Telecommunication Network Design by Jorma Kekalainen

Telecommunication Network Design
Jorma Kekalainen
Course Objective
Understand principal design concepts of
telecommunication networks
Information Sources
1.
Usually, the most precise sources are the original

sources, i.e. standards, recommendations or other
specifications.
You can pull them from the Internet e.g.
ITU-T www.itu.int/ITU-T/
IETF www.ietf.org
3GPP www.3gpp.org
or from elsewhere.
Reading just the general sections in the
specifications and scanning the rest will be usually
enough to get sufficient understanding
2. You can look for material from corresponding
courses in the Internet
3. Some may find that the books are easier to read.
4. Lecture notes are sufficient to pass the course.
Books
Alwayn: Optical Network Design and Implementation
Anandalingam, G.; Raghavan, S. (Eds.): Telecommunications Network
Design and Management
Bose: Breakthrough Perspectives in Network and Data Communications
Security, Design and Applications
Freeman, Telecommunication System Engineering
Graham, Kirkman, Paul: Mobile Radio Network Design in the VHF and
UHF Bands: A Practical Approach
Jurdak: Wireless Ad Hoc and Sensor Networks: A Cross-Layer Design
Perspective (Signals and Communication Technology)
McCabe: Network Analysis, Architecture, and Design
Mishra: Fundamentals of Cellular Network Planning and Optimisation
Olifer N., Olifer V.: Computer Networks: Principles, Technologies and
Protocols for Network Design
Oppenheimer: Top-Down Network Design
Paquet: Campus Network Design Fundamentals
Shooman: Reliability of Computer Systems and Networks: Fault
Tolerance, Analysis, and Design
Spohn: Data Network Design
4

Basic Design Concepts
What is telecommunication network?

A network is a system in which multiple nodes are connected by
communications links.
Node can refer to a switch/router/terminal attached to a
network
A network permits a node to communicate with another node.
Telephone networks and computer networks are
telecommunication networks
The manner in which network nodes are connected is termed
topology
Classification of networks (1)

The switching technique is used to classify
networks
Switching networks
circuit switched or
packet switched
Broadcast networks
Bus
Ring
Star
The switching technique and topology of a network can strongly
impact its performance and robustness.
Networks have performance attributes throughput, error
rates, delay etc.
7
Classification of networks (2)

WAN Wide Area Network spanning national, state,
or municipal boundaries.
MAN Metropolitan Area Network confined to a
metropolitan area, using common protocols.
LAN Local Area Network confined to a room,
floor, building or campus.
PAN Personal Area Network diameter about 1m or
confined to a room.
The boundaries are now becoming increasing blurred
between the categories, e.g. when is the network
a MAN or WAN?
a PAN or LAN?
The long term trend is for universal broadband

connectivity between all these networks.
What is Design?
Dictionary: "Design" refers to the process of
originating and developing a plan or proposal (a
drawing, model, or other description) for a new
object (machine, building, product, etc.).
Designing normally requires considering functional and
many other aspects of an object, which usually
requires considerable
research,
thought,
modeling,
iterative adjustment, and
re-design.
Design as a process can take many forms depending

on the object being designed and individuals
participating it.
10
Design process
In systems engineering and networking, design is a
very systematic process, performed step by step.
The technology in use will usually impose limits on
creativity for the designer.
In many engineering disciplines, design is codified by
standards or design rules, which impose safety
margins and guidelines in the design - This is to
protect against poor design and failure.
A designer must often invest considerable
time,
effort and
resources
to establish a good design for a system.
11
What is Network Design?

Traditionally based on a set of general
rules
80/20
Bridge when you can, route when you must
Focused on capacity planning

No consideration in delay optimization
No guarantee of service quality
Lack of redundancy
A bridge is a device that passes information between two network segments. It

operates at Layer 2 (Data Link Layer) of the OSI reference model.
12
System and systems approach

A network is a system the set of
components that support or provide
connectivity, communications and
network services to users.
In network design, we
Identify components of a system
Understand how they interface with each
other
Network design can then interpret

system planning.
13
Systems Approach for Network

Design
14
What is a communications link?

All communications links comprise
a transmitter,
a channel and
a receiver.
A channel might be
a radio or optical link through the atmosphere,
an optical fibre,
a coaxial cable,
a twisted pair cable, or even
a telephone wire on a pole.
Links have performance attributes

throughput/capacity, error rates, bandwidth etc.
The type of channel determines the design of the
transmitter and the receiver.
15
Elements of a communication system

Send messages or information from a
source to one or more destinations
16
Example: Fast Ethernet LAN

Wavelength used 850 nm
Fiber loss 2.5 dB/km
Four connectors, no splices
Assume BER is 10-9 (one
error for every 109 bits sent)
Silicon pin detector requires
at 100 Mb/s at least -32
dBm (or 630 nW) input
optical power for BER=10-9
Transmitter (LED) couples 20 dBm (10 W) of optical
power into the fibre
Connector loss max 0.7dB
per connector
for 10-11 BER
17
Link design Power budget

Component / loss parameter Output/sensitivity/ Power margin
(dB)
loss
LED output
-20 dBm
PIN sensitivity at 100Mbps
-32 dBm
Allowed loss PT=PS-PR
[-20 - (-32)] dB
12
Source connector loss
0.7 dB
11.3
2 Jumper connector loss
1.4 dB
9.9
Cable attenuation (160 m)
0.4 dB
9.5
Receiver connector loss
0.7 dB
8.8 dB Final
margin
The final power margin is 8.8 dB, which is a sufficient margin for this link.
18
Note. ITU-T Recommendation sets a link margin between 3.0 and 4.8 dB
Measures of network performance

There are various ways we can measure the
performance of a network:
1.
2.
3.
4.
Data throughput or capacity.

Propagation delay between nodes; variability of delay
Error rates per packet, per bit.
Reliability the probability that it will not fail over a finite
time interval.
5. Maintainability time to repair faults, time to effect
upgrades.
6. Security & secrecy, e.g. secure, shielded transfer
the ability to resist unwanted eavesdropping,

message goes to the right receiver,
no intrusion or theft of capacity.
19
Common problem issues for networks

Poor design for performance results in poor response
times on connections.
Poor design for reliability results in frequent outages.
Poor design for growth results in frequent downtime
to effect upgrades.
Poor security results in theft of capacity or user
data.
Poor choices in technology results in early
obsolescence and replacement.
Poor compatibility with end user interfaces results in
downtime, errors and poor performance.
20
Iterative Analysis/Design process

Analyze requirements.
Collect user data.
Analyze future
requirements.
Develop the logical design.

Topology, addressing,
naming, routing, security &
management planning.
Develop the physical design.

Choose technologies,
devices, cable; choose
service providers.
Test, optimize, and document

the design.
Implement a test plan, build
a prototype and document
your work with a network
design proposal.
21
Analysis and Design Processes

Set and achieve goals
Maximizing performance
Minimizing cost
Optimization with trade-offs

Performance analysis or simulation
Recognizing trade-offs
No single best answer
Hierarchies
Provide structure in the network
Redundancy
Provides availability & reliability
22
General Design Approaches

Heuristic by using various principles and
algorithms
Exact by working out mathematical
solutions based on some optimization
method and minimizing certain cost
functions
Simulation often used when no exact
analytical form exists.
Experiments are conducted on simplified
models to see the performance of network
23
Pillars of Science
Traditional scientific and engineering
approach:
Rationalism
Deduction and
Mathematics
Empiricism
Induction and
Observation (experiments)
Simulation
24
How to inspect a system?
25
What is simulation and network simulation?

Simulation
The process of designing a computerized model of
a system (or process) and conducting experiments
with this model for the purpose either of
understanding the behaviour of the system or of
evaluating various strategies for the operation of
the system.
Network Simulation
Simulation is the process of using software and
mathematical models to analyze the behavior of a
network without requiring an actual network.
26
Reasons for network simulation

Need to analyze
Impacts of different network designs
Topologies
Technologies
Protocols
Algorithms
Network configurations
Technology benchmark and comparison
Gain understanding
27
Reasons for network simulation (2)

Need to test
Reliability
Performance
Scalability
Reducing risks before actual deployment.
A simulation tool is more preferable

than to implement an extensive
prototype system as it allows
alternatives to be easily compared.
28
Art + Science
The Art of Network Design
Communication with users
Relations to business goals
Technology choices
The Science of Network Design

Understanding of network technologies
Analysis of capacity, redundancy, delay
29
Example: Scalability of architectures

Here we compare the scalability of client-server architecture
with P2P architecture in the context of distributing a large file
from a single server to a large number of hosts (peers).
In client-server file distribution, the server must send a copy of
the file to each of the peersplacing an heavy burden on the
server and consuming a large amount of server bandwidth.
In P2P file distribution, each peer can redistribute any portion
of the file it has received to any other peers, thereby assisting
the server in the distribution process.
We make the simplifying but generally valid assumption that the
network core has abundant bandwidth, implying that all of the
bottlenecks are in network access.
We also suppose that the server and clients are not
participating in any other network applications, so that all of
their upload and download access bandwidth can be fully devoted
to distributing this file.
30
File distribution
Let us consider a simple quantitative model for
distributing a file to a fixed set of peers for both
architecture types.
The server and the peers are connected to the
network with access links.
Denote the size of the file to be distributed (in bits)
by F and the number of peers that want to obtain a
copy of the file by N.
The distribution time is the time it takes to get a
copy of the file to all N peers.
In the client-server architecture, none of the peers
aids in distributing the file.
31
File Distribution: Server-Client vs P2P

Question : How much time is taken to distribute
file from one server to N peers?
us: server upload
bandwidth (rate)
Server
us
u1
d1
u2
ui: peer i upload

bandwidth
d2
File, size F
dN
uN
di: peer i download

bandwidth
Network (with
abundant bandwidth)
32
Observations (1)
We make the following observations:
The server must transmit one copy of the file to
each of the N peers.
Thus the server must transmit NF bits.
Since the servers upload rate is us, the time to
distribute the file must be at least NF/us.
Let denote the download rate of the peer with the
lowest download rate, that is, dmin = min{d1,d2,...,dN}.
The peer with the lowest download rate cannot
obtain all F bits of the file in less than F/dmin
seconds.
Thus the minimum distribution time is at least
F/dmin.
33
File distribution time: server-client

Server
server sequentially
sends N copies:
NF/us time
client i takes F/di
time to download
F
us
dN
u1 d1 u2
d2
Network (with
abundant bandwidth)
uN
increases linearly in N
(for large N)
Time to distribute F
to N clients using = dcs = max { NF/us, F/min(di) }
i
client/server approach
This provides a lower bound on the minimum distribution
34
time for the client-server architecture.
File distribution time: P2P architectures

Lets now go through a similar analysis for the P2P
architecture, where each peer can assist the server
in distributing the file.
In particular, when a peer receives some file data, it
can use its own upload capacity to redistribute the
data to other peers.
Calculating the distribution time for the P2P
architecture is somewhat more complicated than for
the client-server architecture, since the distribution
time depends on how each peer distributes portions
of the file to the other peers.
Nevertheless, a simple expression for the minimal
distribution time can be obtained.
35
Observations (2)
1) At the beginning of the distribution, only the server has the file.
To get this file into the community of peers, the server must send
each bit of the file at least once into its access link.
Thus, the minimum distribution time is at least F/us.
2) As with the client-server architecture, the peer with the lowest
download rate cannot obtain all F bits of the file in less than F/dmin
seconds.
Thus the minimum distribution time is at least F/dmin
3) Finally, observe that the total upload capacity of the system as a
whole is equal to the upload rate of the server plus the upload rates
of each of the individual peers, that is, utotal =us + u1 + ... + uN.
The system must deliver (upload) F bits to each of the N peers, thus
delivering a total of NF bits.
This cannot be done at a rate faster than utotal .
So, the minimum distribution time is also at least NF/(us+u1+...+uN).
36
Minimum distribution time of P2P

Putting these three observations together, we obtain the
minimum distribution time for P2P, denoted by
dP2P max { F/us, F/min(di) , NF/(us + ui) }

i
Equation provides a lower bound for the minimum distribution

time for the P2P architecture.
It turns out that if we imagine that each peer can redistribute a
bit as soon as it receives the bit, then there is a redistribution
scheme that actually achieves this lower bound.
In reality, where chunks of the file are redistributed rather
than individual bits, equation serves as a good approximation of
the actual minimum distribution time.
dP2P max { F/us, F/min(di) , NF/(us + ui) }

i
37
Numerical example
The following figure compares the minimum distribution time for
the client-server and P2P architectures assuming that all peers
have the same upload rate u.
For the sake of simplicity, we set F/u = 1 hour, us =10u, and dmin
us.
Thus, a peer can transmit the entire file in one hour, the server
transmission rate is 10 times the peer upload rate, and (for
simplicity) the peer download rates are set large enough so as
not to have an effect.
We see from the following figure that for the client-server
architecture, the distribution time increases linearly and
without bound as the number of peers increases.
However, for the P2P architecture, the minimal distribution time
is not only always less than the distribution time of the clientserver architecture; it is also less than one hour for any
number of peers N.
Thus, applications with the P2P architecture can be self-scaling.
This scalability is a direct consequence of peers being
redistributors as well as consumers of bits.
38
File distribution time: P2P

Server
server must send one copy:

F/us =F/(10u) time
client i takes F/di time to
download, so
F/min(di)F/us=F/(10u)
NF bits must be downloaded
(aggregate)
F
us
dN
u1 d1 u2
d2
Network (with
abundant bandwidth)
uN
fastest possible upload rate:
us + ui=(10+N)u Nu, when N>>
dP2P = max { F/us, F/min(di) , NF/(us + ui) }

i F/u}, N>>
= max {F/(10u), F/(10u),
39
Server-client vs. P2P: example

Client upload rate = u, F/u = 1 hour, us = 10u, dmin us
Minimum Distribution Time
3.5
P2P
Client-Server
3
2.5
dsc=NF/(10u)
2
1.5
dP2P=NF/[(10+N)u]
1
0.5
0
0
10
15
20
25
30
35
N
40
Example: Caching
Assumptions
average object size = 100,000
bits
avg. request rate from
institutions browsers to origin
servers = 15/sec
delay from institutional router
to any origin server and back
to router = 2 sec
Consequences
utilization on LAN = 15%
utilization on access link =
100%
total delay = Internet delay
+ access delay + LAN delay
= 2 sec + minutes + milliseconds
origin
servers
public
Internet
1.5 Mbps
access link
institutional
network
10 Mbps LAN
41
Example: Caching (cont)

origin
servers
possible solution
increase bandwidth of
access link to, say, 10
Mbps
public
Internet
consequence
utilization on LAN = 15%
utilization on access link = 15%
Total delay = Internet delay
+ access delay + LAN delay
= 2 sec + msecs + msecs
often a costly upgrade
10 Mbps
access link
institutional
network
10 Mbps LAN
42
Example: Caching (cont)

origin
servers
possible solution: install

cache
suppose hit rate is 0.4
consequence
40% requests will be
satisfied almost immediately
60% requests satisfied by
origin server
utilization of access link
reduced to 60%, resulting in
negligible delays (say 10
msec)
total avg delay = Internet
delay + access delay + LAN
delay = .6*(2.01) secs +
.4*milliseconds < 1.4 secs
public
Internet
1.5 Mbps
access link
institutional
network
10 Mbps LAN
institutional
cache
43
Design of communication systems (1)

Major goals of communication systems designs are
1.
2.
3.
4.
5.
To maximize transmission bit rates

To minimize probability of bit error
To minimize required signal power
To minimize required system bandwidth
To maximize system utilization, e.g., maximum number of
users with minimum delay and maximum resistance against
interference
6. To minimize system complexity and cost.
Note that goals 1 and 2 are in conflict with goals 3

and 4.
Note that goal 5 is conflicting with itself
Note that goal 6 is conflicting with all other goals
44

In practice, we also have some
constraints such as
Nyquist minimum bandwidth
Shannon-Hartley channel capacity
Government regulations (e.g., frequency
allocation, power limit)
Technological limitations (e.g., electronics)
Other system requirements (e.g., wireless,
copper, satellite)
45

Nyquist minimum bandwidth:
R symbols per second could be detected without ISI in an
R/2 Hz minimum bandwidth (called Nyquist bandwidth)
Capacity of ideal Gaussian channel

In the presence of additive white Gaussian noise, an ideal
bandlimited channel of bandwidth W has a capacity C given
by
P: Average transmitted power

N0: Power-spectral density of the additive noise
If the information rate R from the source is less than C (R < C),
it is theoretically possible to achieve reliable transmission
through the channel by means of an appropriate coding
46
Network designers viewpoint

The network is a collection of interconnected devices
which carry data between user equipment.
The network has performance characteristics,
reliability (uptime/downtime), and quality metrics.
The network has to be continuously monitored and
maintained to assure service quality and availability.
The network has to be continuously upgraded to meet
growing user demands for load.
The security of the network has to be maintained.
Hardware, software and channels have to be
integrated together to deliver the desired effect.
The designer sees the network from a technology
perspective and is concerned with its function.
47
End users viewpoint

To the end user the network is a visible as a socket
on the wall or an antenna on a portable device.
The end user perceives the throughput of the
network as increasing or decreasing delays or
response times in applications.
The end user perceives the availability of the
network as the presence or absence of connections,
or the ability or inability to get her/his work done.
The basic question the end user most often asks is
does this network work?
is it running fast enough?
how expensive is it to use this network?
48

General design approach
Design Strategies
What is the basic networking
technology used?
What is the characteristic topology?
What is the cabling or wireless link
technology?
Limitations vs. Strengths?
50
How do I design a network?

1. Identify constraints user needs, budgets,
technology
2. What legacy infrastructure can I exploit cables
etc?
3. Top down approach model loads vs performance to
define link / switch / router capacities
4. Design for evolution project: future growth and
upgrades.
5. Design for maintainability, reliability, supportability
51
Maintainability, Reliability, Supportability

Maintainability is a measure of how easily components can be
repaired or replaced.
Maintainability impacts running costs via man-hours expended
per maintenance action.
Reliability is a measure of how frequently (and severely) the
network breaks down requiring repair.
Reliability impacts running costs via the frequency and severity
of maintenance actions.
Supportability is a measure of how easily replacement
components can be acquired to maintain a network, and becomes
increasingly difficult as equipment become obsolete and parts
difficult to source.
Note: These three items are critical design constraints, but are
often absent in statements of performance and capability need.
A good designer must account for them.
52
Design Process (1)

The first step in any design is to establish and
understand what the system must achieve in
terms of capability and performance.
Demand for network performance usually grows
over time.
Good design allows for low cost incremental
growth in capacity as the load and technology
evolves over time.
Often testing and measurement is required to
establish or validate capability and performance
needs.
53
Design Process (2)

The second step in the design process is to survey
the technology available which can be used to
implement a design.
This means collecting information on the
performance, capabilities and costs of all of the
components which might be used in the design.
For a network this means routers, switches, cables,
management software, and other equipment.
Knowing the strengths and weaknesses of specific
products is critical to success.
Vendors will inflate strengths and diminish or conceal
weaknesses or limitations.
Often testing is necessary to validate survey analysis,

because limitations of components may not be obvious
54
from product documentation or marketing brochures.
Design Process (3)

Once we have understood the demands / needs /
requirements for the product, and the capabilities of
the components available for use in a design, we are in
the position to consider specific designs.
The third step in the design process is to define a
conceptual design which accounts for the
performance / capability needs, and the available
technology.
The conceptual design lacks the detail of a final
design, but is detailed enough to construct a
performance model for the product.
The performance model reconciles demands / needs /
requirements against components.
55
Design Process (4)

Once we have a performance model we can perform
simulations to establish whether the conceptual
design is good enough, or too good.
Design is usually iterative, since the mathematical
modeling of such systems are usually intractable and
finding a perfect solution in one pass is impossible.
The conceptual design is repeatedly adjusted, and
tested by simulation, until an acceptable result is
found.
Often a conceptual design cannot meet needs, and
must be replaced completely.
In most designs, performance must be reconciled
against cost.
56
Design Process (5)

Once an acceptable conceptual design is established,
we can proceed to the detailed design.
In a detailed design, assumed system components,
such as generic routers, switches etc, are replaced
with real components.
Simulations and testing are then required to validate
the detailed design, and confirm that it can fit the
performance model which was met by the conceptual
design.
Often the largest amount of effort expended in the
design process is in the detailed design.
When the detailed design is completed, we have a
product which can be implemented.
57
Iterative Design Process

Demand
Resources
Goals
Design
Activity
Network
Design
Adjust Criteria
Review Results
Design Documents
Specns
Drawings
Plans
Re-design
58
Planning and design
59
Planning and design
60
Analysis
61
Design
62
Implementation
63
Analysis/Design
Analysis processes
Requirements
analysis
Flow analysis
Setting out
blueprints
Design processes
Logical design
Physical design
Routing
Addressing
Implementation and
Test
64
Implementation
Implementation involves purchase of the required
hardware, software, services required to construct
the network.
Once the network is constructed, it must be
acceptance tested to confirm performance.
Acceptance testing usually exposes mistakes in design
and implementation, and is absolutely critical to
success.
Once the network has been tested and commissioned,
it can be handed over to end users.
65
Constraints and Budgets

The first two steps in the design process are
collecting and analyzing information on constraints
upon the design.
Constraints are hard limits you cannot cross without
penalty.
Constraints might be the performance of available
equipment, rental costs on high speed links.
The task of the designer is to find some
configuration of components which fits inside the
constraints.
Most often the budget is the biggest constraint and
the cause of most design failures.
Designers are often asked to produce designs which
cannot be implemented within the available budget.
66
Legacy Infrastructure
More than often a new network design is a
replacement for an existing design.
This may present opportunities to save a lot
of money, or force significant additional
expenditure.
Where existing cabling, racks, air
conditioning, and rooms may be reused, new
replacements are not needed and money is
saved.
Often legacy infrastructure is not adequate
and must be replaced partly or completely,
increasing costs.
67
Top Down Strategy vs. Bottom Up Strategy

A top down strategy in a design means that the
designer starts with a system level model, and works
down into the details progressively.
A bottom up strategy is where the designer first
concentrates on the detail design and then attempts
to construct the system level design.
Some design problems force a bottom up strategy,
but network design is usually not such a problem.
Bottom up design is risky since required system level
performance may be impossible to achieve.
Inexperienced designers often pursue bottom up
design strategies since this postpones difficult
modeling tasks.
68
Questions (1)
How to meet the requirements of the transport
application (like accuracy, throughput, latency,
mobility support...)?
How to represent and use the information?
How to utilize the communication medium?
How to connect users?
How to reach one point from another?
How to coordinate among the transmitters and
receivers?
How to regulate competition among users?
How to make the system robust to failures, attacks,
variations, growth across space and over time?
How to allocate functionalities to layers?
69
Questions (2)
How to describe the channel and estimate its
characteristics (twisted pair, coaxial cable, optic
fiber, radio)?
How fast can data be sent reliably?
How to compress signals?
How to reduce noise (thermal noise, impulse noise ...)
and manage interference (from other users, from
reflections, among symbols)
How to use the communication resources (time,
frequency) efficiently?
What happens when multiple transmitters send data
to multiple receivers?
70
Questions (3)
What topology? Who are transceivers and who are
relays?
Direct link or switched architecture? Circuit switch
or packet switch or something else?
How to divide into (possibly different types of)
subnetworks?
End-to-end control or hop-by-hop control?
How to get on the communication medium?
How to get from one point to another?
How to monitor and adjust overall state of the
network?
How to ensure accurate, secure, timely, and usable
transfer of information across space among
competing users?
71
Customer Needs Drive Basic Decisions

Customer Has:
Needs & expectations
Money
Traffic volumes and characteristics

Functionality and Acceptable Delays
To
Pay
Technical Design
Equipment sizing and numbers
Circuit Dimensioning
Minimises
COSTS!
72
Roles
Customer Has:
Money to pay for products & services
Business Needs & Expectations
Employees (Operators and Users)
Designer Has:
Knowledge
(education &training)
Wisdom
(experience)
Vendors Have:
Products & Services
Skilled Workforce
All are needed

to implement
and support the
network
73
Common design techniques

Key concept: bottleneck
the most constrained element in a system
System performance improves by removing

bottleneck
but creates new bottlenecks
In a balanced system, all resources are

simultaneously bottlenecked
this is optimal
but nearly impossible to achieve
in practice, bottlenecks move from one part of the
system to another
74
example: Ford Model T

Some aspects of TCP/IP design
Jorma Kekalainen
The Internet
What is the Internet?
physical infrastructure
architecture
protocols
software
services/applications
operational practices
standards
All of the above!

76
Standards
Why do we need standards?
electricity plugs
Plugs ore standardized, but only within a country.
The Internet is an international network

need standards between countries
everyone has to agree on one plug
Instead of plugs we standardize protocols

still need plugs, but these are physical layer
a protocol is a more general concept
77
Network Standards Bodies

ISOC (Internet Society)
IESG (Internet Engineering Steering Group)
IETF (Internet Engineering Task Force)
IAB (Internet Architecture Board)

IRSG (Internet Research Steering Group)
IRTF (Internet Research Task Force)
ICANN (Internet Corp. f or Assigned Names and

Numbers) and lANA (Internet Assigned Numbers
Authority)
W3C (WWW standards)
IEEE (Inst. of Electrical and Electronic Engineers)
e.g. IEEE 802.3 (Ethernet)
CCITT, ITU-T (International Telecommunications

Union)
ANSI, OSI (Open System Interconnection)
78
IETF
informal standards body
membership is open to all interested
individuals
few hard and fast rules
publishes RFCs (Request For Comments)
RFC 791: Internet Protocol (IP) Updated in RFC
1391
RFC 793: Transmission Control Protocol (TCP)
Updated in RFC 3168
79
Internet Design Principles

packet switching not circuit switching:
dont reserve bandwidth for a connection
robustness principle:
Be liberal in what you accept, and conservative in what you
send
layered model with a thin waist

end-to-end principle:
smart terminals, dumb network
distributed control
deployment issues:
scale, incremental deployment, heterogeneity
general issues:
simplicity, modularity, performance
80
Packets vs circuits
Some (Bell-heads) believe you need a
dedicated circuit
like a phone line (but higher speed)
Others (Net-heads) think circuits are

a waste of time and money
poor use of resources when traffic is
bursty
81
Packets vs circuits
Circuit switching:
logical equivalent of a phone line connects two (or
more) people.
allows network to control everything (in theory)
allows explicit QoS
needs careful design and admission control
prime example is ATM
Packet switching:
no physical circuit (though there is still an analogue
of a connection)
packets of data are individually switched.
network doesnt do much (in theory)
hard to do QoS, but network is simpler
prime example is IP
82
Packets vs circuits
Doesnt have to be one
or the other
people may run circuit
switched on one layer, and
packet switched on
another.
classic example is IP over
ATM
MPLS creates virtual

circuits between endpoints
though connections are
not between end-users
allows multiplexing of
traffic inside a
connection
multiplexed traffic is less
bursty
Multiprotocol Label Switching (MPLS) is a

mechanism in high-performance
telecommunications networks which directs
and carries data from one network node to
the next. MPLS makes it easy to create
"virtual links" between distant nodes. In an
MPLS network, data packets are assigned
labels. Packet-forwarding decisions are made
solely on the contents of this label, without
the need to examine the packet itself. This
allows one to create end-to-end circuits
83
across any type of transport medium, using
any protocol.
Robustness principle
Be liberal in what you accept, and
conservative in what you send.
if somebody else screws up, dont let this
mess your system up (liberal in what you
accept)
e.g. TCP connection termination
dont cause other systems problems

(conservative in what you send)
e.g. congestion control
84
Layered Architecture
Divide and conquer:
break the overall big
problem into smaller
ones with standardized
interfaces
Each layer provides a
service to upper layers
and utilizes the services
provided by lower layers
Performance may not be
optimal, but makes the
architecture simple and
flexible
85
Layered protocols: OSI model

OSI model breaks functionality into
layers called a protocol stack
86
Layered protocols: TCP/IP model
User Datagram Protocol (UDP), a simple transport protocol used in the Internet. The
Internet Control Message Protocol (ICMP) is chiefly to send error messages
indicating, for instance, that a requested service is not available or that a host or
router could not be reached. The Address Resolution Protocol (ARP) is a protocol for
87
determining a network host's link layer or hardware address when only its Internet
Layer (IP) or Network Layer address is known. Packet over SONET (POS)
Layered protocols
Somewhat like subroutines in programming
Each layer provides services (functions) to higher
layers
Function call interface hides details of how the
service is provided
e.g. network layer asks link layer to transport a packet
across a link, without any network details
the interface is well defined
benefits
reduction in complexity
Communications between peers using

protocols
88
Encapsulation
Lower layers deal with higher layer by
treat information from higher layer as black box
dont look inside data
just treat as bunch of bits
just break data into blocks

encapsulate the blocks, by adding
headers (e.g. addresses)
trailers
when passing back to higher layer

layers strip headers
join blocks back together
89
Layer 1: Physical layer

Function: Transmission of raw bit stream
between devices.
Services: Physical connection, modulation
Issues: # pins/wires, duplex, serial/parallel,
modulation
Media:
copper wire: e.g. coax, twisted pair
fibre optics
free air optics
microwave, satellite,
infra-red
90
Layer 2: Link layer

Function: provide reliable transport of information
between a pair of adjacent nodes.
Services: creates frames/packets, error control, flow
control
Issues: Medium Access Control (MAC),
headers/trailers
Examples:
Ethernet
Token-ring
IEEE 802.11 (Wi-Fi)
FDDI (Fiber Distributed Data Interface)
ATM (Asynchronous Transfer Mode) (also layer 3)
POS (Packet over SONET)
PPP (Point to Point Protocol)
The Medium Access Control (MAC) data communication protocol sub-layer is a sublayer of the Data
Link Layer. It provides addressing and channel access control mechanisms that make it possible for
91
several terminals or network nodes to communicate within a multipoint network, typically a local area
network (LAN) or metropolitan area network (MAN).
Layer 3: Network layer

Function: forwarding packets from end-to-end
Services: packet forwarding, some congestion
control
Issues: determining what routing to use
Examples:
IPv4 (Internet Protocol version 4)
IPv6 (Internet Protocol version 6)
ARP (Address Resolution Protocol)
ATM (Asynchronous Transfer Mode) (also layer 2)
Routing protocols (e.g. OSPF, RIP, EIGRP)
92
Routing details
OSPF - Open Shortest Path First
Open Shortest Path First is a link state (each node possesses information
about the complete network topology), hierarchical IGP (Interior Gateway
Protocol) routing algorithm. Features supported by OSPF include least cost
routing, multipath routing and load balancing.
RIP - Routing Information Protocol

Routing Information Protocol is used to manage router information within a
self contained network such as a corporate LAN (Local Area Network) or an
interconnected group of such LAN. RIP is classified by the IETF (Internet
Engineering Task Force) as one of several internal gateway protocols called
IGP (Interior Gateway Protocols).
IGP - Interior Gateway Protocol

An Interior Gateway Protocol is used to exchange routing information within
an autonomous system.
AS Autonomous System
Autonomous system (Internet), a collection of IP networks and routers
under the control of one entity (typically an Internet service provider or a
very large organization with independent connections to multiple networks,
that adhere to a single and clearly defined routing policy)
EIGRP - Enhanced Interior Gateway Routing Protocol

An advanced version of Interior Gateway Routing Protocol. It provides
superior convergence properties and operating efficiency, and combines the
advantages of link state protocols with those of distance vector protocols
(in distance-vector routing protocols each router does not possess
information about the full network topology).
93
Routing vs. Forwarding

We need to make a distinction
routing = routing protocols build the routing
tables
distributed routing protocol
build forwarding/routing/lookup table
forwarding = send packets to their next hop

lookup destination address in forwarding table
forward packet
94
Layer 4: Transport layer

Function: reliable end-to-end transport of
data
Services: multiplexing, end-to-end error and
flow control
Issues: congestion control algorithm
Examples:
TCP (Transmission Control Protocol)
UDP (User datagram Protocol)
SCTP (Stream Control Transmission Protocol)
RTP (Real-time Transport Protocol)
95
SCTP - Stream Control Transmission

Protocol
Stream Control Transmission Protocol is a
reliable transport protocol operating on top
of IP .
It provides acknowledged error free non
duplicated transfer of data with flow control.
STCP also detects data corruption, loss of
data and duplication of data by using
checksums and sequence numbers.
A selective retransmission mechanism is
applied to correct loss or corruption of the
data.
96
Layer 5: Application layer

E-mail
File transfer (FTP File Transfer Protocol)
Remote terminal (Telnet)
WWW (HTTP Hyper-Text Transfer
Protocol)
File sharing
Video conferences
VoIP (Voice over IP)
97
TCP/IP Encapsulation
Data segment
TCP segment
IP packet
Ethernet frame
98
TCP/IP operation
99
Hourglass IP
robustness against
technological innovations
anyone can innovate at
either end
new applications built by
any students (e.g.
netscape, napster)
new physical/link layers
allows huge
heterogenity
success
100
Broken layering
TCP/IP layers are broken more often than not
ICMP (Internet Control Message Protocol) - uses
IP, but controls its operation
BGP (Border Gateway Protocol) is a routing
protocol (IP layer), but is routed
IP over ATM over IP over ATM over SONET
anything involving MPLS (Multiprotocol Label
Switching)
often services are provided at multiple layers:
error and flow control, e.g. error control in SONET (sortof physical), link layer, IP, TCP
Note: In this light 7-layer OSI model seems

to be too complicated
101
End-to-end principle
Put functionality as high up the stack as
possible.
pushes functionality out towards the end
points results in
dumb network, smart terminals
contrast to PSTN (Telephone Network)
smart network, dumb terminals
also allows survival of partial network

failures
e.g. link failure, we can reroute
102
Distributed control
anything centralized is vulnerable
distribute physical infrastucture
distribute network control
e.g. routing protocols
OSPF (Open Shortest Path First), IS-IS, BGP
(Border Gateway Protocol)
not everything can be completely

decentralized
e.g. NOC, NCC
still can provide redundancy
Intermediate system to intermediate system (IS-IS), is a protocol used by routers to determine the
best way to forward datagrams through a packet-switched network. IS-IS protocol was defined
within OSI reference design.
103
A network operations center (NOC) or Network Control Center (NCC) is one or more locations from
which control is exercised over a computer or telecommunications network.
Deployment issues
scalability: has to work for a large range of networks
(in distance, and number of hosts).
IP creates networks of networks, that can span any scale:
1m > 10 000 km;
1 > 109 hosts;
link speeds 1 kbps 100 Gbps.
incremental deployment: need to be able to deploy

gradually.
constant change in the network
legacy networks wont go away
heterogeneity: different technologies and

applications and link speeds.
see layers 1 and 7 above.
link speeds covering 8 orders of magnitude.
104
Network scales
Geographic scale
PAN Personal Area Network (one room)
LAN Local Area Network (one building)
Ethernet (vast majority), Token ring, Wi-Fi,
CAN Campus Area Network (one campus)

MAN Metropolitan Area Network (one city)
WAN Wide Area Network (bigger than one city)
Number of routers/switches
small < 10
medium 10-100
large> 100
105
IP is not all good

Some things IP does not do well
billing
circuits are easy to bill
packets are not
most Internet charging is flat rate
QoS (Quality of Service)

it is difficult to design network to provide
guarantees of QoS
security (crypto doesnt fix DoS or DDoS

attacks)
A denial-of-service attack (DoS attack) or distributed denial-of-service attack106
(DDoS
attack) is an attempt to make a computer resource unavailable to its intended users.
Note: ToS and QoS

ToS - Type of Service provide an indication of
the quality of service desired.
The performance of a communications channel
or system is usually expressed in terms
of QoS (Quality of Service).
Depending upon the communication system,
QoS may relate to service performance as
SNR (Signal to Noise Ratio), BER (Bit Error
Ratio), maximum and mean throughput rate,
delay, delay variation, reliability, priority and
other factors specific to each service.
107

Network Optimization:
Goals and Constraints
108
Model, Analysis and Design

Empirical data from field trials and
deployments
Computer simulations
Analytic tools
Information theory, coding theory, communication
theory
Queuing theory and other probabilistic tools
Systems theory, graph theory
Optimization theory
109
Optimization
Optimization variables: x
Constant parameters describe objective
function f and constraint set C
Minimize { f (x )}
x
Subject to x C
110
Questions
How to describe the constraint set?
Can the problem be solved globally and
uniquely?
What kind of properties does it have?
Can we numerically solve it in an efficient and
distributed way?
Can we optimize multiple objectives
simultaneously?
Can we optimize over a sequence of time
instances?
111
Applications topics
Theory and algorithms of optimization are useful
Information theory problems,
Transmitter and receiver design,
Channel decoding,
Detection and estimation,
Multiple antenna beamforming,
Network resource allocation and utility maximization,
Wireless power control and medium access,
Network flow problems,
IP routing,
TCP congestion control,
Network architecture and topology design
112
Methodology topics
Linear programming,
Convex optimization,
Quadratic programming,
Geometric programming,
Integer programming,
Robust optimization,
Pareto optimization,
Dynamic programming,
Nonconvex optimization,
Lagrange duality,
Gradient methods,
Interior point methods,
Distributed algorithms,
113
Network Optimization Goals

optimization goals
reduce costs
improve performance
eg. minimize delays or latency
improve reliability or survivability

hard to write as an optimization problem
heuristic approach
redundancy
optimization constraints
technological, geographic, political
Heuristic is an adjective for experience-based techniques that help in problem solving. A heuristic
algorithm is an algorithm that is able to produce an acceptable solution to a problem in many practical
114
scenarios, but for which there is no formal proof of its correctness. Alternatively, it may be
correct, but may not be proven to produce an optimal solution.
Cost in networking
capital
equipment (cables, switches, ...)
premises
land that cables run along (right of ways)
operations
exclude sales and marketing, management, R&D
doesnt depend on network design
salaries of network administrators

repairs and upgrades
design
power
transit (from upstream providers)
fixed
traffic based costs
115
A right-of-way is a strip of land that is granted for transportation purposes, such as for a rail line
Equipment costs
Often assumed to dominate
fixed node costs
cost of a router - often assumed small
need to include premises, installation, etc.
fixed link costs

constant component
BW component
higher bandwidth links cost more
distance costs
straight distance cost
BW x distance cost
116
Link costs
Simple model: cost of a link
where
r = link capacity
d = link distance
the parameters k, , , are constants.
often some terms might be close to zero so ignore
some terms are out of our control, so we ignore
these, or push them into constants
117
Example
Lets consider the problem of business
that wants to connect up two locations
with a 10 Mbps link. What can they do:
118
Example
119
Example
120
Example
We have two possible solutions:
private line
lease or build whole line
cost depends on distance: C = kprivate + privated
VPN
pay for access to network at each end, but not for
the network
no distance dependence: VPN 0
decision: use private line if
kprivate + privated 2kVPN
121
The constants
Assume the simple model, how would you work out k,
, , .
and arise from the costs of building a links.
are the fixed costs: right-of-way, digging cables in, i.e.,
things we need regardless of how much capacity we use.
reflects capacity related costs: e.g., in the old days, if you
wanted two links, you needed two cables. Today, this might
reflect the number of wavelengths you use on a WDM
system.
In reality, we often purchase such links from a

physical layer network provider.
Their costs follow a pricing model that determines and .
122
The constants (2)

Assume the simple model, how would you work
out k, , , .
and k represent the non-distance dependent
costs of a link.
These are usually associated with end equipment,
for instance the WDM multiplexers, and line cards
at the routers that terminate the link:
k is non-capacity dependent costs: cost of getting
someone to install a line card, and spend time
configuring the router.
is capacity related term: higher speed line cards
usually cost more.
123
Link costs alternatives

distance component of physical link
wired: cost of fiber, amplifiers/repeaters,
digging, right of way
wireless: (e.g., free-space optics) free over
short distances
logical link (VPN-like networks)

(simplified) cost depend on capacity, but
not distance
may depend on actual traffic volume
satellites
A virtual private network (VPN) is implemented in an additional software layer (overlay) on top of an
existing larger network providing a secure extension of a private network into an insecure network
such as the Internet. The links between nodes of a virtual private network are formed over logical
124
connections or virtual circuits between hosts of the larger network. The Link Layer protocols of the
virtual network are said to be tunneled through the underlying transport network.
Optimizing for Latency

Another goal for optimization is to
maximize network performance.
Network performance often measured
by latency
Latency is the delay of a packet
crossing the network
Most often we are concerned with
average latency
over all paths through the network
125

Types of delay
propagation:
propagation delay directly related to distance
queueing:
queueing is caused by transient congestion
processing:
packet processing time (address lockup, and header update)
fixed per hop
transmission:
time to transmit packet on the line
= packet size / line rate
126
Example: Different scenarios

ARPANET low speed links (56 kbps), and slow
processors (IMPs)
propagation: coast-to-coast in US - 3Oms
transmission: 1500 x 8/56000 = 0.22 seconds.
queueing: a couple of packets a few seconds
processing: similar order to transmission, but smaller.
Now transmission and queueing times dominate.

Modern national backbone (10 Gbps)
propagation: coast-to-coast in US 3Oms
transmission: 1500 x 8/1e10 = 1.2 ns.
queueing: large buffers (up to 0.2 seconds)
processing: 30 ns.
Now queueing is dominant, unless low load, where

propagation becomes dominant.
IMP-Interface Message Processor, a packet-switching node for connecting
computers to ARPANET (modern term: router)
127

How to reduce
Propagation delay:
cannot speed up light
minimize length of paths
Queueing delay:
reduce queueing by reducing load
Processing delay:
minimize number of hops
Transmission delay:
minimize packet sizes
e.g. VoIP uses small packets
128
Optimizing for survivability

The 6 things network engineers care
about
reliability
reliability
reliability
reliability
cost
dont forget reliability

129
Five 9s
Goal of many telecom level providers is
five nines reliability
e.g. in IP networks
uptime is 99.999%
translates to about 5 minutes downtime per
year
pretty hard to achieve

not just network design
disaster recovery processes
130
Reliability approach
Often not approached using
optimization but
redundancy or standby systems
routers, links, power supplies ..
distribution of control
problem detection and diagnosis
131
Technological Constraints
The other aspect of optimization is the
constraints
max node degree
max number of line cards per router
times max ports per card
max capacity per link

limited by speed of line cards
follows Moores law
e.g. 0C762 40 Gbps
max capacity per router

backplane technology limited (also Moores law)
today, around 10 Tbps
max length of a link (e.g. Ethernet)

Moore's Law describes a long-term trend in computing hardware, in which the number
132
of transistors that can be placed inexpensively on an integrated circuit has doubled
approximately every two years
Non-technological Constraints
geography
cost of cable in oceans is different from land
expensive to lay cable in some places
e.g. downtown Manhattan
politics
internal company organization mandates network
organization
e.g. marketing get a better network than accounting, even
though they have less real need
security
may not want to share network resources outside
of secure building
133
Other Constraints
what if we have more than one objective
e.g. network should be
fastest
cheapest, and
most reliable
multi-objective optimization is hard

use other objectives as constraints, e.g.
best performance within a budget
cheapest network which meets performance
constraints
cheapest network which meets reliability
constraints
134
Other issues
usually there are other inputs to optimization
traffic measurements
not always as easy to get as we think
planning horizon
usually when we design a network it takes some
time to build
often we cant design our network from

scratch
have to deal with legacy equipment
incremental design
135
Simple example
Three node network has three acceptable designs:
4 possible network designs

associated costs have been worked out for each
Easy to choose the second network as the cheapest

136
Bigger problems
Network with N nodes
for small N we can evaluate all designs, and
choose the best
But 2N(N-1)/2 possible network designs

some arent practical
but we still have to check that
Even for N =20 we cant evaluate all of

these
at least not in the life-time of the Universe
137
Optimization
Optimization is about building automated
methods for finding optima of such problems
needs to work quickly (enough)
planning horizon
management requirements
size of the problem
ideally attains provably best solution

cant always do this (in reasonable time)
our problems are often NP-hard
need heuristic (rule of thumb) methods
NP-hard (non-deterministic polynomial-time hard), in computational complexity theory, is a class of

problems that are, informally, "at least as hard as the hardest problems in NP"
Note. A common mistake is to think that the NP in NP-hard stands for non-polynomial. Although it
is widely suspected that there are no polynomial-time algorithms for NP-hard problems, this
138has
never been proven. Moreover, the class NP also contains all problems which can be solved in
polynomial time.
High-level view of problems
139
Simple Set Notation
140
Optimization Notation
141
Other Notation
We usually use
lower case for scalars, e.g., x
lower-case boldface for (column) vectors, e.g., x
upper-case for matrices, e.g., A
When we write x < b we mean every element

of x is less than its corresponding element in
b, so
and similarly for relational operators , ,

142

Routing problem
Logical Router
144
Router Architecture (1)
145
PCI-Peripheral Component Interconnect

High performance architecture (input
and output queueing)
146

High performance architecture (input
and output queueing)
147
Packet processing
In an IP Router
lookup packet destination in forwarding table
up to 150,000 entries
update header (e.g. checksum, TTL)

send packet to outgoing port
buffer packet along the way
For a 10 Gbps line
small 40 byte packets
about 30 million packets per second
you have ~30ns per packet
The TTL (Time To Live) field is set by the sender of the datagram, and reduced by every host on
148
the route to its destination. If the TTL field reaches zero before the datagram arrives at its
destination, then the datagram is discarded and an error datagram is sent back to the sender.
Memory demands
forwarding table can be large
up to 150,000 entries per line card
lookup in 30ns for 10 Gbps line
need fast memory
buffers can be large

0.2 seconds per line card (rule of thumb)
10 Gbps line = 250 M memory (on in and out)
need fast memory (in + out in 30ns)
backplane must be faster than line cards

N times line rate speedup (N linecards)
to guarantee non-blocking switch fabric
149
Mapping the logical to the physical

Network maps (at one layer) can be
quite misleading
150
Circuit switching wont go away

Even for purist IP net-heads
often circuit switching in lower layers
GMPLS - -switching
WDM allows multiple wavelengths of light to share a
single fiber
optical cross-connects switch the light
no electronics involved
purely optical
protocols to set up and tear down optical circuits
packet forwarding on top of these circuits

151
Routing
We need a method to map packet routes to
links
called a routing protocol
several types exist
link state
shortest path
A common approach to routing uses shortestpaths.

The canonical algorithm for solving shortestpath routing is Dijkstras.
152
Notation and Assumptions
153
154
Network Paths
155
Network Paths
156
Network Paths
157
Network Paths
158
Network Paths
159
Network Paths
160
Network Paths
161
Network Paths
162
Network Paths
163
Network Paths
164
165
Routing
166
Routing
167
Routing costs
168
Routing costs
169
Routing problem
170
Routing problem
The Routing Problem: Determine the
optimal routing x to minimize C(f)
171
Linear costs
172
Path lengths
173
Network path-length example
174
175
176
Linear costs => shortest path routing
177
Special case
178
Triangle inequality
179
Dijkstras algorithm
180
Dijkstras algorithm
181
Dijkstras algorithm
182
Dijkstra Example
183
Dijkstra Example
184
Dijkstra Example
185
Dijkstra Example
186
Dijkstra Example
187
Dijkstra Example
188
Dijkstra Example
189
Dijkstra Example
190
Dijkstra Example
191
Dijkstra Result
192
Dijkstra intuition
193
Dijkstra issues
194
Sketch of proof of Dijkstra
195

Microwave Link Design
Jorma Kekalainen
Microwave Communication
A microwave communication system utilizes the
radio frequencies >1 GHz
These radio frequencies are called microwaves
Typically, spanning in microwave communication is
from 2 to 60 GHz.
As per IEEE, electromagnetic waves between 30
and 300 GHz are called millimeter waves or
Extremely High Frequencies (EHF) as their
wavelengths are from 10 to 1 mm.
Frequencies between 300 and 3000 GHz are called
Hyper High Frequencies (HHF) also called
submillimeter waves
197
Microwave Communication
Small capacity systems generally employ
the frequencies less than 3 GHz while
medium and large capacity systems
utilize frequencies ranging from 3 to 15
GHz.
Frequencies > 15 GHz are essentially
used for short-haul transmission.
198
Why radio links

No requirement for right-of-way
Less vulnerable to vandalism
Not susceptible to accidental cutting
of the link
Often more suited to crossing rough
terrain
Often more practical in heavily
urbanized areas
As a backup to fiber-optic cable links
Advantages of Microwave Radio

Less prone to accidental damage
Links across mountains and rivers are
more economically feasible
Two point installation and maintenance
Two point security
They are quickly deployed
200
Competitors
Fiber-optic cable systems provide
strong competition with line-of-sight
(LOS) microwave.
Satellite communications is an extension
of line-of-sight microwave.
Drawback; The excessive delay when the
popular geostationary satellite systems are
utilized.
Line-of-sight Microwave System

Line-of-sight (LOS) microwave provides broadband
bearer connectivity over a link or series of links
upwards into the millimeter spectrum.
Each link can be up to 40 - 50 km long or more
depending on terrain topology and antenna mast
height.
The key term here is line-of-sight (LOS).
It implies that the antenna of the radio link on one
end has to be able to see the antenna on the other
end.
This may not necessarily be true, but it does give
some idea of the problem.
Radio Horizon
The distance to the radio horizon varies with
the index of refraction of the intervening
space.
Under normal atmospheric conditions (k=4/3),
the radio horizon is about 15 percent beyond
the optical horizon because the microwave ray
beam being bent toward the earth.
However, this generalization may be overly
optimistic under certain circumstances.
Radio and Optical Horizon (smooth earth)
Refraction
205
Line-of-Sight Considerations
Microwave radio communication requires a clear line-of-sight
(LOS) condition
Radio LOS takes into account the concept of Fresnel ellipsoids
and their clearance criteria
Fresnel zones are specified employing ordinal numbers that
correspond to the number of half wavelength multiples that
represent the difference in radio wave propagation path from
the direct path
Ideally, the first Fresnel Zone must be clear of all
obstructions.
a+b+/2
r
a
Radius of the 1.Fresnel zone

rF=[*a*b/(a+b)]
206
Example
The link length is 40
km. Calculate the
maximum value of
the first Fresnel
radius, when
a) f=250 MHz
b) f=4 GHz
Note.
a+b+/2
rF
a
207
Radius of the first Fresnel zone

R=17.32(x(d-x)/fd)1/2
where d = distance between antennas (in km)
R= first Fresnel zone radius in meters
f= frequency in GHz
R
x
y
d=x+y
208
Typically the first Fresnel zone is used to
determine obstruction loss
The direct path between the transmitter and the
receiver needs a clearance above ground of at
least 60% of the radius of the first Fresnel zone
to achieve nearly free space propagation
conditions
Clearance is described to ensure sufficient
antenna heights so that, in the worst case of
refraction (for which Earth-radius factor k is
minimum) the receiver antenna is not placed in the
diffraction region
209
Clearance criteria to be satisfied under
normal propagation conditions:
Clearance of 60% or greater at the
minimum k suggested for the certain path
Clearance of 100% or greater at k=4/3
In case of space diversity, the antenna

can have a 60% clearance at k=4/3
210
Microwave Link Design

Microwave Link Design is a
systematic process that includes
Loss/attenuation calculations
Fading and fade margins calculations
Frequency planning and interference
calculations
Quality and availability calculations
211
Microwave Link Design Process

The whole process is iterative and may go through
many redesign phases before the required quality and
availability are achieved
Interference
analysis
Frequency
Planning
Propagation losses
Branching
losses
Other Losses
Link Budget
Quality
and
Availability
Calculations
Fading
Predictions
Rain
attenuation
Diffractionrefraction
losses
Multipath
propagation
212
Hop Calculations
213
Q&A stands here for Quality and Availability
Loss / Attenuation Calculations

The loss/attenuation calculations are
composed of three main contributions
Propagation losses
(Due to Earths atmosphere and terrain)
Branching losses
(comes from the hardware used to deliver the
transmitter/receiver output to/from the
antenna)
214
Loss / Attenuation Calculations

Miscellaneous (other) losses
unpredictable and sporadic in character like fog,
moving objects crossing the path, poor equipment
installation and less than perfect antenna
alignment etc
Detailed contribution of these losses is not

calculated but they are considered in the
planning process as an additional loss
215
Propagation Losses
Free-space loss - when the transmitter and
receiver have a clear, unobstructed line-of-sight
Lfsl=92.45+20log(f)+20log(d) [dB]
where f = frequency (GHz)
d = LOS range between antennas (km)
One estimation of vegetation attenuation
(provision should be taken for 5 years of
vegetation growth)
L=0.2f 0.3R0.6(dB)
f=frequency (MHz)
R=depth of vegetation in meters
216
Diffraction
217
Propagation Losses
Obstacle Loss also called Diffraction Loss or Diffraction
Attenuation.
One method of calculation is based on knife edge
approximation.
Having an obstacle free 60% of the Fresnel zone gives 0
dB loss
First Fresnel Zone
0 dB
0 dB
6dB
16dB
20dB
218
Propagation Losses
Gas absorption
Primarily due to the water vapor and
oxygen in the atmosphere.
The absorption peaks are located around 23
GHz for water molecules and 50 to 70 GHz
for oxygen molecules.
The specific attenuation (dB/km) is
strongly dependent on frequency,
temperature and the absolute or relative
humidity of the atmosphere.
219
Gas Attenuation vs. Frequency

Total specific
gas attenuation 23GHz
1.0 (dB/Km)
T=40oC
RH=80%
0.4
T=30o
RH=50%
Frequency (GHz)
25
50
220
Gas Absorption
221
Propagation Losses
Attenuation due to precipitation
Rain attenuation increases exponentially with rain intensity
The percentage of time for which a given rain intensity is
attained or exceeded is available for 15 different rain
zones covering the entire earths surface
The specific attenuation of rain is dependent on many
parameters such as the form and size of distribution of the
raindrops, polarization, rain intensity and frequency
Horizontal polarization gives more rain attenuation than
vertical polarization
Rain attenuation increases with frequency and becomes a
major contributor in the frequency bands above 10 GHz
The contribution due to rain attenuation is not included in
the link budget and is used only in the calculation of rain
fading
222
Atmospheric attenuation
Rain Attenuation
Two types of attenuating mechanisms:
absorption and scattering caused by the rain
drops.
Horizontally polarized waves are

attenuated more than vertically polarized
waves.
224
Rain attenuation
225
Ground Reflection
Reflection on the Earths surface may give rise to multipath
propagation
The direct ray at the receiver may interfered with by the
ground-reflected ray and the reflection loss can be
significant
Since the refraction properties of the atmosphere are
constantly changing the reflection loss varies.
The loss due to reflection on the ground is dependent on the
reflection coefficient of the ground
The reflection coefficient is dependent on the frequency,
grazing angle (angle between the ray beam and the horizontal
plane), polarization and ground properties
226
Ground Reflection
The grazing angle of radio-relay paths is very small
usually less than 1
It is recommended to avoid ground reflection by
shielding the path against the indirect ray
The contribution resulting from reflection loss is
not automatically included in the link budget.
When reflection cannot be avoided, the fade
margin may be adjusted by including this
contribution as additional loss in the link budget
227
Multipath
228

Path Analysis (Link Budget)
Introduction
The path analysis (or link budget) is carried out to
dimension the link.
This means here to establish operating parameters
such as transmitter power output, parabolic antenna
aperture (diameter), and receiver noise figure, among
others.
The type of modulation and modulation rate (number
of transitions per second) are also important
parameters.
Next table shows basic parameters in two columns.
The first we call normal and would be the most
economic; the second column is titled special,
giving improved performance parameters, but at an
increased price.
LOS Microwave Basic Equipment

Parameters
Space distribution of power
232
The First Fresnel Zone

Design objective:
Full clearance of the 1. Fresnel zone.
233
The First Fresnel Zone

The radiated power is distributed in a
zone surrounding the direct line-ofsight.
234
Example
235
Obstacle Loss
236
Knife edge and earth surface losses
237
True and equivalent earth radius
238
Earth Bulge calculation
239
Free 1.Fresnel zone
The 1st Fresnel shall be free from obstacles

when k=4/3.
On paths over water surfaces or desert
areas, it is recommended to have the 1st
Fresnel zone free from obstacles when k=1.
240
Link Budget
The link budget is a calculation involving the
gain and loss factors associated with the
antennas, transmitters, transmission lines and
propagation environment, to determine the
maximum distance at which a transmitter and
receiver can successfully operate
Receiver sensitivity threshold is the signal
level at which the radio runs for a given biterror rate at a specified bit rate
System gain depends on the design of the
radio
Link Budget
The gains from the antenna at each end are added to
the system gain
larger antennas provide a higher gain
The free space loss of the radio signal is subtracted.

the longer the link the higher the loss
These calculations give the fade margin

In most cases since the same duplex radio setup is
applied to both stations the calculation of the received
signal level is independent of direction
242
Approach
First, we calculate the free-space loss between the transmit
antenna and receive antenna.
This is a function of distance and frequency (i.e., the
microwave transmitter operational frequency).
We then calculate the EIRP (effective isotropically radiated
power) at the transmit antenna.
The EIRP is the sum of the transmitter power output, minus
transmission line losses plus the antenna gain, all in decibel units.
When we add the EIRP to the free-space loss (in dB), the result
is the isotropic receive level (IRL).
When we add the receive antenna gain to the IRL and subtract
the receive transmission line losses, we get the receive signal
level (RSL).
This relationship of path gains and losses is shown in the
following figure.
LOS microwave link gains and losses.

Transmitter output is 1 watt or 0 dBW.
Path Loss
If the isotropic radiator (an antenna that radiates uniformly in
all directions) is fed by a transmitted power Pt, it radiates
power density Pt/4d2 (W/m2) at a distance d, and if a radiator
has a gain Gt, the power flow is enhanced by the factor Gt.
So power density at a distance d is S0= GtPt/4d2 .
Finally, the power intercepted by an antenna of effective cross
section A (related to the gain by Gr = 4A/2) is Pr=AS0=PtGtGr(
/4d)2.
The term 1/(/4d)2 is known as the free-space loss and
represents the steady decrease of power flow (in W/m2) as the
wave propagates.
From this we can derive the commonly used formula of freespace path loss, which reduces to
where L is the free-space attenuation between isotropic

antennas in dB, F is the frequency in GHz, and D is the path
distance in km.
Example
Consider the problem from a different aspect.
It requires 22 dB to launch a wave to just 1 wavelength ()
distant from an antenna.
Thus for an antenna emitting +10 dBW, we could expect the
signal one wavelength away to be 22 dB down, or -12 dBW.
Whenever we double the distance, we incur an additional 6
dB of loss.
Hence at 2 from the + 10-dBW radiator, we would find -18
dBW; at 4, -24 dBW; 8, -30 dBW; and so on.
Now suppose that we have an emitter where F = 1 GHz. What is the
path loss at 1 km?
From rough calculations, the 6-dB relationship is worthwhile and

also gives insight in that if we have a 50-km path and shorten or
lengthen it by a km, our signal level will be affected little.
Calculation of EIRP
Effective isotropically
radiated power is
calculated by adding
decibel units:
the transmitter power
output (in dBm or dBW),
the transmission line losses
in dB (a negative value
because it is a loss), and
the antenna gain in dBi.
dBi = decibels
referenced to an
isotropic (antenna).
EIRP = Trans. output (dBW) - Trans. line loss (dB) + Ant. gain (dB)
Example
If a microwave transmitter has a 1-watt
(0-dBW) power output, the waveguide
loss is 3 dB, and the antenna gain is 34
dBi, what is the EIRP in dBW?
Calculation of Isotropic Receive Level

(IRL)
The IRL is the RF power level impinging on the
receive antenna.
It would be the power we would measure at
the base of an isotropic receive antenna.
The calculation is shown below
lRLdBW = EIRPdBW Path lossdB

Below 10 GHz : Path loss = Free-space loss (FSL)
Calculation of Receive Signal Level (RSL)

The receive signal level (RSL) is the
power level entering the first active
stage of the receiver:
Note: Power levels can be in dBm as well.
Example
Suppose the isotropic receive level
(IRL) is -121 dBW, the receive antenna
gain is 31 dB, and the line losses are 5.6
dB. What is the RSL?
Link Budget
Receive Signal Level (RSL)
RSL = Pout Lctx + Gatx Lcrx + Gatx FSL
Pout = output power of the transmitter (dBm)
Lctx, Lcrx = Loss (cable,connectors, branching unit) between
transmitter/receiver and antenna(dB)
Gatx = gain of transmitter/receiver antenna (dBi)
FSL = free space loss (dB)
Link feasibility formula

RSL Rx (receiver sensitivity threshold)
252
Calculation of Receiver Noise Level

The thermal noise level of a receiver is a function of
the receiver noise figure and its bandwidth.
For analog radio systems, receiver thermal noise level
is calculated using the bandwidth of the intermediate
frequency (IF).
For digital systems, the noise level of interest is in
only 1 Hz of bandwidth using the notation N0, the
noise level in a 1-Hz bandwidth.
The noise that a device self-generates is given by its
noise figure (dB) or a noise temperature value.
Calculation of Receiver Noise Level

Any device above absolute zero generates thermal noise.
We know the thermal noise power level in a 1-Hz bandwidth of a
perfect receiver operating at 1 K is
where Pn is the noise power level.

From numerical value we will recognize this as Boltzmanns
constant expressed in dBW.
We can calculate the thermal noise level of a perfect receiver
operating at room temperature (T=290 K) using the following
formula:
Pn=kTB, k=1.3806504 1023 J/K
Noise Figure
Noise figure simply tells us how much noise has been
added to a signal while passing through a device in
question.
Noise figure (dB) is the difference in signal-to-noise
ratio between the input to the device and the output
of that same device.
We can convert noise figure to noise temperature in
kelvins with the following formula:
where Te is the effective noise temperature of a

device.
Example
Suppose the noise figure of a device is
3 dB. What is the noise temperature?
Lets round 1.995 to 2; thus
Thermal Noise Power

The thermal noise power level of a
device operating at room temperature is
where BW is the bandwidth of the

device in Hz.
Example
A microwave receiver has a noise figure of 8 dB and
its bandwidth is 10 MHz. What is the thermal noise
level (sometimes called the thermal noise threshold)?
If the receiver in the above example was operating in

a digital regime, wed want to calculate N0 (noise
spectral density)
Calculation of Eb/N0 in Digital Radio

Systems
Instead of familiar signal-to-noise ratio (S/N, SNR) in digital
systems we often use Eb/N0, meaning energy per bit per noise
spectral density ratio.
We can relate Eb/N0 to bit error rate (BER) given the
modulation type in question.
Suppose the RSL was 1 watt and we were receiving 1000 bits per
second.
How much energy is imparted to 1 bit? We simply divided 1 watt
by 1000 bits per second. So it is 1 mW.
In radio work we do the division logarithmically because we work
with decibels.
Eb can be stated as follows:
The receive signal level (RSL)
Example
A certain radio system receives 1.544
Mbps and the RSL is -108 dBW. What is
the energy per bit (Eb)?
Eb/N0
We can now develop a formula for
Eb/N0:
Simplifying, we obtain
Note.
Some Notes on Eb/N0

Eb/N0, for a given BER, will be different for
different types of modulation (e.g., FSK, PSK,
QAM, etc.).
When working with Eb, we divide RSL by the
bit rate, not the symbol rate nor the baud
rate.
There is a theoretical Eb/N0 and a practical
Eb/N0.
The practical Eb/N0 is always a greater value
than the theoretical, greater by the
modulation implementation loss in dB.
Digital Modulation of LOS Microwave

Radios
Digital systems, typically standard PCM, are
notoriously wasteful of bandwidth compared to their
analog counterparts.
For example, the analog voice channel is nominally 4
kHz, whereas the digital voice channel, assuming one
bit per hertz of bandwidth is 64 kHz.
This is a 16-to-1 difference in required bandwidth.
Thus various national regulatory authorities require
that digital systems must be bandwidth conservative.
One term that is used in this connection is bit
packing.
This means packing more bits into a hertz of
bandwidth.
How does bit packing work?

In the binary domain we can estimate bandwidth to
approximately equate to 1 bit/Hz.
For example, if we were transmitting at 2.048 Mbps wed
need 2.048 MHz of bandwidth.
Quadrature phase shift keying (QPSK) is one example

of higher levels of modulation.
In this case we achieve a theoretical packing of 2 bits/Hz.
Again, if we are transmitting 2.048 Mbps with QPSK we
would need 2.048 MHz/2 or 1.024 MHz.
QPSK is one of a family of modulation schemes that

are based on phase shift keying (PSK).
Block diagram of a QPSK modulator

With binary PSK (BPSK) we might assign a binary 1 to the 0
position (i.e.. no phase retardation) and a binary 0 to the 180
phase retardation point.
For QPSK, the phase-circle is broken up into 90 segments.
In this case, for every transition we transmit 2 bits at a time.
QPSK modulator really only consists of two BPSK modulators

where one is out of phase with the other by 90.
8-PSK
In 8-ary PSK modulation case the phase
circle is broken up into 45 phase
segments.
Now for every transition, 3 bits at a
time are transmitted.
The bit packing in this case is 3 bits per
Hz.
QAM
We can construct a hybrid waveform covering both
amplitude modulation as well as phase modulation.
This family of waveforms is called quadrature
amplitude modulation (QAM).
For example, 16-QAM has 16 different state possibilities:
eight are derived for 8-ary PSK and two are derived from
the two amplitude levels.
Here for each state transition, 4 bits are transmitted at
once.
The bit packing in this case is 4 bits/Hz.
Some digital LOS microwave systems use 256-QAM

and 512-QAM, theoretically achieving 8 bits/Hz and
9 bits/Hz of bit packing.
Example
Theoretical/Practical bandwidth
The difference between theoretical bit
packing and the practical deals with filter
design.
For QAM-type waveforms, depending on
design, practical bit packing may vary from
1.25 to 1 .5 the baudrate-bandwidth.
The extra bandwidth required provides a
filter with spectral space to roll-off.
In other words, a filters skirts are not
perfectly vertical.
Example
Suppose we are using a 48-Mbps bit stream to input
to our transmitter which was using 16-QAM
modulation.
Its baud rate, which measures transitions per second,
would be 48/4 megabauds.
If we allowed 1 baud/Hz, then 12 MHz bandwidth
would be required.
If we used a roll-off factor of 1.5, then the practical
bandwidth required would be 18 MHz.
Carry this two steps further to 64-QAM.
Here the theoretical bit packing is 6 bits/Hz and for
the 48-Mbps bit stream, a practical 12-MHz
bandwidth would be required.
baud (unit symbol "Bd") is synonymous to symbols per second. It is the unit
of symbol rate, also known as baud rate or modulation rate; the number of
distinct symbol changes (signaling events) made to the transmission medium
per second in a digitally modulated signal or a line code
BER vs. Eb/N0 for various PSK and QAM
As an amount of signal elements (symbols) M=2n increases (e.g., M = 64), for

a given error rate, Eb/N0 increases
Example
A digital link operates in the 7-GHz
band with a link 37 km long.
The bit rate is 1.544 Mbps and the
modulation is 64-QAM.
The specified BER for the link is 10-7
and the modulation implementation
loss is 2 dB.
The receiver noise figure is 8 dB.
The antennas have 35-dB gain at each
end, and transmission line losses are
1.8 dB at each end.
What link margin can be expected
assuming the transmitter has a 1-watt
output?
Solution
First turn to next figure and derive
the required Eb/N0.
This is about 19.5 dB; add to this the
modulation implementation loss of 2
dB and the result is that the required
value for Eb/N0 is 21.5 dB.
Example (cont.)
The next step is to calculate a
candidate RSL value.
We know that Eb must be 21.5 dB above
N0.
We can calculate N0 because we have
the receiver noise figure (8 dB)
Example (cont.)
Thus RSL, in this case, is 10 log(1.544106)
greater than Eb, because
So
Example (cont.)
We will hold this minimum RSL value for future
reference (we use this reference to calculate the
prevailing link margin), and now we turn to the transmit
side of the link.
Assume the transmitter has a 1-watt output or 0 dBW.
Calculate EIRP (effective isotropically radiated power)
in dBW
Example (cont.)
Calculate the free-space loss (path loss):
Calculate the IRL (isotropic receive level) :

lRLdBW = EIRPdBW Path lossdB
Example (cont.)
Calculate RSL (Receive Signal Level) :
Calculate the link margin: RSL-RSLreference
Antenna Gain
Often we are faced with the problem of
What antenna gain will provide the
margin or provide the gain necessary to
meet performance objectives?
Parabolic Antenna Gain

At a given frequency the gain of a parabolic antenna is a
function of its effective area and may be expressed by the
formula
where G is the gain relative to an isotropic antenna, A is the

area of antenna aperture, is the aperture efficiency, and is
the wavelength at the operating frequency.
Commercially available parabolic antennas with a conventional
horn feed at their focus usually display a 55% efficiency or
somewhat better. With such an efficiency, gain (G, in decibels)
is then
where D is measured in meters and F in gigahertz.
Calculation of Antenna Gain (=0.55)
c
=
f
9 2
FGHz 10
2
G = 10 log 4 (D / 2 ) 0.55
8
3 10
2 0.55 100
G = 20 log D + 20 log FGHz + 10 log

9
G = 20 log D + 20 log FGHz + 17.8
Example
What size antenna would be required in
the preceding example? Let G = 35dB
and F=7GHz.
D1m
Antenna types
Parabolic dish antennas, with waveguide (horn)
feeds, are probably the most economic
antennas for radio links operating from 3 GHz
upward.
Below 3 GHz, coaxial feeds are used, and often
the antennas are Yagis.
Coaxial cable transmission lines deliver the RF
energy from/to transmitter/receiver to the
antenna in this range.
Above 3 GHz, coaxial cable becomes too lossy
and waveguide is more practical.
Other types of antennas may also be used, such
as the horn, and spiral.
Besides cost and gain, other features are
front-to-back ratio, side lobes, and efficiency.
Power budget
LFSL = 92.4 + 20 log FGHz + 20 log Dkm
Prx = Ptx + 2Gant 2 L feed LFSL
Margin = Prx Pthreshold
LFSL: free space loss

Dkm: distance
FGHz: frequency
Prx: received power [dBm]
Ptx : transmitted power [ dBm]
Pthreshold: receiver threshold [dBm]
M : margin
Gant: antenna gain
Lfeed: feeder cable loss
Link Budget
The fade margin is calculated with respect to
the receiver threshold level for a given biterror rate (BER).
The link is available if received signal is above
receiver threshold level, otherwise the link
goes down
Link feasibility formula
RSL Rx = receiver (sensitivity) threshold
285
Radio path link budget

Transmitter 1
waveguide
Transmitter 2
Splitter
Splitter
Receiver 1
Antenna Gain
Branching
Losses
Propagation
Losses
Output
Power (Tx)
Receiver 2
Branching
Losses
Received
Power (Rx)
Fade Margin
Receiver threshold Value

286
Fading
On very short radio paths below about 10 GHz, the signal level
on the distant end receiving antenna, assuming full LOS
conditions, can be predicted (calculated) accurately (about 1dB
inaccuracy).
If the transmitter continues to give the same output, the
receive signal level will remain constant.
As the path is extended, the measured RSL will vary around a
median.
The signal level may remain at that median for minutes or hours,
and then suddenly drop and then return to the median again.
In other periods and/or on other links, this level variation can be
continuous for periods of time.
Drops in level can be as much as 30 dB or more.
This phenomenon is called fading.
The system and link design must take fading into account when
sizing or dimensioning the system/link.
Fades
As the RSL drops in level, so does the Eb/N0.
As the Eb/N0 decreases, there is a
deterioration in error performance; the BER
degrades.
Fades vary in depth, duration, and frequency
(i.e., number of fade events per unit of time).
We cannot eliminate the fades, but we can
mitigate their effects.
The primary tool we have is to overbuild each
link by increasing the margin.
Availability
Link margin is the number of dB we have as a surplus in the link
design.
We could design an LOS microwave link so we just achieve the
RSL at the distant receiver to satisfy the Eb/N0 (and BER)
requirements using free-space loss as the only factor in link
attenuation (besides transmission line loss).
Unfortunately we will only meet our specified requirements
about 50% of the time.
So we must add margin to compensate for the fading.
We have to determine what percentage of the time the link
meets BER performance requirements.
We call this time availability.
If a link meets its performance requirements 99% of the time,
then it does not meet performance requirements 1 % of the
time.
We call this latter factor unavailability.
Signal Power (dBm)
Scales of Fading
Large scale fading component
Small scale fading

component
290
Link Margin = Fade Margin

To improve time availability, we must increase the link margin,
often called the fade margin.
How many additional dB are necessary?
There are several approaches to the calculation of a required
fade margin.
One of the simplest and most straightforward approaches is to
assume that the fading follows a Rayleigh distribution, often
considered worst-case fading.
If we base our premise on a Rayleigh distribution, then the
following fade margins can be used:
Probability calculations
Probability calculations
Example
Example
Calculate the following table a) with given time availability and b)
with given fade margin
Time availability
Probability that R-distributed field is less

than abscissa value
Per-hop time availability

More often than not, LOS microwave systems consist of multiple
hops.
Here our primary interest is the time availability at the far-end
receiver in the system after the signal has progressed across all
of the hops.
From this time availability value we will want to assign an
availability value for each hop or link.
Suppose a system has nine hops and the system time availability
specified is 99.95%, and we want to calculate the time
availability per hop or link.
The first step is to calculate the system time unavailability.
This is simply 1.0000 0.9995 = 0.0005.
We now divide this value by 9 (i.e., there are nine hops or links):
Now we convert this value to time availability:
Multipath
The most common cause of
fading is multipath
conditions.
As the term implies, signal
energy follows multiple paths
from the transmit antenna to
the receive antenna.
Two additional paths, besides
the main ray beam, are shown
next figure.
Most of the time the
delayed signal energy (from
the reflected/refracted
paths) will be out of phase
with the principal ray beam
which causes fading.
Multipath
300
Multipath Propagation
Reflection (odd Fresnel zone numbers) and
Scattering
Reflected signal strength is dependent on:

The total reflection coefficient (reflection and/or
scattering)
The total phase shift
301
Multipath Propagation
Reflection (even Fresnel zone numbers) and
Scattering
Reflected signal strength is dependent on:

The total reflection coefficient (reflection and/or
scattering)
The total phase shift
302
Dispersion
In digital systems, there is the
additional impairment of dispersion
caused by multipath.
Of course, the delayed energy arrives
later, spilling into the next bit or binary
symbol position, increasing the
probability that that bit decision will
be in error.
Overbuilding
Probably the most economic way to overbuild a link is to increase the
antenna aperture.
Every time we increase the aperture doubling the diameter of the
parabolic dish, we increase the gain by 6 dB
It is recommend that apertures for LOS microwave antennas not

exceed 4 m.
Not only does the cost of the antenna get notably greater as aperture
increases over 3 m, but the equivalent sail area of the dish starts to
have an impact on system design.
Wind pressure on large dishes increases tower twist and sway, resulting
in movement out of the capture area of the ray beam at the
receive antenna.
This forces to stiffen the tower, which could dramatically increase
system cost.
Also, as antenna aperture increases, gain increases and beamwidth
decreases.
Overbuilding
Other measures we can take to overbuild a link are:
Insert a low-noise amplifier (LNA) in front of the receiver mixer.
Improvement: 6 - 12 dB.
Use an HPA (high-power amplifier).

Usually a traveling-wave tube (TWT) amplifier; 10 watts output.
Improvement: 10 dB.
Implement FEC (forward error correction).

Improvement: 15 dB.
Implement some form of diversity.
Space diversity is preferable in many countries.
Can be a fairly expensive measure.
Improvement: 5 - 20 dB or more.
Fading conditions
It should be appreciated that fading varies with path
length, frequency, climate, and terrain.
The rougher the terrain, the more reflections are
broken up.
Flat terrain, and especially paths over water, tends to
increase the incidence of fading.
For example, in dry, windy, mountainous areas the
multipath fading phenomenon may be nonexistent.
In hot, humid coastal regions very strong fading may
be expected.
Fading types
Fading is defined as the variation of the
strength of a received radio carrier signal
due to atmospheric changes and/or ground
and water reflections in the propagation
path.
Four fading types are considered while
planning links.
They are all dependent on path length and
are estimated as the probability of
exceeding a given (calculated) fade margin
307
Fadin Margin
308
Fading margin
Safety margin.
Should be large
enough to guarantee
that quality and
availability
objectives are met
during fading
conditions.
Typical value 40 dB
309
Fading
310
Fading types
Multipath fading
Flat fading
Frequency-selective fading
Rain fading
Refraction-diffraction fading
311
Multipath fading
Multipath fading is the dominant fading
mechanism for frequencies lower than 10 GHz.
A reflected wave causes a multipath, i.e. when a
reflected wave reaches the receiver as the
direct wave that travels in a straight line from
the transmitter
If the two waves reach the receiver out of phase
they weaken the overall signal.
As a thumb rule, multipath fading, for radio links
having bandwidths less than 40MHz and path
lengths less than 30km is described as flat
instead of frequency selective
312
Flat and Frequency Selective fading

Flat fading
A fade where all frequencies in the channel
bandwidth are equally affected.
If necessary, flat fade margin of a link can be
improved by using larger antennas, a higher-power
microwave transmitter, lowerloss feed line and
splitting a longer path into two shorter hops
On water paths at frequencies above 3 GHz, it is
advantageous to choose vertical polarization
Frequency-selective fading
There are amplitude and delay distortions across
the channel bandwidth
It affects high capacity radio links
313
Multipath Fading
314
Rain fading
Rain attenuates the signal caused by the scattering
and absorption of electromagnetic waves by rain
drops
It is significant for long paths (>10km)
It starts increasing at about 10GHz and for
frequencies above 15 GHz, rain fading is the
dominant fading mechanism
Rain outage increases dramatically with frequency and
then with path length
Microwave path lengths must be reduced in areas
where rain outages are severe
The available rainfall data is usually in the form of a
statistical description of the amount of rain that falls
at a given measurement point over a period of time
315
Reducing the Effects of Rain

Multipath fading is at its minimum during
periods of heavy rainfall, so entire path fade
margin is available to combat the rain
attenuation
Route diversity with paths separated by more
than a few km can be used successfully
Vertical polarization is far less susceptible to
rainfall attenuation (40 to 60%) than
horizontal polarization.
316
Fading Margin
Fading margin > Rain fading + Multipath
fading
Rain fading
Dominant cause of fading for f> 10 GHz
Multipath fading
Dominant cause of fading for f< 10 GHz
Common fading margin: 25 - 40 dB

317
Refraction Diffraction Fading

Also known as k-type fading
For low k values, the fictive Earths surface becomes curved and
terrain irregularities, man-made structures and other objects
may intercept the Fresnel Zone.
For high k values, the fictive Earths surface gets close to a plane
surface and better LOS (lower antenna height) is obtained
The probability of refraction-diffraction fading is therefore
indirectly connected to obstruction attenuation for a given value
of Earth radius factor
Since the Earth-radius factor is not constant, the probability of
refraction-diffraction fading is calculated based on cumulative
distributions of the Earth-radius factor
318
Interference fade margin

To accurately predict the performance of a digital
radio path, the effect of interference must be
considered.
Interference in microwave systems is caused by the
presence of an undesired signal in a receiver.
When this undesired signal exceeds certain limiting
values, the quality of the desired received signal is
affected.
To maintain reliable service, the ratio of the desired
received signal to the (undesired) interfering signal
should always be larger than the threshold value.
319
Interference fade margin

In normal unfaded conditions the digital signal
can tolerate high levels of interference but in
deep fades it is critical to control
interference.
Adjacent-channel interference fade margin
(AIFM) (in decibels) accounts for receiver
threshold degradation due to interference
from adjacent channel transmitters
Interference fade margin (IFM) is the depth
of fade to the point at which RF interference
degrades the BER to 10-3 .
320
Far-end Interference
321
Near-end Interference
322
Adjacent channel interference
323
Output Power
324
Output Power
325
Antennas
326
Threshold Degradation
327
Threshold Degradation
328
Link Budget
329
Microwave Link Multipath Outage

Models
A major concern for microwave system users
is how often and for how long a system might
be out of service.
An outage in a digital microwave link occurs
with a loss of digital signal frame sync for
more than 10 sec.
Digital signal frame loss typically occurs when
the BER increases beyond 10-3.
330
Outage and Availability

Outage (Unavailability) (%) = (SES/t) x 100
where
t = time period (expressed in seconds)
SES = severely errored second (1-s intervals in which the BER
exeeds 10-3 )
Availability is expressed as a percentage as :
A = 100 - Outage (Unavailability)
A digital link is unavailable for service after a ten consecutive
BER>10-3 SES outage periods.
331
Quality and Availability (Q&A)

The main purpose of the quality and availability
calculations is to set up reasonable quality and
availability objectives for the microwave path.
The ITU-T recommendations G.801, G.821 and
G.826 define error performance and availability
objectives.
The objectives of digital links are divided into
separate grades: high, medium and local grade.
The following grades are usually used in wireless
networks
Medium grade for the access network
High grade for the backbone network
332
Improving the Microwave System

Diversity Improvement
Space Diversity
Polarity Diversity
Frequency Diversity
Route Diversity
Hybrid Diversity
Media Diversity
Hardware Redundancy
Hot standby
Repeaters
333
Diversity Operation
Diversity reception means the simultaneous reception of the
same radio signal over two or more paths.
Each path is handled by a separate receiver chain and then
combined in the radio equipment so that effects of fading are
mitigated.
The separate diversity paths can be based on space, frequency,
and/or time diversity.
The simplest form of diversity is space diversity.
The two diversity paths in space diversity are derived at the
receiver end from two separate receivers with a combined
output.
Each receiver is connected to its own antenna, separated
vertically on the same tower.
The separation distance should be at least 70 wavelengths and
preferably 100 wavelengths.
Probably, fading will not occur on both paths simultaneously.
Diversity Antenna Systems

Combining
Combined signal
fed to receiver
Signal 2
Signal 1
Signal Strength
C o m b in e d s ig n a l
S ig n a l 1
S ig n a l 2
T im e
335
Concept of Diversity
Fade
Transmission
media 1
Information
Receiver
Transmission
Tmedia 2
Peak
336
Diversity
Primarily used to reduce the effects of
multipath fading.
337
Diversity
338
A space-diversity configuration
Antenna separation
340
Correlation between branches

The branch correlation coefficient (r) represents the degree of
similarity between the signals from two different receiving
branches.
The correlation coefficient ranges from 0 to 1.
r=1 means the signals from two different branches behave
exactly the same. In this case, the signals are coherent.
r=0 means the signals from two different branches behave
completely different. In this case, the signals are uncorrelated.
To achieve the best performance, a diversity antenna system
is required to provide uncorrelated signals.
For r=1, the diversity antenna becomes ineffective in combating
the multipath fading.
A diversity antenna system can perform satisfactorily provided
that r<0.7.
341
Probability that R-distributed field is less than abscissa

value with different number of diversity channels
Number of diversity channels
Frequency Diversity
Frequency diversity is more complex and more costly than space
diversity.
It has advantages as well as disadvantages.
Frequency diversity requires two transmitters at the near end
of the link.
The transmitters are modulated simultaneously by the same
signal but transmit on different frequencies.
Frequency separation must be at least 2%, but 5% is
preferable.
The two diversity paths are derived in the frequency domain.
When a fade occurs on one frequency, it will probably not occur
on the other frequency.
A frequency-diversity configuration
Frequency Diversity
The more one frequency is separated from the other, the less
chance fades will occur simultaneously on each path.
Frequency diversity is more expensive, but there is greater
assurance of path reliability.
It provides full and simple equipment redundancy and has the
great operational advantage of two complete end-to-end
electrical paths.
In this case, failure of one transmitter or one receiver will not
interrupt service, and a transmitter and/or a receiver can be
taken out of service for maintenance.
The primary disadvantage of frequency diversity is that it
doubles the amount of frequency spectrum required in this day
when spectrum is at a premium.
In many cases it is prohibited by national licensing authorities.
For example, the U.S. Federal Communications Commission (FCC)
does not permit frequency diversity for industrial users.
Hot-standby
The full equipment redundancy aspect is very attractive to the
system designer.
Another approach to achieve diversity improvement in
propagation plus reliability improvement by fully redundant
equipment is to resort to the hot-standby technique.
On the receive end of the path, a space-diversity configuration
is used.
On the transmit end a second transmitter is installed as in in the
previous figure, but the second transmitter is on hot standby.
This means that the second transmitter is on but its signal is not
radiated by the antenna.
On a one-for-one basis the second transmitter is on the same
frequency as the first transmitter.
On failure of transmitter 1, transmitter 2 is switched on
automatically.
One-for-N hot standby

One-for-N hot standby is utilized on large radio link systems
employing several radio carriers, where the cost for duplicate
equipment for each channel may be prohibitive.
On such multi-RF channel arrangements it is customary to assign some
sort of priority arrangement.
Often the priority channel enjoys the advantage of frequency diversity,
whereas the other RF channels do not.
On failure of one of the other channels, the diversity improvement is
lost on the priority channel, with the diversity pair switched to carry
the traffic on the failed pair.
In another arrangement the standby channel carries low-priority
traffic and does not operate in a frequency diversity arrangement,
while providing protection for perhaps two or three other channels
carrying the higher-priority traffic.
On failure of a high-priority channel, the lower-priority channel drops
its traffic, replacing one of the RF channels carrying the more
important traffic flow.
Once this occurs, the remaining channels operate without standby
equipment protection.
Diversity Improvement
Propagation reliability improvement can be
exemplified as follows.
If a 50 km path required a 51 dB fade margin to
achieve a 99.999% reliability on 6.7 GHz without
diversity, with space diversity on the same path, only
a 33-dB fade margin would be required for the same
propagation reliability, namely, 99.999%.
For frequency diversity in the nondiversity condition,
assuming Rayleigh fading, a 30-dB fade margin would
display something better than a 99.9% path
reliability.
But under the same circumstances with frequency
diversity, with only a 1% frequency separation,
propagation reliability on the same path would be
improved to 99.995%.
LOS Microwave Repeaters

Digital LOS microwave repeaters completely
demodulate the incoming signal to baseband (i.e., to
the raw electrical signal of 1s and 0s.).
This full demodulation causes regeneration of the 1s
and 0s.
The outgoing signal is squared up and retimed.
The regenerated baseband signal is remodulated,
upconverted and retransmitted on a different
frequency.
Repeaters
Repeater receives signal
Repeater amplifies
the signal and
rebroadcasts the
signal
Poor Coverage area
Repeater units are designed to receive signals from a donor site,

amplify and rebroadcast the donor sites signals into poor coverage
areas or to extend the coverage range of the donor site.
350
Path/Site survey
This can turn out to be the most important step in
the design of an LOS microwave link (or hop).
Much of the survey is to verify findings and
conclusions of the path profile.
Of course each site must be visited to determine the
location of the tower and the radio equipment
shelter.
Site/path survey personnel must personally inspect
the sites in question, walking/driving the path or
flying in a helicopter, or a combination of these.
The positions must be accurately enough (<1m) crosschecked in three dimensions.
Frequency planning
The objective of frequency planning is
to assign frequencies to a network
using as few frequencies as possible
and in a manner such that the quality
and availability of the radio link path is
minimally affected by interference.
352
Frequency Planning and Frequency

Assignment
To derive optimum performance from an LOS microwave system,
the design engineer must set out a frequency-usage plan that
may or may not have to be approved by the national regulatory
organization.
The problem to choose frequencies has many aspects.
In practice, the useful RF spectrum is limited to about 150 GHz.
The upper limit is technology-restricted.
To some extent it is also propagation-restricted.
There is the problem of congestion
Those frequencies above 10 GHz could be called rainfallrestricted, because above 10 GHz excess attenuation due to
rainfall can become an important design factor.
Around urban and built-up areas, frequency assignments below
10 GHz are hard to obtain from national regulatory authorities.
However, if we plan properly for excess rainfall attenuation,
nearly equal performance is available at those higher
frequencies.
Some LOS Microwave Frequency Bands
Channel Plan
355
Frequency selection
Scenario 1:
Frequency band and sub-band decided.
Perfect. Early ordering of equipment possible.
Scenario 2:
Frequency band unknown.
Critical. Major changes of assumed bands may
alter the network design.
Application to frequency authority must be
made.
356
Frequency Planning
Determining factors for the influence of
interference:
The carrier to interference ratio (C/I) at the
input of the interfered receiver
The receiver selectivity i.e. receiver bandwidth
and the frequency separation between carrier and
interfering signal
The spectrum width of the interfering signal
The influence of interference can be reduced

by increasing C/I and/or the frequency
separation.
357
Frequency planning/Interference
reducing methods
Frequency (channel) separation
Transmitter attenuation (at interfering
transmitter)
Cross-polar antenna discrimination (V/H)
Co-polar antenna discrimination (directivity)
Larger antennas (smaller beam width and
higher antenna gain)
High performance antennas (side- and back
lobe attenuation)
Reduced spectrum width (low deviation)
358
Frequency planning
The following aspects are the basic
considerations involved in the assignment of
radio frequencies
Determining a frequency band that is suitable for
the specific link (path length, site location, terrain
topography and atmospheric effects)
Prevention of mutual interference such as
interference among radio frequency channels in
the actual path, interference to and from other
radio paths, interference to and from satellite
communication systems
Correct selection of a frequency band allows the
required transmission capacity while efficiently
utilizing the available radio frequency spectrum
359
Frequency channel arrangements

The available frequency band is subdivided
into two halves, a lower and an upper duplex
half.
The duplex spacing is always sufficiently
large so that the radio equipment can
operate interference free under duplex
operation.
The width of each channel depends on the
capacity of the radio link and the type of
modulation used.
360
Frequency planning tools

Frequency planning of a few paths can be
carried out manually but, for larger
networks (= for complex microwave
design), it is profitable to employ some
software transmission design tool.
Tools include standards, different
diversity schemes, diffraction and
reflection (multipath) analysis, rain
effects, interference analysis etc.
361
Path/Site survey
Line of sight must be precisely studied
reiteration for each obstacle in the LOS
microwave path
at least 0.6 of the first Fresnel zone must be
added on top of obstacles including 20 m for trees
and a few meters more for growth if in a
vegetated area (to avoid foliage loss penalties)
often it is advisable to add a few meters of safety
factor on top of the first Fresnel zone clearance
to avoid any diffraction loss penalties.
System tests
A series of tests should be carried out to
verify if the link (or system) meets the
performance requirements.
The first is the measurement of receive
signal level (RSL) at a links far-end receiver.
The second test is the bit error rate test
(BERT).
Ideally, the tests should be done over time.
This means to run the BERT continuously for
at least a few days or more to capture the
effects of fading.
Basic Recommendations
Use higher frequency bands for shorter hops
and lower frequency bands for longer hops.
Avoid lower frequency bands in urban areas.
In areas with heavy precipitation, if possible,
use frequency bands below 10 GHz.
Use protected (stand-by) systems (1+1) for all
important and/or high-capacity links.
Leave enough spare capacity for future
expansion of the system.
364
Common Planning Considerations

Perform detailed path surveys on ALL microwave
hops.
Maps are used only for initial planning, as a first
approximation.
Below 10 GHz, multipath outage increases rapidly
with path length.
Multipath effect can be reduced with higher fade
margin.
If the path has excessive path outage the
performance can be improved by using one of the
diversity methods.
However, diversity is an expensive way of improving
the performance of the microwave link and it should
be used carefully and as a last resort.
365
Difficult Areas for Microwave Links

In areas with lots of rain, use the lowest frequency
band allowed for the project.
Microwave hops over or in the vicinity of the large
water surfaces and flat land areas can cause severe
multipath fading.
Reflections may be avoided by selecting sites that
are shielded from the reflected rays.
366

Introduction to Cellular System
Design
Jorma Kekalainen
Introduction
The objective of a cellular system is to provide quality
communication to the maximum number of users in a defined area.
The number of users supported by the system can be increased by
using more frequencies.
Frequency resources are however always limited.
Hence RF Planning engineers are required to maximise spectrum
efficiency.
In order to accommodate a maximum number of subscribers per
network, the available frequencies need to be reused as often as
possible.
This creates interference towards other cells, which have
detrimental impact to the link quality.
Finding the optimum compromise between dense re-use and least
368
interference is the objective of frequency planning.
Introduction
The system design and planning of the system has to be done so as
to reuse the frequencies as often as possible while keeping the cochannel and adjacent channel interference within acceptable
limits.
Also a minimum received signal level has to be provided throughout
the coverage area of the network.
Frequency planning can be done
Manually by skilled expert RF Engineers.
With powerful planning tool (here used acronym CellCAD)
having the option of automated frequency planning.
369
Cellular network design

Cellular network design is a complicated multidiscipline process.
The fundamental properties of antenna operation is only the
first step in that process
Wave propagation is a much more complicated process than one
might expect by a simple statement like the farther away one
goes the weaker the signal becomes the rate at which such
weakening occurs is very important!
Signal strength is actually AVERAGE signal strength that one
is actually talking about.
Understanding the fluctuations (i.e., fading) around this average
is of paramount importance.
Fading is a catch-all term and one must realize that there are
several different types of fading characteristics of the channel.
Different fading characteristics require different mitigation
solutions.
Mechanisms causing signal variability
Propagation loss
Cellular network design

Cellular network planning and design is
accomplished in a systematical and structured
manner, analogous to that of building a house
You cannot start putting the roof together if you
have not built the foundation!
Similarly, you can not start building the foundation
if you have not made plans in accordance with local
rules, available resources etc.
In network planning these resources involve

both and spectrum (not to mention real
estate etc.)
Pre-design
During pre-design, project leaders study business
planning, a prediction of the number of subscribers,
subscriber usage, target customers, spectrum and
pricing.
Objectives
Main phases
Prephase: Preparation (D0)
Initial design: Phase 1 (D1)
Implementation: Phase 2 (D2)
Optimization: Phase 3 (D3)
Ongoing system improvement: Phase 4 (D4)
Design preparation (Prephase)

Setup HW, SW & DBs
Determine design standards and objectives
Perform model optimization
Link budget analysis and cell size estimation
Design preparation (Prephase)

The design starts with the preparation stage.
Set-up hardware, software, and databases including
measurement test equipment
Determine Design Standards and Objectives:
We select (with the clients approval) a specific technology for
our wireless system and establish fundamental parameters that
affect the quality of the system and define the yearly goals
Perform Model Optimization:

We add or subtract dBs to the signal strengths predicted by
simulations to match the signal strengths measured by the test
equipment.
Estimate Cell Count:

Given an area that requires coverage for a number of
subscribers using a specific mobile, we estimate the number of
cells required.
Initial design (D1)

Coverage design
Capacity analysis
Frequency planning
Preparing site maps
Initial design (D1)

In coverage design, we select site locations and
antenna configurations to provide sufficient signal
strength for mobile stations to communicate in
defined areas.
In capacity analysis, we identify each site that
requires more than the maximum number of channels
allowed and either off-load the excess channels to
surrounding sites or add sites.
In frequency planning, we carefully assign frequencies
to stations without creating co-channel and adjacent
channel interference.
We prepare search area maps to guide real estate
people to the area where we need a site.
Design objectives
Determining our design objectives and
standards is a major step of our design.
We need to know:
How much received power is strong enough?
Where?
How much spectrum or how many channels are
available? Where?
How small a signal to interference ratio is
tolerable? Where?
How much and what kind of demand and growth is
expected?
Objectives
Coverage, Capacity and Interference

(CCI) objectives
The levels of tolerance and desired quality are
location dependent.
CCI objectives
We define different levels of quality
tolerance in different areas.
Typically, we expect to provide better quality
in the downtown area and major highways than
in the outskirts of town.
Coverage objective is the minimum signal
strength that a mobile station requires to
communicate with the base station.
Capacity and interference objectives are sets
of thresholds.
CCI objectives
Databases
To perform simulation and design, we need information in databases about the physical
environment, people, and resources in the target area.
CellCAD or corresponding design tools display some of these databases, such as
terrain, in the form of a color map and others, such as antennas, in a table form.
Example: Terrain database
Example: Terrain database

Terrain database is a matrix that shows the
elevation of different points in the target
area.
CellCAD software uses the terrain database
to predict signal loss from a base station to a
mobile station.
We use the terrain database to help decide
where to place a site.
We can view terrain databases in CellCAD as a
color map where each color represents a
range of terrain elevations.
Example: Morphology database

Morphology database is information about land usage
in the target area.
Different environments attenuate RF signal strength
by different amounts.
In CellCAD, we can assign a dB adjustment to each
type of land usage and CellCAD will automatically
adjust the signal strengths predicted in the area of
each type of land usage.
Example: Antenna database
Antenna database is a table of available antennas with their gains at different angles.
We can view the antenna gains, horizontal patterns or vertical patterns in CellCAD.
Example: Demographic database

Demographic database contains information
about the distribution of people, their income,
and their major commuting paths for a
market.
We assume that demographics data
correlates with subscriber density.
We estimate the correlation, called
demographic weighting factors.
Lacking real subscribers, we estimate
subscriber density and call traffic using
demographic analysis.
Example: Structure database
A structure database contains all the existing structures, i.e., towers and buildings, their
location, ground elevation, and radiation center.
We will use this database in the site selection stage of coverage design.
Loss model optimization
Model optimization is the process of adjusting our loss prediction model to conform with
measurement data collected through drive tests in the target area.
Using a correction factor we incorporate the best dB adjustment.
Morphological classification
We classify the target into only three categories
based on land usage:
urban,
suburban, and
rural.
To balance economics and statistical reliability of

measurement data (a representative sample), at least
two sites from each category for model optimization
are measured.
It is very important to select the right sites to test.
The site chosen should be representative of the
target and the environment type.
Urban areas have man-made structures and little
vegetation.
Local clutter (obstacles) interferes the line-of-sight (LOS)
from the base stations antenna to mobile stations antenna,
since the radiation center is only slightly higher than the
building heights.
Urban sites have relatively small coverage areas.
Suburban areas have a balance of man-made

structures and vegetation.
More LOS areas than in urban areas.
Suburban sites have larger coverage areas than urban areas.
Rural areas have open regions with few man-made

structures.
LOS almost everywhere in the coverage area.
Rural sites have the largest coverage areas.
Correction factors
Morphologically diverse area needs several
correction factors.
Regions with different morphological characteristics require unique correction factors.
Link budget analysis and cell size

estimation
We complete a link budget analysis
to choose an EIRP for a base station
to determine a coverage threshold
to estimate average cell radius
dB and dBm
Note: dBm is a unit for power but dB is a unit for power gain or loss.
Link budget analysis
EiRP = Effective isotropic Radiating Power, BS = Base Station, MS = Mobile Station
Link Budget Analysis (LBA)

The objective of the LBA is to determine the
transmit power required from the BS so that
the maximum allowable path loss in the
downlink is equal to the maximum allowable
path loss in the uplink.
By balancing the maximum allowable path
losses in the two directions, we ensure that
everywhere the BS provides acceptable
coverage for a MS, the MS will have enough
power to reach the BS receivers.
Link Budget Analysis (LBA)

In LBA, we subtract losses and add gains due to
various components of the BS & MS hardware to
compute the maximum allowable path loss from the
transmit antenna to the receive antenna, for both the
uplink and the downlink.
The maximum allowable path loss in the uplink is the
difference between the power out of the MS and the
minimum required power at the BSs antenna.
Similarly, the maximum allowable path loss in the
downlink is the difference between the power
transmitted out of the BS antenna and the minimum
required power at the MSs antenna.
We assume that the environment losses are the same
for the downlink and uplink.
Coverage classes
The mobile station requires at least a signal strength on the street or at the mobile
stations receiver. To evaluate coverage of a mobile station in a car, we added 10 dB to
the power required on the street. To evaluate coverage of the mobile station in the
building, we added 20 dB to the power required on the street.
Hand-off window
Hand-off window
Switching back and forth between two
stations due to signal level fluctuations
increases the mobiles chances of dropping a
call and overhead signaling to complete a
hand-off.
Therefore, multiple hand-offs (ping-pong
effect) are undesirable.
The hand-off window ensures that the signal
strength of a neighboring station is at least 3
dB stronger than the signal strength of the
serving station before the call is handed off
to the neighboring station.
Hand-off window
Comment: Hand-off window

The threshold that defines the dotted circles
is the threshold necessary for good coverage
(derived from the LBA).
The threshold that defines the solid circles is
the coverage threshold plus the hand-off
window.
By making the solid circles touch each other,
we provide sufficient overlap between the
dotted circles for hand-off purposes.
Theoretical cell size estimation
After LBA is done and the balanced path is achieved from path loss and the propagation
loss model (e.g. HATAs model) we can estimate the cell radius.
Note that this method is crude and does not account for the terrain.
Cell size estimation

Cells radius may vary
in different directions
We need to take the

average of radii in
different directions
After running propagation for a site with a typical antenna type, radiation center, and
EiRP in CellCAD, we can plot the signal predictions to use for estimating a cell size.
The coverage threshold plus the hand-off window is the threshold that determines the
cell radius.
Because the cell radius varies in different directions, we take the average of radii in
several directions.
Cell size estimation

After measuring the cell radius for several test sites,
we average the radii to estimate the typical cell
radius.
We estimate a cell size for each area based on a
morphology category and a receiving environment for
the typical mobile station.
Reduce R to meet capacity

requirements
Each station has an upper limit to the amount of call
traffic that the station can serve.
Fill up the coverage objective area with cells of

predicted size and count the number of sites
Estimated cell count
Multiple cell sizes

Different needs:
SeveraI LBAs
Several path losses
SeveraI cell size estimations
If the target area is not homogenous in terms of morphology, call traffic demands,
dominant type of phones, and phones environment, we estimate several cell sizes.
After classifying the areas, we estimate a unique cell size for each area.
Cell count estimation with

multiple cell sizes
After estimating one cell size for each region, we fill
up the coverage objective area with cells of
estimated sizes and count the total number of cells
Iteration is the essential of initial

design (D1) phase
Coverage Design
In coverage design, we select site locations and
antenna configurations, in a sequential manner, to
provide sufficient signal strength for mobile stations
to communicate in defined areas.
Topographical maps
Topographical maps show the ground elevation using
contour lines.
Coverage design getting started
We calculate our coverage threshold in the link budget analysis.

Coverage thresholds are different for different receiving environments and technology
standards. Note: par=nominal value.
Coverage design I
For regularity of design, we use a hexagonal grid pattern to guide our selection of sites.
This regularity helps us in the next stages of design, i.e., capacity analysis and
frequency planning.
Setting hexagonal grids

We build a hexagonal grid using the estimated cell size for each
region defined by the morphology category and receiving
environment.
Selection of the first site

Typically, we start selecting sites from the core of the target
and continue towards the boundaries.
We follow this approach because normally coverage in the core
area is most important and less flexible.
Site selection considerations

Candidate sites
Structure database
Trade off rules
Zoning issues
Target/market visit information
Average antenna height
Special considerations in
coverage design
Areas that are physically far apart may
become close from a propagation viewpoint
because of water enhancement.
The EM loss over water is less than over dry
ground and CellCAD accounts for this
difference.
In site selection, we need to watch for extra
propagation from sites close to water because
the propagation could cause interference.
coverage design: Tilt
During the D1 stage of coverage design, we design sites using one type of omni
antenna placed within a range of heights.
In the D1 stage, we rarely downtilt an antenna because we save time by waiting to
recommend antenna downtilts during the D2 stage when we have real site locations.
Some antenna rules

Select all radiation
centers within a
given range
Use only one type of
omni & directional
antenna
Always start with
omnis.
No antenna tilt is
recommended.
coverage design
Special considerations in coverage

design: Tunnels and bridges
To get coverage over a bridge, in a tunnel, or
trenched roadways, we select sites that are
in line of sight of the bridge, tunnel, or
trenched roadway.
CellCAD predicts signal strength over terrain
or around buildings, but not in tunnels or over
bridges.
We rely on field measurements to determine
if we have coverage over a bridge or in a
tunnel.
Propagation in tunnels
Propagation inside tunnels is inhibited because of
excessive loss at obstructions, bends, and corners.
For propagation studies, a tunnel is often modeled as
a smooth-walled, glossy, straight, and homogeneous
wave guide.
For such a model, attenuation varies as the inverse
square of frequency.
Tunnels
Propagation in tunnels is difficult and special modeling
and design methodology must be used.
In the mobile communications industry, this has
evolved into a specialized domain, with many
techniques including such design approaches as leaky
cables, closely located repeaters, etc.
Example: Lincoln Tunnel

From a study in Lincoln Tunnel, Manhattan in
New York, propagation loss decreases as
frequency increases with the following
equation
Above 2 GHz, attenuation is linearly proportional

to frequency with the approximate formula of
a1.2.
Example: Lincoln Tunnel

Right figure shows loss as a function of antenna separation
at 980 MHz.
Attenuation varies inversely as the fourth power with
distance resembling the plane earth model
Propagation over water
Foliage
Foliage losses
The effects of foliage are considerable at UHF and
higher frequencies.
Foliage effects vary with seasons, and thus present
special design problems.
Sometimes, significant design alterations are
necessary as the seasons change, thus the term
multi-season design.
coverage design: Delay spread
In selecting site locations, we need to look for reflecting
surfaces, i.e., (glass) buildings and (snowcap) mountains, that
create time dispersion.
Capacity analysis
Objective:
Estimate traffic/channel loading per station
Put additional sites wherever needed
that too heavy traffic load can distribute fairly
Demographic weighting factors

Demographic maps show bin by bin distribution of
information about people such as vehicular traffic,
income, business, and population.
The weighted average of these maps is our estimate
for subscriber density.
We choose weighting factors based on experience in
similar existing markets.
Subscriber density maps
The resulting subscriber density is a map or matrix with the number of subscribers in
each bin.
Distribution of traffic for bins

on the cell boundaries
On the cell boundaries, we distribute demand in a bin to several
base stations that could possibly serve the area.
We measure the demand on the station either as number of
subscribers or erlangs.
Coverage area and server list

The stations coverage area is the sum of the bins that the
station potentially could serve.
Traffic parameters
Traffic parameters
The incoming traffic load is a random process,
therefore, we should employ a probabilistic
model with parameters such as:
Average number of users in the system.
Average amount of delay for each call.
Average total number of subscribers predicted or
planned.
Probability of blocking a call request
System queuing strategy
Call arrival rate
Blocked, connected and held calls

In general, the system cannot serve or connect some
of the arrived calls.
Either the system immediately clears the unserved
calls or holds the unserved calls in a finite queue
before clearing them.
If the system has a queue, the call can stay in the
queue for a limited period of time before the system
either serves or clears the held call.
Blocked, connected and held calls
Erlang and subscriber usage

Erlang: is a quantitative measure of telephone
traffic load
1 Erlang = One circuit occupied for one hour
An Erlang has no units

# Erlangs = average call arrival rate times average
call duration
1 Erlang = 1 call/min * 1 min/call or 2 calls/min *
0.5 min/call
Subscriber usage is measured in Erlangs per

Subscriber.
Traditionally, we estimate call traffic at the
busiest hours.
Subscriber usage
Erlangs/station
For each station or cell, CellCAD calculates the call traffic in Erlangs by multiplying the
number of subscribers in the cell area by the subscriber usage.
Traffic models
Example
Limitations
Hardware limitations, spectrum availability, selected reuse
scheme, and zoning issues are inputs that we use to calculate the
maximum number of channels on each station.
Channel off-loading
Reduce the Coverage Area of Station by:
Reducing its Antenna Height (effective)
Using an Antenna with lower gain (not effective)
Reducing its power output (not effective and not
recommended)
Down tilting (effective, not to be used in D1)
To fill the gaps,increase the coverage area of

neighboring sites by increasing their antenna
height.
Channel off-loading
First, we try to off-load traffic from the overloaded
site to neighbor sites.
Cell splitting
Second, we try to add a site to off-load traffic from
the overloaded site.
Capacity analysis Frequency

planning
After capacity analysis we should have a list of the
channel loading for all the sites in the target area,
which along with the site configuration is used for
frequency planning.
In frequency planning, we carefully assign frequencies
to stations without creating co-channel and adjacent
channel interference.
RF planning
The goal is to achieve optimum use of resources and maximum
revenue potential whilst maintaining a high level of system
quality.
Full consideration must also be given to cost and spectrum
allocation limitations.
A properly planned system should allow capacity to be added
economically when traffic demand increases.
As every urban environment is different, so is every macrocell and
microcell network.
Hence accurate planning is essential in order to ensure that the
system will provide both the increased capacity and the
improvement in network quality where required.
454
RF planning
RF planning plays a critical role in the cellular design
process.
By doing a proper RF planning we can reduce a lot of
problems that we may encounter in the future and also
reduce substantially the cost of optimization.
On the other hand a poorly planned network not only
leads to many network problems , it also increases the
optimization costs and still may not ensure the desired
quality.
455
Tools for RF planning

Tools used for RF planning
Network Planning Tool
Propagation Tool
Traffic Modeling Tool
Project Management Tool
456
Network Planning Tool

Planning tool is used to assist engineers in designing
and optimizing wireless networks by providing an
accurate and reliable prediction of coverage, doing
frequency planning automatically etc.
With a database that takes into account data such as
terrain, clutter, and antenna radiation patterns, as
well as a graphical interface, the Planning tool gives
RF engineers a state-of-the-art tool to:
Design wireless networks
Plan network expansions
Optimize network performance
Diagnose system problems
Some tools available in the market.

Also many vendors have developed Planning tools of
their own.
457
Frequency planning
How many channels are needed and where?
Station by station reuse penalty?
Exclusion zones?
Which RF channels should be used and where?
Big picture
Before assigning frequencies we create a

reuse matrix to decide which sites need to
be sectored.
We sectorize the sites identified through the
reuse matrix.
We try Automatic Frequency Planning in
CellCAD. If APP does not work, we assign
frequencies manually.
Interference
analysis
Interference
reduction
Automatic frequency
planning
Manual frequency
planning
Carrier to interference ratio

Carrier to interference ratio (C/I) is the
ratio of total power from the desired station
(C) and from all undesired stations
broadcasting on the same frequency or
adjacent frequencies.
The effect of input tuned filters in rejecting
adjacent channels is considered in term of
their adjacent channel rejection factor.
CellCAD calculates one C/I for co-channels
and another C/I for adjacent channels.
Carrier to interference ratio
Reuse distance
Reuse distance
We increase our system capacity when we decrease
the distance between sites that use the same RF
channel.
Close reuse means small D/R, not small D.
The trade-off to increased capacity is an increase in
interference because the co-channel sites have less
distance between them.
Reuse patterns
A reuse pattern means that we can use all our spectrum on every
cluster of K sites.
4, 7, 9, and 12 are typical cell reuse patterns.
C/I
This theoretical equation based on simplistic assumptions provides a rough estimate of

an appropriate reuse factor for the selected C/I objective.
Sectorization
Smart or adaptive antennas

Feedback and control
Adaptability
Adaptable beam forming and steering
Interference suppression
Dump vs. smart antenna
Dump sectorized antenna
Smart multi-beam antenna
Smart antenna
Profits
Gain
Beamwidth
Interference
reduction
Adaptive coverage
Loss
Cost
Complexity
Likely server
Likely server
For interference calculations, CellCAD determines the serving
area based on likely server or secondary threshold.
We use interference based on likely server most often.
In a bin for interference based on likely server, CellCAD
1) filters out all stations that do not have a signal strength above
the system threshold before identifying possible servers (>Tsys)
2) identifies station A as a potential server if its signal strengths is
greater than its primary threshold.
3) identifies the best server, the station with the strongest signal
strength, and classifies this station as a potential server.
4) identifies station A as a potential server if its signal strength is
greater than the signal strength from the best server minus station
As hand-off window.
Note, to define the same serving area as likely servers, set the
secondary threshold equal to the secondary threshold for
demographics based on cell hand-off.
Interference weighting factors

In different parts of the target area, we can vary our
tolerance and express it as a percentage of all
covered bins that can have interference (
C/I<Threshold)
Example
We quantify tolerable interference as a percentage
of all covered bins that can have interference.
Percentage interference
=10/32*100=31%
Example: Average C/I
Average C/I over interference bins:
Two tier buffer of cells
Two tier buffer of cells

We are interested to know how many times a
frequency group is used among potentially interfering
stations.
Theoretically for a 7 cell reuse every frequency group
is used three times within the first two tiers of cells.
Therefore if we assume that interference is limited
to the cells within the first two tiers around each cell
then R=3.
Theoretically for 4 cell reuse it can be shown that
R=2.
In a real scenario based on practice we choose R=2.
Automatic Frequency Planning

(AFP)
Adjacent channel assignment
Redesign and optimization
Redesign and optimization

If interference problems can not be solved after
sectorization procedure we may have to change
radiation center or other optimization based
adjustments.
Eventually if adjustments do not solve the problem we
need to reselect the problem site.
Frequency group table

How do we group available RF channels to assign one
group to each station?
Example: Frequency group table
Example: Incorrect frequency

grouping
Adjacent frequency assignments

Assigning adjacent channels to neighboring stations
causes problems especially during hand-offs.
Frequency assignment I:
Automatic Frequency Planning
works
Frequency assignment I: Manual

operation
Frequency assignment II
Model optimization
Model optimization is the process of adjusting
our loss prediction model to conform with
measurement data that are collected through
drive tests in the target area.
As a result of this measurement integration
process, we find a correction factor, the
standard deviation of prediction errors, and a
color plot that shows the value of error at
each point on the drive-test route.
Measurement integration
Measurement integration for model optimization
Drive-test plan
After selecting the sites to be drive-tested,
the next step is to formulate a drive test plan
for each site.
The following rules should be observed.
The drive test plan should include both radial and
circumferential routes.
It should also include wide & narrow roads.
It is important that radial roads be drive-tested
till the edge of the cell boundary.
It is beneficial to measure beyond the coverage
contour up to the interference contour of the site.
These contours are based on set threshold values.
Drive-test plan
Correction Factor
Every measured point is compared to its
corresponding predicted signal strength and
the error as well as co-ordinates are
recorded in a file in CellCAD.
The model is adjusted by adding a correction
factor to the predicted signal strength
values.
The correction factor to be added is the
average value of the differences between
measured and predicted signal strength
values.
Correction Factor
Predicted power loss (Lp)
Correction Factor and

distribution of errors
Desired test data

We would like to have zero mean prediction error.
Adding a correction factor shifts the mean of error distribution
to zero but does not change the variance of error.
Decreased variance of prediction error means an increase in the
models accuracy.
If after measurement integration the result of a drive tests
show a large standard deviation, it means there is an error in
drive test or we may need to put breakpoints and consider
separate correction factors for areas between each two break
points.
For DO we do not use break points.
Any drive test data with standard deviation more than 10 dB is
not acceptable.
Recap: Main design phases

D0: Preparation
D1: Initial design
D2: Implementation
D3: Optimization
D4: Ongoing system improvement
Recap: Design preparation (D0)

Setup HW, SW & DBs
Determine design standards and objectives
Perform model optimization
Link budget analysis and cell size estimation
Recap: Results from initial design

(D1)
Coverage objective maps
Traffic parameters
Reuse factor
Spectrum
C/I thresholds
Interference weighting factor
Antenna parameters
Demographic weighting factors
Recap: Initial design (D1)

Coverage Design
Capacity Analysis
Frequency Planning

Design of Optical Fiber Systems
Jorma Kekalainen

Optical Fiber Systems
Operating Wavelength Bands
Advantages of Fiber
Advantages of the fiber transmission media
Low transmission loss (typically 0.2-0.5 dB/km)
Allows longer distances between repeaters or amplifiers
By comparison, Cat. 5 UTP (copper pairs) have loss of 7
dB/km to 220 dB/km in 64 kHz-100 MHz range
504
Advantages of Fiber
505
Advantages of Fiber
Larger information carrying capacity
506
Note
The Infrared Data Association (IrDA) defines physical specifications
communications protocol standards for the short-range exchange of
data over infrared light, for uses such as personal area networks
(PANs).
IrDA is a very short-range example of free space optical
communication.
Ultra-Wideband (UWB) is a technology for transmitting information
spread over a large bandwidth (>500 MHz)
UWB can be used at very low energy levels for short-range highbandwidth communications
802.20 is a proposed IEEE specification for boosting IP-based datatransmission rates for mobile users in wireless metropolitan area
networks (WMANs).
Formally known as "The Standard Air Interface for Mobile Broadband
Wireless Access Systems Supporting Vehicular Mobility - Physical and
Media Access Control Layer Specification," 802.20 would support
transmission speeds of up to 1M bit/sec in the 3-GHz spectrum band.
The goal is to have a ubiquitous data wireless network that can support
real-time traffic with extremely low latency at 20 milliseconds or less.
Advantages of Fiber
Immunity to electromagnetic interference
Can be placed alongside powerlines or close to radiative equipment
More secure to eavesdropping

Smaller size and weight
Example: 700 km of copper cabling weights 20 tonnes, while same
cable run with fiber weighs 7 kg
508
Basics of Fiber Propagation

An optical fiber is composed of:
Cylindrical core: refractive index n11.5
Cladding: refractive index n2 < n1
Buffer (or primary coating): protects fiber from
damage
509
Fiber Refractive Index Profile

Core Refractive Index (n1)
Cladding Refractive Index (n2)
Step Index Profile
Graded (Quadratic) Index Profile
510
Light Transmission in Fiber

A simple explanation via Ray Optics
511
Modes of a Fiber
What makes a fiber single mode or
multimode?
512
Fiber Attenuation
As light travels along a fiber, its power
decreases exponentially with distance L
513
Fiber Attenuation
Attenuation coefficient preferably
expressed in units of dB/km
dB is logarithmic unit for representing gain
or loss
dBm is logarithmic unit for absolute signal
power in mW
514
Fiber Attenuation
What causes fiber loss?
Absorption
Scattering
Bending
515
Fiber Dispersion
Dispersion = different
components of the
signal travel at
different velocities
Pulses spread in time
Causes intersymbol
interference (ISI)
more errors
Limits possible distance
and bit rate
516
Fiber Limitations
Link performance is limited by:
517
Wavelength-Division Multiplexing
Wavelength-division multiplexing or WDM
Frequency-division multiplexing in the optical
domain
Multiple information-bearing optical signals
transported on a single strand of fiber
518
519
Current WDM systems
Dense WDM (DWDM)
ITU-T G.694.1 grid with
channel spacing 200
GHz
C- and L-band (1530-1625
nm) operation
Coarse WDM (CWDM)

ITU-T G.694.2 grid with
2500 GHz channel spacing
0-, E-, S-, C- and L-band
(1260-1625 nm) operation
520
Intercontinental Optical Networks

About 70% of earths surface covered
by water
For intercontinental (global)
connectivity communication links must
traverse water
Overhead in the sky via LEO, MEO or
GEO satellites
Underwater using submarine cables
521
Intercontinental Optical Networks

Majority (~90%) of intercontinental traffic now
carried on fiber rather than satellite
522
Backbone Optical Networks

Continental backbone network providing connectivity
between different countries
523
Backbone Optical Networks

National backbone network interconnecting cities and
main towns of a country
Corenet (VR Group)

fiber backbone
Optical Metro Networks

Provide connectivity within a city/metro or region
524
Access Networks
Access network are last leg of
telecommunications network
Between service provider distribution facility and
users home or business
Other names:
last mile
local loop
first mile
Fiber is increasingly deployed now in access

networks
525
Access Networks
DSL = Digit Subscriber Line, DSLAM = Digit Subscriber Line Access Multiplexer, MxU
526 and
= a generic term for Multiple Tenant Unit and Multiple Dwelling Unit, SME = Small
Medium Enterprise

Fiber Optic System Design
Design Approach
The first step in designing a fiber-optic
communication system is to establish the basic
system parameters.
Among these we would wish to know at the outset:
Type, bit rate and format of signal to be transmitted
(e.g., analog, PCM, SONET, SDH or digital TV).
System length, fiber portion, end-to-end.

Growth requirements (additional circuits, increased bit
rates).
This could mean total number of fibers, use of WDM or both.
Tolerable signal impairment level stated as signal-to-noise

ratio or BER at the electrical output of the terminal-end
detector.
BER for link budgets

The link BER should be established based on end-toend requirements.
It is recommended using one of the standards
accepted worldwide
E.g. 10-12 or 10-10 as a link BER requirement.
Here well set the BER for practice link budgets at

10-10.
Throughout the design procedure the system
engineer is working in the power-limited domain or
dispersion-limited domain.
At the lower bit rates (<622 Mbit/s) one can expect
to work in the power-limited domain under all
circumstances.
ISI
System design can eliminate a number of
major causes of dispersion.
Lets simply look at dispersion as a delay.
We have a stream of bits.
The first bit in the stream does its job, but there
is some power from that bit that is delayed which
slips into the time slot of bit two.
If there is sufficient power from bit one in time
slot for bit two, the receiver is confused and may
make an incorrect decision.
As the link bit rate is increased, pulse widths get
shorter and the problem of dispersion becomes
more acute.
Eliminate dispersion
One example of eliminating the cause of dispersion
deals with the type of fiber we select.
To eliminate multimode dispersion, use single mode
fiber.
We then can turn to using the zero dispersion
wavelength which is at approximately 1310 nm for
production silicon fiber.
By doing this we remove the opportunity to use the
low loss hand at about 1550 nm.
To overcome this shortcoming, we spend more on
fiber and buy dispersion-shifted fiber.
That moves the zero dispersion wavelength to the
1550 band by changing the fiber geometry.
Chromatic dispersion
The designer is now left with chromatic
dispersion.
This is the phenomenon where even with
the narrow line width of a laser diode
different frequencies appearing in that
spectral line travel at different
velocities.
Important factors
The designer must select the most economic alternatives among
the following:
Fiber parameters: single mode or multimode: if multimode, step
index or graded index: number of fibers.
Transmission wavelength: 820 nm, 1 330 nm or 1 550 nm.
Source type: LED or semiconductor laser: there are subsets to
each source type.
Detector type: PIN or APD.
Use of EDFA (amplifiers).
Repeaters, if required and how they will be powered.
Modulation will probably be intensity modulation (IM), but the
electrical waveform entering the source is important (coding).
EDFA
In the Erbium Doped Fiber Amplifier
(EDFA) the core of a silica fiber is
doped with Erbium ions and can be
efficiently pumped with a laser at a
wavelength of 980 nm or 1,480 nm.
EDFA exhibits gain in the 1,550 nm
region.
Splices and connectors

There is the splice and connector trade-off
as well as the type of splice and type of
connector.
Permanently installed systems would opt for
splices because of lower insertion losses.
Temporarily installed systems such as used by
the military in the tactical environment may
prefer connectors because of ease and speed
of mating and demating.
Loss Design
As a first step, assume that the system is power-limited.
Probably a majority of systems being installed today can stay in
the power-limited regime if single mode fiber is used with
semiconductor lasers.
With systems operating at such high data rates such that
chromatic dispersion may be a problem, the selection of the
laser and the design of the transmitter itself become important.
It is highly desirable to minimize the line width, and we can
achieve very narrow widths by using a distributed feedback
(DFB) laser and an external modulator.
It also may be wise to select dispersion-shifted fiber, where
the zero dispersion line is shifted to the low-loss 1550-nm hand.
We now get the best of both worlds for extra cost.
Link budget
The next step is to develop a link budget, which in format is very
similar to the link budgets in LOS microwave and satellite
communications.
It is a tabular format where the first entry is the transmitter
output.
If the transmitter initially selected is a laser diode, a 0-dBm
output is a good starting point.
For shorter links with lower bit rates, the LED transmitter
should be a candidate because of cost, lower complexity and
longer life.
Then all the losses/gains of the link are entered, enumerating
and identifying each.
Among these losses we would expect to find the following items:
1. Connector Loss
There are two connectors, one at the
output of the transmitter pigtail and
one at the input of the receiver pigtail.
A pigtail is a short length of factoryinstalled fiber
Budget 0.5 dB for each.

Enter minus signs for losses, plus signs
for gains.
2. Fiber Loss
The fiber selected for the link will have
a loss specified by the manufacturer
given in dB/km for a particular
wavelength of interest.
Multiply that value by the length of the
link plus 5% for slack.
3. Splice Loss
Assuming the link is more than 1 or 2 km
long, there will be a splice to connect
the fiber from one reel to the fiber of
the next reel.
Good fusion splices have a very low
insertion loss.
Budget 0.1 dB each.
Multiply this value by the number of
splices in the link.
4. Amplifier Gain
Links longer than 30 to 100 km (depending on
design) will use an amplifier.
Budget +30 dB for each in-line amplifier.
There is a trade-off where the amplifier is
installed.
Another candidate location is at the input to the
receiver to extend the receiver threshold
Another is at the output of the transmitter to
increase output value.
5. Dispersion Compensation Loss

Budget 1.0 dB for this value.
6. Link Margin Reserve

This dB loss value is set aside for the
following contingencies:
Cable reel loss variability.
Future added splices due to cable repair and their
insertion loss.
Component degradation over the life of the
system.
This is particularly pronounced for LED output.
Many designers reduce the margin value as

much as possible.
ITU-T recommends 3.0 dB.
Major differences
One major difference between a link
budget for LOS radio and for fiber
optics is that there is no fading on a
fiber-optic link.
Another consideration is the use of an
optical power attenuator on very short
links so as not to overload the receiver.
Link budget calculation

The next step is to sum the losses and gains
with the output value of the transmitter.
The summed value should be stated in dBm.
This value helps the system designer to select
the type and make of the receiver, whether a
PIN diode or an APD.
The receiver threshold in dBm is established
for the bit rate and BER desired.
For example values at 10-10 BER for some of the
more popular SONET/SDH bit rates.
Receive Levels, BER Values, and Bit Rates for PIN Diode
and APD Light Detectors
Extinction ratio penalty

One imagines that with intensity modulation, a
binary 1 is represented by the on condition
and a binary 0 by an off condition.
This is not true in the case of the off
condition.
The laser is not completely off: it still radiates a
small amount.
We take this low output level into account as

an equivalent loss, a extinction ratio penalty,
stemming from the laser diode extinction
ratio.
Extinction ratio
The extinction ratio is then
defined as
where P0 is the laser diode

output power for the binary
0 condition and P1 is the
output power for the binary 1
condition.
We would want the value of
rex to be as small as possible.
In many practical cases it is
less than 0.05. This derives a
power penalty of <0.4 dB
according to the following
curve
Example
The link operates at 2.5 Gbps and is 300 km long.
The link operates in the 1550-nm band and the loss is
0.25 dB/km.
The total fiber necessary is 1.05 300 km or 315 km.
This includes the necessary slack.
There will be 314 fusion splices and two connectors.

The light source is a distributed feedback laser diode
with an extinction ratio penalty of 0.4 dB.
The link BER is 10-10.
The candidate light detector is a PIN diode and with
the specified BER the receiver threshold is -23 dBm.
Distributed feedback lasers (DFB) are the most common transmitter type in DWDMsystems. To stabilize the lasing wavelength, a diffraction grating is etched close to
the p-n junction of the diode. This grating acts like an optical filter, causing a single
wavelength to be fed back to the gain region and lase. DFB lasers are the workhorse
of demanding optical communication.
Example: Fiber-Optic Link Budget
Shortfall compensation
To compensate for the shortfall of 3.55 dB the
following are possible measures that can be taken:
1. Shorten link by 15 km.
2. Increase gain of each amplifier by 1.2 dB, assuming that
EDFAs were operating at less than full gain.
3. Increase output power of laser diode transmitter by 3.55
dB.
This may shorten the life of the device.
4. Reduce margin accordingly.

Highly undesirable.
5. Turn to using an APD rather than a PIN diode.

The threshold is found to be -32 dBm or a 9-dB improvement.
The APD has a shorter life than a PIN diode: its reliability is
lower: it is more expensive and it is temperature and humiditysensitive.
On the positive side, the margin has been increased by 5.45 dB.
Link and System Design

Simple fiber-optic communications link
Short distance
Low bit rate
Point-to-point
Major concern is to ensure sufficient received power

Link power budget analysis
552
Link Power Budget
553
Link Power Budget

A power budget for an amplified WDM
link
554
Detailed System Design

In an amplified WDM link, there is more to
worry about than just power budget
Non-ideal optical devices (transmitters, filters
etc.)
Fiber links longer
Increased transmission loss
Dispersion and fiber nonlinearity more severe
Tightly packed wavelength channels

Interference between different channels
Cascaded optical amplifiers

Provide gain but also noise and unequal gain at different
wavelengths
Gain affected by power transients
555
Detailed System Design
556
Nonlinear Schrodinger equation

Propagation of optical pulse over fiber modeled by
the nonlinear Schrodinger equation
Maxwells equations in cylindrical coordinates and with
boundary conditions of fiber optic cables
Some terms ignored for pulses >10ps (<100 Gbit/s NRZ)
557
Case Study
The following parameters are established for a longhaul single-mode optical fiber system operating at a
wavelength of 1.3 m.
Estimate
1) the maximum possible link length without repeaters
when operating at 35 Mbit/s (BER 10-9).
It may be assumed that there is no dispersion equalization
penalty at this bit rate.
2) the maximum possible link length without repeaters

when operating at 400 Mbit/s (BER 10-9) and
assuming no dispersion equalization penalty.
3) The reduction in the maximum possible link length
without repeater of (2) when there is a dispersionequalization penalty of 1.5 dB.
It may be assumed for the purpose of this estimate that
the reduced link length has the 1.5 dB penalty.
Solution
1)
At 35 Mbit/s,
Solution
2) At 400 Mbit/s,
3) Add the dispersion-equalization power penalty DL
There is a reduction of 3 km in the maximum possible

line length without repeater.
WDM Network Design

Optical layer (WDM network) design
Realizing connections over physical fibers using WDM
network elements (OADMs, OXCs, OLTs etc.)
Lightpath = an optical connection assigned a distinct
wavelength over a span of fiber(s)
562
WDM Network Design

Physical topology
Optical nodes and fiber topology supporting the creation of
lightpaths
Lightpath (logical or virtual) topology

Topology seen by the client layer equipment (e.g. IP routers)
563
WDM Network Design

In practice WDM network design split into separate
(manageable) design problems
Less complex to solve
A WDM network may be realized by solving:

Physical topology design (PTD) problem
Lightpath topology design (LTD) problem
Routing and wavelength assignment (RWA) problem
PTD problem solved initially during network

construction and later when physically expanding the
network
The LTD and RWA are solved more frequently to
enable optimum provisioning of optical capacity
564
Physical Topology Design (PTD)
565
Light path Topology Design (LTD)

LTD problem
Interconnecting client layer equipment to realize a light path
topology that meets client layer traffic requirements
Routing client layer packets or TDM circuits over the light
path topology
LTD constraints
Flow conservation (net flow of traffic in/out of nodes)
Link capacity
Node degree (number of in/output links)
566
Routing & Wavelength Assignment (RWA)

RWA problem Given a physical network topology and a
set of end-to- end light path requests determine
route and wavelengths
Routing sub-problem = how to route a light path through
various intermediate nodes within a network
Wavelength assignment sub-problem = how to assign
wavelength channels to different light paths
Possible light path request blocking (denial) if no free channels
available
567
Routing & Wavelength Assignment (RW)

Constraints in routing sub-problem
Route (fiber) length
Accumulated fiber impairments (EDFAs, OEOs, FEC etc.
required)
Hop number (nodes passed)

Accumulated crosstalk, insertion loss, polarization dependent
loss etc.
568

Constraints on wavelength sub-problem
Number of wavelengths
Only 18 channels available on CWDM wavelength grid
<200 channels available on DWDM wavelength grid
Wavelength continuity
Same wavelength channel must be used between end-points
Distinct wavelength assignment

Wavelength only used by one lightpath on any link
Wavelength assignment for a 3 node network

569

RWA problem may be classified as static or dynamic
Static RWA problem
Entire set of connections (demand traffic matrix) known in
advance
Routing and wavelength assignment performed offline
Possible over estimates in requirements = idle capacity
Dynamic RWA problem

Routing and wavelength assignment performed online
Time varying demand matrix
Lightpath established for each connection as it arrives and is
released after some finite time efficient capacity
utilization
Connection request blocking is possible
570

Wavelength converters (WCs)
Devices that change the wavelength of input signal to a
different wavelength on the ITU grid
Implemented using transponders (OEOs) or preferably alloptical WC technologies
Has significant impact on RWA problem
571

WCs eliminate wavelength continuity constraints of
the wavelength assignment sub-problem of RWA
Increases wavelength channel reuse reduce number of
required wavelengths
Reduces probability of connection blocking
Wavelength assignment for a 3 node linear point-to-point network, where

572 2
wavelength channels needed between node 1 and node 3.

Deployment of full WCs
in all network nodes is
costly and/or infeasible
Strategic placements of
WCs in parts of network
with heavy loads
Sparse WC placement
problem
There has been many
proposals for WC
placement algorithms
573

Alternatively deploy cheap but limited or fixed WCs
Wavelength conversion classified according to
flexibility of output wavelength:
574

Telecommunication Network Design by Jorma Kekalainen

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Telecommunication Network Design by Jorma Kekalainen

Transféré par

Droits d'auteur :

Formats disponibles

Telecommunication Network Design

Usually, the most precise sources are the original

Telecommunication Network Design

What is telecommunication network?

Classification of networks (1)

Classification of networks (2)

The long term trend is for universal broadband

Design as a process can take many forms depending

to establish a good design for a system.

What is Network Design?

Focused on capacity planning

A bridge is a device that passes information between two network segments. It

System and systems approach

Network design can then interpret

Systems Approach for Network

What is a communications link?

Links have performance attributes

Elements of a communication system

Example: Fast Ethernet LAN

for 10-11 BER

Link design Power budget

PIN sensitivity at 100Mbps

Allowed loss PT=PS-PR

Source connector loss

2 Jumper connector loss

Cable attenuation (160 m)

Receiver connector loss

Measures of network performance

Data throughput or capacity.

the ability to resist unwanted eavesdropping,

Common problem issues for networks

Iterative Analysis/Design process

Develop the logical design.

Develop the physical design.

Test, optimize, and document

Analysis and Design Processes

Optimization with trade-offs

General Design Approaches

How to inspect a system?

What is simulation and network simulation?

Reasons for network simulation

Reasons for network simulation (2)

A simulation tool is more preferable

The Science of Network Design

Example: Scalability of architectures

File Distribution: Server-Client vs P2P

ui: peer i upload

di: peer i download

File distribution time: server-client

File distribution time: P2P architectures

Minimum distribution time of P2P

dP2P max { F/us, F/min(di) , NF/(us + ui) }

Equation provides a lower bound for the minimum distribution

dP2P max { F/us, F/min(di) , NF/(us + ui) }

File distribution time: P2P

server must send one copy:

fastest possible upload rate:

us + ui=(10+N)u Nu, when N>>

dP2P = max { F/us, F/min(di) , NF/(us + ui) }

Server-client vs. P2P: example

Minimum Distribution Time

Example: Caching (cont)

Example: Caching (cont)

possible solution: install