
IXC
J2EE
J2ME
J2SE
JAIN
494
integrated services
interoffice facility
intelligent optical switch
Internet protocol
Internet protocol private branch exchange
Internet protocol communications
Internet protocol device control
Internet protocol data record
intelligent peripheral
intelligent premises gateway
initial public offering OR Internet protocol over optical
Internet protocol over ATM
Internet protocol quality of service
Internet protocol security
IP telephony
Internet protocol version 6
Internet package exchange
infrared
indefeasible right to user
information service OR interim standard
intermediate system to intermediate system
industry standard architecture
Internet server application programmer interface
integrated service carrier OR International Softswitch Consortium
integrated service development framework
integrated services digital network
ISDN basic access
ISDN primary rate access
intelligent signaling endpoint
industrial, scientific, and medical OR integrated service manager
International Organization for Standardization
integrated software on silicon
Internet service provider
ISDN user part
independent software vendor
information technology OR Internet telephony
Internet telephony service provider
information technology infrastructure library
International Telecommunication Union
ITUTelecommunication Standardization Sector
Internet television
interactive voice response
interactive voice-response unit
interworking function
interworking gateway
interworking unit
interexchange carrier
Java Enterprise Edition
Java Micro Edition
Java Standard Edition
Java APIs for integrated networks
JCAT
JCC
JDBC
JDMK
JMAPI
JMX
JPEG
JSCE
JSIP
JSLEE
JTAPI
JVM
kbps
kHz
km
L2F
L2TP
LAC
LAI
LAN
LANE
LATA
LB311
L-band
LBS
LC
LCD
LCP
LCUG
LD
LDAP
LDCELP
LDP
LDS
LE
LEAF
LEC
LED
LEO
LEOS
LER
LES
LIDB
LL
LLC
LMDS
LMN
LMOS
LMP
LMS
LNNI
LNP
LNS
LOI
LOL
LOS
LPF
LQ
LRN
LRQ
LSA
LSB
LSMS
LSO
LSP
LSR
LT
LTE
LUNI
LX
M2PA
M2UA
M3UA
MAC
MADU
MAN
MAP
MAS
Mb
MB
MBAC
MBGP
MBone
Mbps
MC
MCC
MCU
MDF
MDSL
MDTP
MDU
MEGACO
MEMS
MExE
MF
MFJ
MG
MGC
MGCF
MGCP
MHz
MIB
496
MII
MIME
MIMO
MIN
MIPS
MIS
MITI
MLT
MM
MMDS
MMPP
MMS
MMUSIC
MNC
MOM
MON
MOP
MOS
MOSFP
MOU
MPC
MPEG
MPI
MPLambdaS
MPLS
MPOA
MPOE
MPOP
MPP
MPx
MRC
MRS
MRSP
ms
MSC
MSF
MSIN
MSNAP
MSO
MSP
MSPP
MSS
MSSP
MTA
MTBF
MTP [x]
MTTR
MTU
MVL
MWIF
MZI
N11
NA
NAFTA
NANC
NANP
NAP
NARUC
NAS
NASA
NAT
NATA
NBN
NCP
NCS
NDA
NDMU
NDSF
NE
NEAP
NEBS
NEL
NEXT
NFS
NG
NGCN
NGDLC
NGF
NGN
NGOSS
NHRP
NI
NIC
NID
NIDS
NIIF
NIS
NIU
nm
NML
NMS
NND
NNI
NNTP
NOC
NOMAD
NP
NPA
NPAC
NPN
NPREQ
NPV
498
(refers to FCCmanaged dialable service codes such as 311, 411, and 911)
network adapter
North America Free Trade Agreement
North American Numbering Council
North American Numbering Plan
network access point
National Association of Regulatory Utility Commissioners
network access server
National Aeronautics and Space Administration
network address translation
North American Telecommunications Association
node-based network
network control protocol
national communications system OR network connected server
national directory assistance
network data managementusage
non-dispersion-shifted fiber
network element
non-emergency answering point
network-equipment building standards
network-element layer
near-end crosstalk
network file system
next generation
next-generation converged network
next-generation digital loop carrier
next-generation fiber
next-generation network
next-generation operations system and software OR next-generation OSS
next-hop resolution protocol
network interface
network interface card
network interface device
network intrusion detection system
Network Interconnection Interoperability Forum
network information service
network interface unit
nanometer
network-management layer
network-management system
name and number delivery
network-to-network interface
network news transport protocol
network operations center
national ownership, mobile access, and disaster communications
number portability
numbering plan area
Number Portability Administration Center
new public network
number-portable request query
net present value
NRC
NRIC
NRSC
NRZ
NS/EP
NSAP
NSAPI
NSCC
NSDB
NSP
NSTAC
NT
NTN
NTSC
NVP
NZDSF
O&M
OA&M
OADM
OAM&P
OBF
OBLSR
OC[x]
OCBT
OCD
OCh
OCR
OCS
OCU
OCX
OD
ODBC
ODSI
OE
OEC
OECD
OEM
OEO
OEXC
OFDM
OIF
OLA
OLAP
OLI
OLT
OLTP
OMC
OMG
OMS SW
OMS
OMSSPRING
ONA
ONE
ONI
ONMS
ONT
ONTAS
ONU
OP
OPEX
OPS
OPTIS
OPXC
ORB
ORT
OS
OSA
OSC
OSD
OSGI
OSI
OSMINE
OSN
OSNR
OSP
OSPF
OSS
OSS/J
OSU
OTM
OTN
OUI
O-UNI
OUSP
OVPN
OWSR
OXC
P&L
PABX
PACA
PACS
PAL
PAM
PAMS
PAN
PANS
PBCC
PBN
PBX
PC
PCF
PCI
500
PCM
PCN
PCR
PCS
PDA
PDC
PDD
PDE
PDH
PDN
PDP
PDSN
PDU
PE
PER
PERL
PESQ
PFD
PHB
PHY
PIC
PICS
PIM
PIN
PINT
PINTG
PKI
PLA
PLC
PLCP
PLL
PLMN
PLOA
PM
PMD
PMDC
PMO
PMP
PN
PNNI
PnP
PO
PODP
POET
POF
POH
POIS
PON
PoP
POP3
POS
PosReq
POT
POTS
PP
PPD
PPP
PPPoA
PPPoE
PPTP
PPWDM
PQ
PRI
ps
PSAP
PSC
PSD
PSDN
PSID
PSN
PSPDN
PSQM
PSTN
PTE
PTN
PTP
PTT
PUC
PVC
PVM
PVN
PWS
PXC
QAM
QOE
QoS
QPSK
QSDG
RAD
RADIUS
RADSL
RAM
RAN
RAP
RAS
RBOC
RCP
RCU
RDBMS
RDC
RDSLAM
REL
502
position request
point of termination
plain old telephone service
point-to-point
partial packet discard
point-to-point protocol
point-to-point protocol over ATM
point-to-point protocol over Ethernet
point-to-point tunneling protocol
point-to-pointwavelength division multiplexing
priority queuing
primary rate interface
picosecond
public safety answering point
Public Service Commission
power spectral density
public switched data network
private system identifier
public switched network
packet-switched public data network
perceptual speech quality measure
public switched telephone network
path terminating equipment
personal telecommunications number service
point-to-point
Post Telephone and Telegraph Administration
public utility commission
permanent virtual circuit
parallel virtual machine
private virtual network
planning workstation
photonic cross-connect
quadrature amplitude modulation
quality of experience
quality of service
quaternary phase shift keying
QoS Development Group
rapid application development
remote authentication dial-in user service
rate-adaptive DSL
remote access multiplexer
regional-area network
resource allocation protocol
remote access server
regional Bell operating company
remote call procedure
remote control unit
relational database management system
regional distribution center
remote DSLAM
release
RF
RFC
RFI
RFP
RFPON
RFQ
RGU
RGW
RHC
RIAC
RIP
RISC
RJ
RLL
RM
RMA
RMI
RMON
ROADM
ROBO
ROI
ROW
RPC
RPF
RPR
RPRA
RPT
RQMS
RRQ
RSU
RSVP
RSVPTE
RT
RTCP
RTOS
RTP
RTSP
RTU
RxTx
RZ
SAM
SAN
SAP
SAR
S-band
SBS
SCAN
SCCP
SCCS
SCE
SCF
radio frequency
request for comment
request for information
request for proposal
radio frequency optical network
request for quotation
revenue-generating unit
residential gateway
regional holding company
remote instrumentation and control
routing information protocol
reduced instruction set computing
registered jack
radio in the loop
resource management
request for manual assistance
remote method invocation
remote monitoring
reconfigurable optical add/drop multiplexer
remote office/branch office
return on investment
right of way
remote procedure call
reverse path forwarding
resilient packet ring
Resilient Packet Ring Alliance
resilient packet transport
requirements and quality measurement system
round-robin queuing or registration request
remote service unit
resource reservation protocol
resource reservation protocoltraffic engineering
remote terminal
real-time conferencing protocol
real-time operating system
real-time transport protocol
real-time streaming protocol
remote test unit
receiver/transmitter
return to zero
service access multiplexer
storage-area network
service access point OR session announcement protocol
segmentation and reassembly
short band
stimulated Brillouin scattering
switched-circuit automatic network
signaling connection control part
switching control center system
service-creation environment
service control function
503
SCL
SCM
SCN
SCP
SCR
SCSI
SCSP
SCTP
SD
SD&O
SDA
SDB
SDC
SDF
SDH
SDM
SDN
SDP
SDRP
SDSL
SDTV
SDV
SE
SEC
SEE
SEP
ServReq
SET
SFA
SFD
SFF
SFGF
SG
SG&A
SGCP
SGSN
SHDSL
SHLR
SHV
SI
SIBB
SIC
SICL
SID
SIF
sigtran
SIM
SIP CPL
SIP
SIPT
504
SISO
SIU
SIVR
SKU
SL
SLA
SLC
SLEE
SLIC
SLO
SM
SMC
SMDI
SMDS
SME
SMF
SML
SMP
SMPP
SMS
SMSC
SMTP
SN
SNA
SNAP
SNMP
SNPP
SNR
SO
SOA
SOAC
SOAP
SOCC
SOE
SOHO
SON
SONET
SOP
SP
SPC
SPE
SPF
SPIRITS
SPIRITSG
SPM
SPOP
SPX
SQL
SQM
SRF
SRP
SRS
srTCM
SS
SS7
SSE
SSF
SSG
SSL
SSM
SSMF
SSP
STE
STM
STN
STP
STR
STS
SUA
SVC
SW
SWAN
SWAP
SWOT
SYN
TALI
TAPI
TAT
Tb
TBD
Tbps
TC
TCAP
TCB
TCIF
TCL
TCM
TCO
TCP
TCP/IP
TCPAM
TDD
TDM
TDMA
TDMDSL
TDOA
TDR
TE
TEAM
TED
TEM
506
TFD
THz
TIA
TIMS
TINA
TINA-C
TIPHON
TIWF
TKIP
TL1
TLDN
TLS
TLV
TMF
TMN
TMO
TN
TNO
TO&E
TOD
TOM
ToS
TP
TPM
TPSTC
TR
TRA
TRIP
trTCM
TSB
TSC
TSI
TSP
TSS
TTC
TTCP
TTL
TTS
TUI
TUP
TV
UA
UADSL
UAK
UAWG
UBR
UBT
UCP
UCS
UDDI
UDP
toll-free dialing
terahertz
Telecommunications Industry Association
transmission impairment measurement set
Telecommunications Information Networking Architecture
Telecommunications Information Networking Architecture Consortium
Telecommunications and Internet Protocol Harmonization over Networks
trunk interworking function
temporal key integrity protocol
transaction language 1
temporary local directory number
transparent LAN service OR transport-layer security
tag length value
TeleManagement Forum
telecommunications management network
trans-metro optical
telephone number
telecommunications network operator
table of organization and equipment
time of day
telecom operations map
type of service
twisted pair
transaction processing monitor
transmission control specifictransmission convergence
technical requirement OR tip and ring
technology readiness assessment
telephony routing over Internet protocol
two-rate tri-color marker
telecommunication system bulletin
terminating call screening
time slot interchange
telecommunications service provider
Telecommunications Standardization Section
Telecommunications Technology Committee
test TCP
transistor-transistor logic
text-to-speech OR TIRKS table system
telephone user interface
telephone user part
television
user agent
universal ADSL
user-authentication key
Universal ADSL Working Group
unspecified bit rate
ubiquitous bus technology
universal computer protocol
uniform communication standard
universal description, discovery, and integration
user datagram protocol
507
UDR
UI
ULH
UM
UML
UMTS
UN
UNE
UNI
UOL
UPC
UPI
UPS
UPSR
URI
URL
USB
USTA
UTOPIA
UTS
UWB
UWDM
V&H
VAD
VAN
VAR
VAS
VASP
VBNS
VBR
VBRnrt
VBRrt
VC
VCC
VCI
VCLEC
VCO
VCR
VCSEL
VD
VDM
VDSL
VeDSL
VGW
VHE
VHS
VITA
VLAN
VLR
VLSI
VM
508
VMS
VoADSL
VoATM
VoB
VoD
VoDSL
VoFR
VoIP
VON
VoP
VOQ
VoT1
VP
VPDN
VPI
VPIM
VPN
VPR
VPRN
VRU
VSAT
VSI
VSM
VSN
VSR
VT
VTN
VToA
VVPN
VXML
W3C
WAN
WAP
WATS
WB DCS
WCDMA
WCT
WDCS
WDM
WECA
WEP
WFA
WFQ
WIM
WiMAX
WIN
WLAN
WLL
WMAP
WML
WNP
voice-mail system
voice over ADSL
voice over ATM
voice over broadband
video on demand
voice over DSL
voice over frame relay
voice over IP
voice on the Net
voice over packet
virtual output queuing
voice over T1
virtual path
virtual private dial network
virtual path identifier
voice protocol for Internet messaging
virtual private network
virtual path ring
virtual private routed network
voice response unit
very-smallaperture terminal
virtual switch interface
virtual services management
virtual service network
very short reach
virtual tributary
virtual transport network
voice traffic over ATM
voice virtual private network
voice extensible markup language
World Wide Web Consortium
wide-area network
wireless application protocol
wide-area telecommunications service
wideband DCS
wideband CDMA
wavelength converting transponder
wideband digital cross-connect
wavelength division multiplexing
wireless Ethernet compatibility alliance
wired equivalent privacy
work and force administration
weighted fair queuing
wireless instant messaging
Worldwide Interoperability for Microwave Access
wireless intelligent network
wireless local-area network
wireless local loop
wireless messaging application programming interface
wireless markup language
wireless local number portability
509
WRED
WS
WSP
WTA
WUI
WVPN
WWCUG
WWW
XA
XC
XD
xDSL
XML
XNS
XPM
XPS
xSP
XT
XTP
Y2K
510