Vous êtes sur la page 1sur 17

WCDMA Advantages

and a Comparison to TD-CDMA 1.

03/20/13

Introduction
1.1 UMTS main requirements
The main requirements on UMTS/IMT-2000 is the ability to support high bit rates and new multi-media services. The goal is to support at least 384 kbit/s with wide area coverage, and up to 2 Mb/s with local area coverage. The main application for the high bit rates will probably be on packet access, e.g. supporting access to internet/intranet. The radio-interface should also be very flexible, to support new unforeseen services. This means that the radiointerface should support mixed services, with different bit-rate/quality requirements. It also means that that bitrate variability should be supported, e.g. variable rate voice codecs. The flexibility aspect also includes efficient packet transmission, both for short and long packets of data. In the GSM migration to UMTS it is not only the common core network evolution that is important. Another aspect is dual-mode terminals. In the early deployment of a UMTS system it is crucial to have GSM as fallback for coverage. Also the multi-media services should be kept when going out of UMTS coverage, possibly with a bit-rate reduction. This also means that handover between GSM and UMTS must be supported.

1.2 Outline
In Chapter we give an intuitive motivation why Wideband CDMA is the best proposal for UMTS and IMT2000. Chapter describes the WCDMA presented by the Alpha Concept Group in SMG2, see also []. The key issues of WCDMA is described in Chapter . In the SMG2 process it seems like the Alpha and the Delta group has got the most attention so far. Therefore the Delta concept is described in Chapter , see also [], followed bybefore a comparison between the Alpha and the Delta concepts is givenshown in Chapter .

2. Motivation for WCDMA


To provide the required capabilities on a basic technology, there are basically two options available, namely the wideband versions of TDMA and CDMA (WTDMA and WCDMA). It is interesting to observe that CDMA becomes a much better proposition and a much more viable alternative as a technology choice when the bandwidth is increased, i.e. WCDMA is a better technology than narrowband CDMA (NCDMA), see Figure 1. This means WCDMA offer both higher user data rates and improved coverage and capacity than does NCDMA.
Bit-rate (log scale)

WCDMA

Improved bit rate due to Higher BW and better capacity Improved Coverage due to frequency diversity and coherent uplink (> 3dB)

NCDMA

Coverage (log scale)

Unfortunately, the same is not true of TDMA. Coverage and capacity is not improved with WTDMA compared Figure 1: WCDMA vs. NCDMA with NTDMA, see Figure 2. This is due to limitations of Bit-rate equalizers and peak power restrictions. Moreover, it is (log scale) not a new finding; it was discovered and debated already W TDMA cannot provide full coverage due to in 1987 in the "big" European discussion over the peak-power and relative merits of WTDMA vs. NTDMA; finally, equalizer limitations W TDMA NTDMA was selected for GSM. When comparing the merits of different technologies for UMTS, the following issues are very important to consider: SERVICE FLEXIBILITY The ability to handle different services
Figure 2: WTDMA vs. NTDMA COST OF INFRASTRUCTURE The number of base station sites required to provide the service and the cost of these sites
N-CDM A

NTDMA cannot provide very high bit-rates

NTDMA

Coverage (log scale)

03/20/13

CAPACITY The number of users that can be handled by the system SPECTRUM FLEXIBILITY The efficiency of the spectrum planning process and the minimum spectrum needed for an operator to be able to provide the service in the first place COST OF TERMINALS The cost and size of terminal equipment for difference services, including dual-band operation TECHNOLOGY MATURITY AND RISK The time-to-market and risks involved GLOBAL STANDARD The opportunity to obtain an internationally-accepted standard in all parts of the world where the IMT-2000 frequency band will be available We believe that the our proposal WCDMA proposal fulfills these requirements, and is therefore the best radio access choice for UMTS. This will be further elaborated in this paper.

3. WCDMA System description


The WCDMA technology currently being developed in ETSI/SMG2 Concept group Alpha is the result of a ten year long research effort in Europe as well as worldwide. Ericsson has been active in research on WCDMA in European research programs such as Race 1 and Race 2/CODIT, and has also developed a Wide Band Testbed, which demonstrates multimedia services over a WCDMA radio interface. In Concept group Alpha, a large number of participating organisations have contributed with inputs to the concept development discussions. A merging process, taking all these contributions into account was finalised, September 15-16, where all participants agreed on one common WCDMA concept [].
P f = Codes with different spreading, giving 8-500 kbps

....

4.4-5 MHz

High rate multicode user Variable rate users 10 ms frame

Figure 3: Wideband CDMA, ETSI Concept alpha

Table 1: Basic parameters of ETSI/SMG2 Concept group Alpha, WCDMA

W CDM A ETSI Concept Multiple Access scheme Duplex schem e Chip rate Carrier spacing Frame length Inter-BS synchronization Coherent detection Multi-rate/variable-rate Service m ultiplexing Packet access

/ ARIB Core-A

W ideband CDMA FDD / TDD 4.096 Mcps ( expandable to 8.192 M cps and 16.384 M cps ) 4.4-5.2 MHz 10 ms No accurate synchronization needed UL: DL: Pilot bits Pilot bits Multi-code + O rthog. Variable SF Code and/or tim e m ultiplexed Dual-mode, comm on/dedicated channel Mobile Assisted, Supported with slotted mode

03/20/13

Interfrequency handover

In parallel to the ETSI work, a merge has also been done with the ARIB Core A, the WCDMA concept being standardised in Japan. Today the ARIB Core A WCDMA concept is in all essential parameters identical to the ETSI Alpha WCDMA concept. The plan is to have a complete merge of the WCDMA specifications between ETSI and ARIB. A merge is made possible by organisations that are active in both ETSI and ARIB, e.g. Ericsson, Nokia, Panasonic, NEC, Fujitsu etc. The basic parameters of WCDMA, defined by ETSI/SMG2 Concept group Alpha are given in Table 1. Figure 3 shows a WCDMA system with multiple services multiplexed on one carrier. Further, in Figure 4 the Physical Layer multiplexing is described in detail. Important to notice in Figure 4 is that uplink and downlink uses different multiplexing of DPDCH and DPCCH (Dedicated Physical Data Channel and Dedicated Physical Control Channel) in order to minimise mobile complexity and EMC in the mobile station. Pilot bits are included in both downlink and uplink in order to facilitate use of adaptive antennas and allow coherent detection. Also fast power control is supported on both up- and down-links, and by introducing explicit rate information fast variable rate services are supported efficiently. For more information regarding the details of the WCDMA concept see the Wideband Direct-Sequence CDMA Evaluation Document [].

4. WCDMA Key issues


4.1 Service and Operator Flexibility
One of the most important characteristics of WCDMA is the fact that power is the common shared resource, see also Figure 5. This makes WCDMA very flexible in handling mixed services and services with variable bit-rate demands. Radio resource management is done by allocating power to each user (call), and to ensure that the maximum interference is not exceeded. No allocation of codes, time-slots etc. is needed as the bit-rate demand changes, which means that the physical channel allocation remains unchanged even if the bit-rate changes. Further, WCDMA requires no frequency planning, since one cell re-use is applied This flexibility is supported in WCDMA, since it uses Orthogonal Variable Spreading Factor codes, OVSF, for channelisation of different users. These codes have the characteristic to maintain orthogonality between users (or different services allocated to one user) even if they operate at different bit-rates. One physical resource can thus carry multiple services with variable bit-rates. As the bit-rate demand changes, the power allocated to this physical resource is adjusted so that quality-of-service is guaranteed at any instant of the connection.
Translates into

Varying User Bit Rate

Varying power level Varying spreading factor

But, same physical resource


Bit Rate Translates into Power Level Spreading Factor

Downlink - Time multiplexed control and data


DPCCH DPDCH

Figure 5: WCDMA, Bit-rate flexibility Service 2 Service 1

Pilot

PC

RI

Data

I&Q
Service 3

Uplink - I/Q Code multiplexed control and data


DPDCH
Service 1

Data
Service 2 Service 3

I
RI

DPCCH

Pilot

PC

Pilot

4.1.1 Mixed services and service multiplexing

Pilot bits Enables coherent detection Facilitates adaptive antennas

PC

Fast power control 1600 / second

RI

Rate information Fast variable rate High flexibility

WCDMA is designed for wideband multimedia services and supports bit-rates of at least 384 kbit/s with good coverage and full mobility. Up Figure 4: WCDMA, Physical Layer multiplexing to 2 Mb/s is supported with one 5 MHz carrier with local coverage. For a mixed service connection, each service is optimised to its own specific requirements (BER, maximum delay, throughput etc.). After channel coding and interleaving, the different services are multiplexed together onto one physical channel, cf. Figure 6.

03/20/13

A typical scenario for a fully utilized WCDMA system includes a mix of simultaneous high speed packet data users and low rate voice user connections, cf. Figure 7. This is done maintaining high capacity and coverage for every service. Since power is the common resource, multiplexing of services with very different characteristics can be achieved in an efficient way, utilizing the full capacity of the WCDMA-carrier.

Bit Rate Service 1 Power Level

Po w er L eve ls fro mM S R e ceiv ed P ow er Lev els a tB T S

CA

CB
Bit Rate Service 2

CA
CC

CD

Figure 6: WCDMA support of multimedia services

Figure 7: WCDMA support of simultaneous voice and high speed packet data services

4.1.2 Capacity gains for variable rate, real time services or link improvements
In WCDMA, a direct gain in capacity from speaker inactivity can be seen as the transmit power is reduced in speech pauses. Reduced power on the DPDCH-physical channel in combination with interferer diversity, i.e. many simultaneous speech users on the same WCDMA carrier, means that the number of users can be increased with close to a factor of two. This is done without any extra signaling between mobile and base station. Further, any link improvements due to e.g. adaptive antenna arrays, gives immediate improvements of capacity/coverage/quality for all the users in the radio network. This is a very important feature of the WCDMA technology, and comes from the fact that WCDMA operates with a re-use of one and has only one shared resource (power).

4.2 WCDMA is optimised for packet services


With the WCDMA dual-mode packet-access scheme, packet transfer can take place on both common and dedicated channels, see Figure 8. In this way, the packet access can be optimized for fast access/response as well as for maximum throughput. The mode of operation is adaptively chosen, based on estimated packet traffic characteristics and does not require any explicit user interaction. With common-channel packet transmission an uplink packet is appended directly to a Random Access burst. Common-channel packet transmission is typically used for short infrequent packets where fast access is most important and where the link maintenance needed for a dedicated channel should be avoided. With dedicated-channel packet transmission, an initial Random Access request is used to set up a dedicated channel for the packet transmission. On this dedicated channel, closed-loop power control is in operation and a sequence of packet transfers can take place without any intermediate random-access requests. In this mode of operation, packets will either be transmitted in a scheduled or non-scheduled way, depending on the amount of capacity needed for the packet transmission.

03/20/13

Random-access burst

User packet

Arbitrary time

Random-access burst

User packet

Common channel

a) Common-channel packet transmission


Scheduled packets Non-scheduled packet Access request User packet : Link maintenance User packet Access request User packet

Dedicated channel

b) Dedicated-channel packet transmission


Figure 8: WCDMA dual-mode packet-access scheme

4.3 Spectrum allocation and guard bands


The spectrum allocation of WCDMA is done using a 200 kHz carrier raster, enabling flexibility in handling interference between different allocations. The 3 dB bandwidth of the WCDMA carrier is 4.096 MHz, but normally a carrier spacing larger than this is used. For multiple carriers that are co-sited, the carrier spacing iscan be 4.4 MHz. This is based on the adjacent channel interference considerations for coverage and capacity. At this spacing, the worst case sensitivity loss is negligible, while the capacity loss is less than 1%.

Other operator

UMTS Operator

Scenario Co-ordinated Uncoordinated


5 MHz

Carrier spacing UMTS UMTS GSM 4.4 MHz 5.0 MHz 3.0 MHz

UMTS UMTS UMTS

Uncoordinated
UMTS Operator

Other GSM operator

3 MHz

For carriers that are not co-sited, larger spacing is needed. A model analysing the interference into a WCDMA receiver from an Figure 9: Spectrum allocation and guard bands for WCDMA. adjacent carrier has been done, assuming a certain Minimum Coupling Loss (MCL) between transmitter and receiver. Both MS-to-BTS and BTS-to-MS interference is considered. As a reference, the minimum coupling loss between two GSM systems has been estimated from the GSM 05.05 specifications of receiver sensitivity and spectrum masks. They imply that for GSM to GSM non-coordinated systems, a requirement of 86 dB MCL is necessary for a reasonable carrier spacing of 0.6 MHz ( 1.5 MHz spacing is required to achieve a better MCL). As a result, an MCL value of 86 dB is chosen as a reference. When the MCL analysis is applied to WCDMA, the 86 dB value is reached at a carrier spacing of 5 MHz between two non-coordinated UMTS systems. For non-coordinated cell layers for the same operator (micro and macro cells) the same 5 MHz spacing applies. Since the WCDMA carrier raster is the same as for GSM (200 kHz), it is possible to locate a WCDMA close to any GSM spectrum allocation. This is important in the case of re-farming GSM spectrum for UMTS. MCL analysis shows (assuming 86 dB MCL) that the spacing between a WCDMA carrier and the closest allocated GSM carrier needs to be only 3 MHz. This facilitates an efficient re-farming of GSM spectrum. 5.6 MHz of the spectrum needs to be cleared for the first WCDMA carrier.

03/20/13

4.4 Coverage (Infrastructure cost)


4.4.1 Link budget comparison WCDMA vs GSM
The coverage of WCDMA is very competitive and also makes it possible to re-use GSM 1800 cell sites when migrating from GSM to UMTS services. Table 2 shows uplink budgets for WCDMA with GSM speech service as a reference. The uplink is the coverage limiting link. Assumptions for the comparison are that the average mobile output power is equal in WCDMA and GSM. The WCDMA Eb/N0 values are from the ETSI Evaluation Report, while GSM sensitivity is based on a GSM implementation. More assumptions and comments for the coverage analysis can be found in Section . A WCDMA speech service will thus tolerate a few dB higher path loss than a GSM speech service. This means that WCDMA gives better speech coverage than GSM, re-using the same cell sites when being deployed in the same or a nearby frequency band (e.g. GSM1800 vs. the UMTS band). Table 2 also shows that a 144 kbit/s circuit switched data service can operate with at least the same coverage as a GSM speech service, thereby reusing the GSM cell sites.
Table 2: Uplink budgets for GSM and WCDMA for speech, circuit switched data and packet switched data (ETSI environment Pedestrian A, 3 km/h).

Link budgets Average TX power per traffic ch. TX antenna gain Body loss RX antenna gain Cable and connector losses Required Eb/No RX sensitivity Interference margin Handover gain Log-normal fade margin Maximum path loss dBm dBi dB dBi dB dB dBm dB dB dB dB

GSM Speech 13 kbit/s (Peak) 30 0 3 10 2 3,5 -111,2 2,0 4,7 11,3 137,6

WCDMA WCDMA Speech Circuit 13 kbit/s 144 kbit/s 21 24 0 0 3 0 10 10 2 2 4,1 1,3 -123,8 -116,1 4,0 4,0 6,0 6,0 11,3 11,3 140,5 138,8

The conclusion is that WCDMA voice and 144 kb/s circuit switched data coverage will be somewhat better than GSM. A 384 kb/s data service can also reuse GSM sites, using a low gain mobile antenna and an improved base station antenna.

4.5 Capacity
In this section, WCDMA is compared to NCDMA and conclusions are drawn on capacity improvements, for more information on WCDMA capacity see Section . Narrowband CDMA, i.e. IS-95, and GSM are roughly equal in terms spectrum efficiency for speech services []. For a 13 kbit/s speech service the capacity will roughly be 3.5-4 times analog AMPS.
Wideband spreading (Uplink and Downlink) Frequency Diversity Statistical Averaging Gain 2-3 dB ~1 dB

1.2288 Mcps

4.096 Mcps

Coherent Uplink
Data

Gain

Coherent Demodulation 2-3 dB Wideband CDMA on the other hand has a significant DPDCH gain from the wideband carrier that reduces the fading Total Gains of the radio signal, see Figure 10. The wider carrier DPCCH Uplink 4-6 dB also leads to a larger number of users on each carrier Downlink 3-4 dB 1600 Power Control which give improved statistical averaging of Commans / Pilot symbols second (Also downlink) interference from other users. Further, by introducing give coherent demodulation coherent demodulation in the uplink capacity and coverage gains are achieved. The coverage gain is probably most important since the uplink usually is limiting the link budget. The capacity gain for the Figure 10: Capacity gains in WCDMA compared to
Pilo t PC RI

Narrowband CDMA

03/20/13

uplink, from coherent demodulation can be up to a factor of two. Also, fast power control on the down-link will give improved performance at low Doppler speeds, a feature which is not used in NCDMA. In total for a speech service, the cell capacity of WCDMA is expected to be at least two times (3 dB) that of NCDMA. Also for high bit-rate packet services, WCDMA will give very good capacity figures. In the ETSI WCDMA evaluation report the simulations of a 384 kbit/s packet data service show that for a low mobility radio channel, a total bit-rate of 1.9 Mb/s is available on the downlink of each WCDMA carrier []. For a typical WWW-browsing application this would mean that 130 simultaneous active users can be supported with one WCDMA-carrier 1 . Each user having 384 kbit/s packet access with immediate response. In the above mentioned capacity figures for WCDMA, improvements from techniques like adaptive antenna arrays, multi-user detection or down-link antenna diversity has not been taken into account. WCDMA has built in support for such techniques, which makes it a future proof technology. Adaptive antenna arrays can be introduced efficiently, since the down-link physical channel carries dedicated pilot symbols. Spreading with short codes, makes multi-user detection feasible and with the OVSF channelisation codes, orthogonal transmit diversity is possible to introduce in the down-link. All these link improvements will give immediate improvements of capacity/coverage/quality for all the users in the radio network, when introduced. This is a very important feature of the WCDMA technology, and comes from the fact that WCDMA operates with a reuse of one and has only one shared resource (power).

4.6 Soft Handover


4.6.1 Soft handover or Macro diversity
Soft handover or macro diversity provides a true seamless handover between different base stations. There are two types of soft handover to be considered; softer handover (the mobile station communicates simultaneously with co-located base stations) and soft handover (the mobile station communicates simultaneously with noncollocated sectors, i.e. different BTS). Soft handover introduces some overhead in terms of transmission links, since a call requires more than one link on the average. The number of required transmission links is directly proportional to the number of mobiles engaged in soft handover. The actual number depends on solutions for handling handover. Different solutions may be used for different types of traffic, circuit or packet. For example, soft handover may not be required for packet services. The WCDMA concept includes several control mechanisms to reduce the number of users in soft/softer handover without any decrease in cell capacity. The main ones are; Fast power control (closed loop) in both up- and downlink. This feature reduces the need to connect to more base stations because the fading is reducedcombat by the fast power control, meaning that the active set size can be reduced without any substantial loss of diversity. Hence, the cell capacity is maintained while decreasing the transmission link capacity requirement. Relative comparison between all candidate base stations. The soft/softer handover algorithm prevents the MS from connecting to unnecessary base stations, i.e. base station that not contribute to the macro diversity gain. In general, one can say that the WCDMA concept is restrictive in the to use of soft/softer handover. When comparing the WCDMA and IS-95-A soft handover algorithms it shall be noted that IS-95-A uses absolute thresholds (load dependent) and WCDMA uses relative thresholds (which means less MSs in soft/softer handover). As seen in Figure 11 below the IS-95-A pilot strength (i.e. pilot C/I) used in the handover algorithm has a direct impact on the fraction of users engaged in soft handover. However, this also means that a load dependency is built into the IS-95 system, which can be hard to control in some cases. Soft handover normally reduces the interference and thereby increases the capacity, but the IS-95 handover algorithm reduces the number of MS in soft/softer handover at increased system load. Note that the power (coverage) of WCDMA synchronisation channel (SCH) has no effect on the WCDMA soft handover zone (shaded in the figure) since this is fully controlled by the handover algorithm threshold. The SCH controls the maximum cell size only.
1

The traffic model assumed is a user that actively looks for information in the world wide web. The average page size is 40 kbytes. Approximately every 6:th page is read more carefully, giving av average time per page of 22 seconds.

03/20/13

Simulation and field measurements of the IS-95 system often state that more than 50% (handover threshold, load and MS speed dependent) of the MSs are engaged in soft/softer handover. System simulations of the WCDMA algorithm show that only 30-40% (handover threshold dependent) of the MSs will be in soft/softer handover.

IS-95-A, Absolute thresholds


Pilot strength Cell load

WCDMA, Relative thresholds


SCH coverage

Absolute Threshold

Relative Threshold

Figure 11. A comparison between IS-95-A and WCDMA soft handover algorithm The WCDMA technology facilitates a very good support for utilising variable rate voice coding and packet services. This together with the use of efficient transport technologies such as ATM/AAL2 will save transmission by at least 50%. This means that the extra transmission introduced by macro diversity is well compensated by the variable rate coding and ATM transport gain.

4.6.2 Cell-breathing
The uplink coverage of a CDMA system will have a link budget dependence on the system load, so-called cellbreathing. However, the admission and congestion control functions in the WCDMA concept will automatically control the system load, i.e. prevent that coverage holes are created due to an overload situation. The actual setting of the maximum load threshold, used by the admission and congestion control, can be assisted by means of a radio network planning tool. Also a C/I based handover algorithm will naturally hand over mobile stations at the cell border to potentially less loaded cells, i.e. load sharing. This means that the WCDMA system will provide very good coverage in rural areas (low traffic density) as well as high capacity in urban areas (high traffic density), due to the relative small site-to-site distance used in urban areas. In a system with dominating data traffic, this works to the advantage of having asymmetric traffic. If more downlink than uplink traffic exists, the uplink coverage will improve compared to the downlink. This is essential, since the main limitation for coverage of data services is the mobile power and hence the uplink. By trading the uplink system load for coverage, a very competitive coverage of high rate data services can be achieved. The WCDMA concept also supports the possibility of multi-user detection (MUD) in the uplink. Applying MUD in the uplink improves the coverage in loaded system, since the interference in the base station is cancelled.

4.7 Interfrequency handover, HCS


The support of seamless interfrequency handover is a key feature of WCDMA that distinguishes it from current second generation Narrowband CDMA. Interfrequency handover is a necessary feature for the support of Hierarchical Cell Structures, HCS, with overlapping micro and macro cells operating on different carrier frequencies, see Figure 12 below. With the introduction of HCS, a cellular system can provide a very high system capacity through the micro cell layer, at the same time offering full coverage and support of high mobility by the macro layer. Interfrequency handover is then needed for a handover between the different cell layers. A second application where interfrequency handover is necessary is the hot-spot scenario, where a certain cell that serves a high traffic area, uses additional carriers compared to the neighbouring cells, see Figure 13.

03/20/13

Figure 12: Hierarchical Cell Structures

To support seamless interfrequency handover, it is vital that a mobile terminal can carry out measurements on other carriers than the one currently being used for communication. This feature is supported through the downlink slotted mode. In slotted mode, the downlink transmission is compressed in time, thereby creating a short idle slot, during which the mobile station can carry out interfrequency measurements without the need for the extra

f1 f1 f1 f1,f 2 f1 f1 f1

Idle period available for interfrequency measurements

Instantaneous Rate/Power

Figure 13: Hot spot scenario

Tf Normal transmission Slotted transmission

Figure 14: WCDMA, slotted downlink transmission for interfrequency measurements

complexity associated with a second receiver dedicated for interfrequency measurements, see Figure 14. Note that the slotted mode does not cause any loss of data, since the information bit-rate, and hence the power, is increased during transmission of the compressed slots.

4.8 Support of a TDD mode


With the support for TDD (Time-Division Duplex) operation, in addition to the normal cellular FDD operation, WCDMA allows for optimal utilization of the available UMTS spectrum. The excess part of the lower UMTS band can be used for TDD operation Due to the asymmetry of many UMTS services, with more downlink than uplink traffic, additional uplink spectrum can be released and be used with the TDD mode.

The WCDMA TDD mode has been designed in harmonization with the FDD mode. As a consequence, the cost and size of a dual-mode FDD/TDD terminal will be only marginally higher than a single-mode FDD terminal. The TDD mode will reuse the same frame and slot structure as the FDD mode where each slot can be individually allocated to either uplink or downlink according to the operator requirements, see Figure 15. Seamless handover from TDD to FDD and vice verse will be supported based on the same cell-search scheme used within the FDD mode.


Slot allocation for high mobility

Slot allocation for large coverage

Slot allocation for assymetric traffic

Figure 15: WCDMA-TDD, slot allocation examples

03/20/13

10

5. TD-CDMA, short description


5.1 Introduction
In the work in ETSI/SMG2 on finding the best candidate for UMTS / IMT2000, the concept group delta, TD-CDMA, is promoted by Siemens as being backwards compatible to GSM. For more details on the concept see []. The main similarity to GSM is that the length of time slots and number of time slots per frame in TD-CDMA is equal to that in GSM, see Figure 16. However, the bandwidth is different fromto GSM.
f = 1 tim e slot, 1 code , ~16 kbps user data
M ult slot / M ulti code user

1.6 M H z M ulti code user Single slot / Single code user 577 s M ulti slot user 4,62 m s

Furthermore, many fundamental parts of the system definition of TD-CDMA are still open, leading to great difficulties in doing performance evaluation of Figure 16: TD-CDMA, ETSI concept delta the system concept. One example of this difficulty is the calculation of minimum spectrum requirements in [] where the link-level simulations were performed assuming perfect frequency hopping giving a certain re-use, and the number of carriers needed to deploy the system were calculated assuming that no frequency hopping is applied.

The most important areas where information on the system definition is missing for TD-CDMA are the following: Logical and physical channel structures Common control channels (BCCH and RACH) and theare unclear m. Mapping of logical channels onto physical channels is not described in detailare unclear. Physical channel structure The multiframe structure is not defined. Service mapping on the physical channel is incomplete. Resource allocation and variable data rates A procedure for capacity requests and allocation for variable rate is described. However, several open issues remain with this scheme, e.g. regarding link adaptationmapping on physical layer. The principles for handling the multi-dimensional resource allocation in the code, time, power and modulation domains are not described. Planning and link adaptation of the 2 Mb/s service will be a problem due to the assumption of 16 QAM modulation, which leads to higher C/I requirements. Link adaptation procedures have not been described. Radio resource management The frequency planning is supposed to be done automatically through slow Dynamic Channel Allocation. This needs to be described in more detail, since it is not clear how the very few 1.6 MHz carriers available can be allocated dynamicallyespecially in case of 200 kHz BCCH carriers, where these need to be planned together with 1.6 MHz carriers.

5.2 Resource allocation complexity


The radio resource allocation in the TD-CDMA concept requires effective handling of three dimensions (frequency, time and code). This is likely to cause severe constraints in terms of how efficiently the available radio resource can be utilised. In addition, multiple modulation- and coding schemes (link adaptation), frequency hopping and time slot hopping lead to an overall very complex scenario when it comes to the allocation of a single radio resource. UMTS will have to operate with a large number of service types, with different needs of resource configurations, as well as in scenarios with mixed services.
Resource allocation is in three dimensions with restrictions in both time and frequency!
144 kbps How are resources for the new 144 kbps user allocated? code f t

Figure 17: TD-CDMA resource allocation example

03/20/13

11

Also, it can be foreseen that several mobile classes will be defined for TD-CDMA, with constraints in number of time slots or codes used, frequency/time hopping restrictions and link adaptation possibilities. A large number of mobile classes may be required due to a large impact on MS -complexity and power consumption as a function of e.g. how many time-slots are received. In Figure 17 an example is given where a 144 kbit/s user, restricted to use only one time slot, is not able to fit into a TD-CDMA carrier even though sufficient capacity is unused at the time. Finally, when using link adaptation (change in modulation and/or coding) each service will change their C/I requirement as well as resource configuration continuously when adapting the link. Hence, the shape of the already occupied resources in Figure 17 will then change in time, leading to an even further complex scenario of the radio resource allocation for TD-CDMA.

5.3 Packet support


The support of packet services requires efficient random access in the system in order to not introduce delays in the packet transmission. Being one of the key UMTS requirements, it is essential to get more information on TD-CDMA packet support and its performance. No RACH is defined. Both logical and physical channel descriptions are needed. In packet mode (Non Real Time operating mode) messages like Capacity Allocation (CA), Forward Order (FO) etc. are defined. No information is given, on which logical channels these commands are sent. Further, it is important for performance, how the logical channels are mapped on the physical layer.

6. Comparison WCDMA vs TD-CDMA


6.1 Minimum spectrum requirements
For WCDMA it was stated earlier that the requirement on guard-bands, and consequently the carrier spacing, depends on the allocation scenario, see Section . For coordinated WCDMA - WCDMA the carrier spacing will be 4.4 MHz, e.g. two co-sited WCDMA carriers. For uncoordinated operation, e.g. two different operators or two different cell layers, the requirement is 5 MHz. Finally, for un-coordinated operation between WCDMA and GSM the required carrier spacing is 3 MHz. The minimum spectrum allocation needed is 5 MHz, although for commercial operation an allocation of 10-15 MHz is desirable, to support e.g. HCS. In the analysis of the TD-CDMA concept in the -group, ideal frequency hopping (FH) has been assumed in the link simulations. In this case a 1/3 re-use, with a fractional load below 100%, may be feasible. Looking at the spectrum requirements however shows that this assumption is not realistic. For performance close to ideal FH at least 3 carriers are needed for hopping. With 3 carriers/sector and a 3-sector re-use, a total of 9 carriers is needed, i.e. 14.4 MHz (9*1.6). Furthermore, according to the -group the carrier spacing needed between operators is 3.2 MHz, i.e. one carrier guard. In total the minimum spectrum required will be 16 MHz, for only one cell layer. The requirement for BCCH allocation is unclearnot included and should be added. A minimum allocation of 16 MHz is clearly not realistic. The consequence of that is that FH can not be used in practice for the TD-CDMA concept. When FH is not used the performance will degrade due to reduced frequency diversity and interference averaging. In practice this means that at least a 6 sector re-use must be deployed. With one carrier/sector the minimum requirement will then be 11.2 MHz, including guard-band (6*1.6+1.6). See Figure 18.
1.6 M H z

Guard

11.2 M H z

Figure 18: Minimum spectrum allocation for one cell-layer, TDCDMA

03/20/13

Guard

12

Two interesting comparisons between WCDMA and TD-CDMA are shown in Figure 19 and Figure 20. In the first example the minimum spectrum requirement for an operator using two cell layers, one macro- and one micro layer. WCDMA requires in this case 10 MHz spectrum, while TD-CDMA requires 22.4 MHz, i.e. more than twice the spectrum is needed. Note that for the TD-CDMA/FH case the total spectrum requirement would be 32 MHz! In the second example it is shown how much spectrum must be cleared when re-farming GSM1800. Originally the operator has 15 MHz, or 74 GSM carriers (including guard). In the WCDMA case the 28 carriers havehas to be cleared (5.6 MHz), in accordance with the 3 MHz carrier spacing requirement between WCDMA and GSM. Note that WCDMA is expected to provide at least twice the capacity to GSM. For TD-CDMA 55 GSM carriers need to be cleared (11.0 MHz), according to a 1.7 MHz carrier spacing requirement between TDCDMA and GSM stated in the -group, and a 6-sector re-use.

WCDMA

6.2 Coverage and, link budget comparisonlink budget


5.6 MHz 9.4 MHz, 46 carriers

28 carriers are cleared

WCDMA
Other operator Macro layer Micro layer Other operator

TD-CDMA
The two spectrum allocations give approximately the same capacity.
55 carriers are cleared

5 MHz

5 MHz

5 MHz

10 MHz

11.0 MHz + BCCH carriers? Micro layer

4.0 MHz, 19 carriers

+ BCCH carriers?

TD-CDMA
Macro layer

Figure 20: Spectrum re-farming example, WCDMA vs TDCDMA


1.6 MHz
f

comparison

The coverage of a cellular system is calculated through the link budget, which is affected by a number of Figure 19: Spectrum requirements example, WCDMA vs parameters. Most of these parameters are independent TD-CDMA of the radio access technique, such as body loss and antenna gain. They are therefore not relevant for a comparison. Parameters that may differ between different concepts are mobile output power (including peak power constraints), Eb/N0 performance, Handover gain and Interference margin. Average mobile output power is here assumed to be equal in WCDMA and TD-CDMA, and also for GSM. There are thus no constraints on the peak power used because of e.g. the TDMA duty cycle. The Eb/N0 values are from the respective ETSI Evaluation Reports. In addition to this, TD-CDMA has a loss added for not using frequency hopping, since frequency hopping is optional and can not be applied unless very large frequency allocations are used. The ETSI Evaluation Report values are with ideal frequency hopping. In the delta evaluation report, it is stated that up to 5 dB loss is expected. A conservative value of 1.5 dB is used here and is simply added to the Eb/N0 values from the delta evaluation report. Handover gain has so far in the ETSI evaluation been estimated by an analytical two-cell model, giving the multi-server gain from being able to connect to two instead of one base station. Hard handover is modeled with a 3 dB hysteresis giving a 4.7 dB handover gain and soft handover is modeled as an ideal handover (0 dB hysteresis) giving a 5.0 dB handover gain. The soft handover advantage of 0.3 dB is however underestimated, since it does not take into account the macro diversity combining and the gain from avoiding handover delays. The macro diversity combining gain is here estimated to be 1 dB in the pedestrian environment, which is added to the soft handover gain, giving a 6.0 dB value. The interference margins of the two systems are estimated at a 60% system load. All values are for the pedestrian environment (Outdoor to indoor A) at 3 km/h mobile speed.

22.4 MHz

03/20/13

13

The speech service investigated in Table 3 is a 13 kbit/s speech service with 20 ms interleaving. The Eb/N0 values are however from 8 kbit/s simulations for the UMTS proposals, with the respective mobile sensitivity values re-calculated for the higher data rate.
Table 3: Coverage analysis for a 13 kbit/s speech service

Link budgets, uplink Average TX power per traffic ch. TX antenna gain Body loss RX antenna gain Cable and connector losses Required Eb/No RX sensitivity Interference margin Handover gain Log-normal fade margin Maximum path loss

dBm dBi dB dBi dB dB dBm dB dB dB dB

(Peak)

GSM 30 0 3 10 2 3,5 -111,2 2,0 4,7 11,3 137,6

WCDMA TD-CDMA 21 21 0 0 3 3 10 10 2 2 4,1 9,8 -123,8 -118,1 4,0 2,0 6,0 4,7 11,3 11,3 140,5 135,5

It should be noted that some GSM numbers are not comparable to the UMTS system values. The GSM Mobile power and Receiver sensitivity are peak power values, while WCDMA and TD-CDMA have average powers. GSM Eb/N0 is per modulated bit, while the others have Eb per information bit. The bottom line maximum path loss values are however comparable. It can be seen that WCDMA has higher maximum path loss value than GSM and 5 dB better than TD-CDMA. This implies a coverage advantage for WCDMA. Table 4 shows a coverage comparison between UMTS data services (144 kbit/s circuit switched) and a GSM speech service. The assumptions are the same as in the speech comparison, with a few exceptions. The UMTS data terminals have 3 dB higher output power than the GSM voice terminal. Also, the body loss of 3 dB for speech services has been removed, assuming a terminal on e.g. a table. The bottom line for the data service is that WCDMA can have the same coverage for a data service as for GSM speech, while it is 5 dB better than TD-CDMA. A 5 dB difference in link budget corresponds to 93% more base stations for TD-CDMA, assuming 35 log(d) propagation.
Table 4: Coverage analysis for 144 kbit/s circuit switched data

Link budgets, uplink Average TX power per traffic ch. TX antenna gain Body loss RX antenna gain Cable and connector losses Required Eb/No RX sensitivity Interference margin Handover gain Log-normal fade margin Maximum path loss dBm dBi dB dBi dB dB dBm dB dB dB dB

GSM Speech (Peak) 30 0 3 10 2 3,5 -111,2 2,0 4,7 11,3 137,6

WCDMA TD-CDMA LCD144 LCD144 24 24 0 0 0 0 10 10 2 2 1,3 7,0 -116,1 -110,4 4,0 2,0 6,0 4,7 11,3 11,3 138,8 133,8

6.3 Capacity
In Table 5 below, a comparison of cell capacity for a 13 kbit/s speech service has been made between WCDMA and TD-CDMA. For reference, also analog AMPS, GSM and IS-95 have been included in the table, see also []. It can be seen that here that WCDMA will give approximately double the cell capacity of TD-CDMA. For a 13 kbit/s speech coder, the capacity of WCDMA is 8 times analog AMPS.

03/20/13

14

In the calculation of number of carriers that can fit into a 15 MHz total spectrum allocation, a reduction due to guard bands to adjacent systems have been taken into account (except for AMPS). The number of channels per carrier for WCDMA is eight times the value for IS-95 (four times carrier bandwidth and two times spectrum efficiency). Furthermore, it can also be derived from uplink link-level and system-level simulations of an 8 kbit/s speech service, see []. The channel propagation model used, Vehicular A is described in ETSI document 04.02. The ETSI Alpha group simulations show a capacity of 80-150104 speech users per carrier for WCDMA depending on environment. When converted from 8 kbps to 13 kbps speech coding, this is of the same order as the 64 users/carrier value in Table 5.. To arrive at the capacity figures for a 13 kb/s speech coder this figure has been scaled, assuming identical Eb/No requirements. The downlink capacity of WCDMA for the same service/propagation characteristics shows a similar figure. For TD-CDMA, the number of channels per carrier is based on the assumption that 8 codes per time slot is used and that 100% load is possible at the given frequency re-use. The capacity number is then slightly below that in the ETSI delta group evaluation [3], since Frequency Hopping is here assumed to be unrealistic.
Table 5: Capacity comparison for a 13 kbit/s speech services

Total operator bandwidth Number of carriers Number of channels per carrier Re-use pattern Carriers / Sector Speech channels / Sector Capacity gain [* AMPS]

WCDMA 15 MHz 3 64 1/1 3 192 8

TD-CDMA 15 MHz 8 64 6
(on average)

GSM 15 MHz 69 8 6
(on average)

IS-95 15 MHz 11 8 1/1 11 88 3.7

AMPS 15 MHz 500 1 7/21 24 24 1

1.33 85 3.5

11.5 92 3.8

The re-use of WCDMA is 1. For TD-CDMA, there are no results available showing that 1/3 re-use operation is possible with 15 MHz spectrum allocation. aA re-use >1/3 is required sinceand thus frequency hopping is not possible due to lack of carriers. This would lead to a minimum re-use similar to an optimised GSM-system, i.e. re-use 6 on average. It is not clear how a re-use of 6 can be deployed for a TD-CDMA system, but this fact has not been taken into account in the comparison. The conclusion that TD-CDMA without frequency hopping operates with a similar re-use as an optimised GSM-system is based on the following reasoning: TD-CDMA has the same transmission efficiency as GSM, see Figure 21 (same bandwidth per user, same downspread C/I). Other factors that affect re-use are orthogonality within cell, frequency diversity, interference averaging, time dispersion diversity and power control. There is no advantage in favour of TDCDMA for any of these factors; GSM is orthogonal within the cell and TD-CDMA has a similar characteristic with the JD-receiver, frequency diversity is not as good as GSM for a non-hopping TD-CDMA system, interference averaging gains come from frequency hopping and spreading codes respectively, equalizers handle time dispersion in both cases and power control is similar in the two systems. If a more conservative GSM system is analysed, i.e. with a 9or 12-sector re-use, the capacity numbers for GSM would be

GSM
Carrier BW #slots / carrier #users / slot BW / user Spreading factor Interference level Down-spread C/I 200 kHz 8 1 25 kHz 1x I0 C/I0

TD-CDMA
1.6 MHz 8 8 25 kHz 8x 8*I0 C/I0
Code

t
f

GSM
Carrier BW #slots / carrier #users / slot BW / user Spreading factor Interference level Down-spread C/I 200 kHz 8 1 25 kHz 1x I0 C/I0

TD-CDMA
1.6 MHz 8 8 25 kHz 8x 8*I0 C/I0
Code

t
f

03/20/13

15
Figure 21: GSM vs TD-CDMA transmission efficiency

reduced accordingly. This does however not influence the conclusion that WCDMA has twice the capacity compared to TD-CDMA.

6.4 Support of 2 Mb/s


The support for high data rates is a key service requirement of UMTS. These services are typically packet services where short transmission bursts are separated by relatively long idle periods. Consequently, high capacity is not the key issue for these services. Instead the key issue is to support a high maximum rate in order to reduce packet delay and thus increase customer satisfaction. With WCDMA it has been shown that 2 Mb/s user rate can be provided with 5 MHz total spectrum allocation, see []. For TD-CDMA, the use of higher order modulation is proposed. Higher-order modulation is significantly more sensitive to interference, i.e. a significantly larger re-use is needed if full load shall be possible. The conclusion is that the TD-CDMA proposal requires more than twice the spectrum allocation (>2*11.2 MHz) compared to WCDMA, to support a 2 Mb/s service in both uplink and downlink. For WCDMA a total spectrum allocation of 2*5 MHz is sufficient for the support of 2 Mb/s services. The reason for the large difference in minimum spectrum requirements is again related to re-use. For the transmission of 2 Mb/s user rate, a certain bandwidth is needed. For WCDMA, the entire system bandwidth (in this case 5 MHz) is available in each cell. This bandwidth is sufficient for the transmission of 2 Mb/s user rate, although not in all cells simultaneously. For TD-CDMA, which uses a fixed reuse, only a small portion of the total system bandwidth is available in each cell. Consequently, a significantly larger system bandwidth is needed for TD-CDMA to support 2 Mb/s user rates, even at very low load.

6.5 Terminal aspects


The most important factor for terminal cost and size is the maturity of the technology, i.e. how far one has come on the learning curve. Research on WCDMA has been carried out for many years in Europe. A number of test-beds have been built in both Europe and Japan (Ericsson has a WCDMA test bed running for more than two years). Although WCDMA is not a simple technology, it does not rely on any UN-proven high risk y technologies, such aslike Joint Detection (JD) in TD-CDMA. In the evaluation so far, one striking example is the baseband processing requirements, shown in Figure 22. The Joint Detection algorithm required by TD-CDMA is not only un-proven in large scale field trials, it is also extremely complex compared to the straight forward RAKE receiver in WCDMA (approximately 10 times more complex). Figure 22, demodulator complexity of an 8 kbit/s speech service. Parameters used in the computations:

Million real multiplications/second


80 70 60 50 40 30 20 10 0

WCDMA:
Single code, 256 spreading factor 24 kbit/s gross rate R=1/3 convolutional code. (code (From group Evaluation Report)
Type 1

TD-CDMA:
Single slot/code, burst type 1 22.4 kbit/s gross rate R=1/3 convolutional code 80% are floating point multiplications. (From group Evaluation Report)

Type 2

W C D MA WCDMA

W -TD MA/S TD-CDMA

Another terminal aspect is the implementation of service UMTS/GSM dual mode terminals. Firstly, for both WCDMA and TD-CDMA it can be stated that: The baseband processing is completely different to GSM (RAKE / JD vs. Viterbi Equalizer) Different filter bandwidths (5 MHz / 1.6 MHz vs. 0.2 MHz)

Figure 22: Demodulator complexity, 8 kbit/s speech

03/20/13

16

Different carrier frequencies (2 GHz vs. 900 / 1800 / 1900 MHz) The conclusion is that there are no fundamental differences in additional hardware requirements. Secondly, for WCDMA the handover measurements for UMTS-GSM handovers is accomplished by using measurement slots. For TD-CDMA, one would use the time between TX and RX slots (Note however that for multi-slot allocation a separate receiver is needed even for normal handover measurements). Therefore, there are no major differences regarding handover measurements. The conclusion is that the extra complexity due to dual-mode implementation is similar in the two cases.

7. Conclusions
The motivation for WCDMA as the preferred UTRA concept has been given. The concept has been briefly described, and the key issues pointed out. In the comparison with the TD-CDMA concept, WCDMA is the clear winner. The comparison can be briefly summarised as: WCDMA requires less than half the spectrum WCDMA requires half the number of base stations WCDMA provides double capacity WCDMA has much more flexible resource allocation WCDMA has lower terminal complexity Finally, WCDMA is a more mature concept, giving less cost and faster time to market. Furthermore, WCDMA is the only candidate that can get global acceptance.

8. References
[1] Concept Group Alpha - Wideband Direct-Sequence CDMA, Evaluation Document, Draft 1.0, ETSI SMG2 Tdoc SMG2 270/97 [2] Concept Group Alpha - Wideband Direct-Sequence CDMA, Answers to Operators Key questions, Version 1.0, ETSI SMG2 Tdoc SMG2 268/97 [3] Concept Group Delta, Wideband TDMA/CDMA, Evaluation Report - Part 1, V 0.2, ETSI SMG2 TDoc SMG2 UMTS D 9 / 97 [4] Answers to Operators Key questions to the UTRA Concept groups of Concept Group Delta, Wideband TDMA/CDMA, ETSI SMG2 Tdoc SMG2 308/97 [5] Ericsson, PCS1900, The New Personal Communications System for North America, Technical Analysis Appendix, December 1994

03/20/13

17

Vous aimerez peut-être aussi