Capacity gain of an uplink-synchronous WCDMA system under ch

982IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY,VOL.53,NO.4,JULY2004 Capacity Gain of an Uplink-Synchronous WCDMA System Under Channelization Code Constraints

JoséOutes Carnero,Klaus Ingemann Pedersen,Member,IEEE,and Preben E.Mogensen,Member,IEEE

Abstract—The performance of an uplink-synchronous wide-band code-pision multiple-access(WCDMA)system is evaluated for radio environments with low temporal dispersion. The capacity gain of synchronous WCDMA is evaluated theoreti-cally under certain constraints and by means of extensive dynamic system level simulations for more advanced scenarios.The effect of channelization code shortage,the impact of the dispersive radio channel on the orthogonality of received signals,and soft han-dover are some of the considered effects.The potential capacity gain is found to equal35.8%in a multicell scenario,conditioned on an infinite number of channelization codes per cell.For a more realistic scenario with channelization code constraints,the capacity gain is reduced to9.6%.The absolute number of users per cell,relative to the available number of channelization codes within each scrambling code group,is therefore found to be an important metric.This further suggests that the capacity gain of synchronous WCDMA decreases when other capacity-enhancing techniques are deployed,such as uplink antenna persity,soft handover,voice activity detection,etc.The presented simulation results in the case where soft handover is not considered accurately match the analytical findings.

Index Terms—channelization code restriction,noise rise,orthog-onal codes,system level,uplink capacity,uplink synchronization.

I.I NTRODUCTION

D URING the last two decades,several uplink capacity

enhancement techniques have been studied for wide-band code-pision multiple-access(WCDMA)systems.To mention a few,application of antenna array techniques have been studied in,e.g.,[1]and[2],and multiuser detection schemes have been addressed in[3]–[5],among others.An alternative approach is to synchronize the uplink so that signals transmitted from different user equipments(UE)1within the same cell are time aligned at the base station(BS).This opens for the utilization of orthogonal codes for own-cell UE separation,so the own-cell interference is,in principle,completely mitigated.The same approach is used in the downlink of the universal mobile telecommunication system(UMTS),where signals from the same BS are separated by time-synchronized orthogonal codes derived from the set of Walsh codes[6].Previous studies have demonstrated that it is possible to achieve accurate symbol

Manuscript received May14,2002;revised May9,2003,November25,2003, and March24,2004.

J.Outes Carnero is with the Department of Communication Technology,Aal-borg University,Aalborg DK-9220,Denmark(e-mail:outes@kom.aau.dk). K.I.Pedersen is with Nokia Networks,Aalborg DK-9220,Denmark.

P.E.Mogensen is with the Department of Communication Technology,Aal-borg University,Aalborg DK-9220,Denmark,and also with Nokia Networks, Aalborg DK-9220,Denmark.

Digital Object Identifier10.1109/TVT.2004.830944

1Note that the term user equipment is synonymous with mobile station(MS).synchronization in the uplink by using a maximum likelihood acquisition algorithm[7]or a simple delay-tracking loop for low-mobility UEs[8].Reliable synchronization can also be obtained in environments with temporal dispersion in the radio channel by estimating both the line-of-sight(LoS)component and the multipath parameters[9].The potential capacity gain of using uplink synchronized WCDMA(compared to asyn-chronous schemes)was first evaluated in[10]for a single-cell scenario,assuming an infinite number of available orthogonal codes.Further,the uplink-synchronized WCDMA is currently being discussed as a candidate feature within the standard-ization body,the third-generation partnership project(3GPP) [8].In3GPP,uplink-synchronized WCDMA is denoted the uplink-synchronous transmission scheme(USTS).

In this paper,we will quantify the capacity gain of uplink synchronized WCDMA in a multicell environment,assuming that UEs can only be synchronized to one cell.As will be demonstrated,this assumption impacts on the potential per-formance gain of using soft handover(SHO).The orthogonal codes used for own-cell separation of UEs are assumed to be Walsh codes.This implies a finite set of orthogonal codes, which tends to severely limit the capacity gain of uplink-syn-chronous WCDMA.Scenarios with single-and dual-antenna reception at the BS are addressed.The potential capacity gain of synchronous WCDMA is theoretically derived and evaluated by means of Monte Carlo simulations,using a dynamic network simulator.

This paper is organized as follows.Section II discusses various aspects of an uplink-synchronous WCDMA system.

A theoretical analysis of the potential capacity gain of up-link-synchronous WCDMA is presented in Section III.The simulation methodology and the considered scenarios used in the dynamic system simulator are described in Section IV. Both theoretical results and simulation results are presented and compared in Section V(a subset of the results presented in Section V has appeared in the3GPP technical report[8]). Concluding remarks and discussion are presented in Section VI.

II.U PLINK S YNCHRONISM IN WCDMA

In a conventional uplink-asynchronous direct-sequence WCDMA system,every UE is allocated a unique pseudonoise (PN)sequence.Adopting the UMTS terminology,we will refer to the PN sequences as scrambling codes.In uplink-syn-chronized WCDMA,UEs within the same cell share the same scrambling code,while using different orthogonal channelization codes derived from the set of Walsh codes. The channelization codes are utilized to separate physical

0018-9545/04$20.00?2004IEEE

OUTES et al.:CAPACITY GAIN OF AN UPLINK-SYNCHRONOUS WCDMA SYSTEM983

channels from UEs.Hence,the narrow-band signal from a UE is both multiplied by a channelization code(Walsh code) and a scrambling code.The scrambling code is cell specific; the number of available channelization codes sets an upper limit of the maximum number of UEs per cell.However,this limitation can be lifted by introducing several scrambling code groups within a cell.This implies that a certain set of UEs are transmitted under one scrambling code while another set of UEs is transmitted under different scrambling codes. The introduction of multiple scrambling codes within a cell eliminates the constraint on the maximum number of UEs due to channelization code shortage.However,this is obtained at the expense of an increased multiple-access interference (MAI),since signals transmitted under different scrambling codes are nonorthogonal.As will be demonstrated in the later sections,the limited number of channelization codes under a single scrambling code is a major limitation on the capacity gain of introducing USTS in UMTS.

In principle,the application of synchronized WCDMA in flat fading radio channels completely cancels the own-cell interfer-ence experienced at the BS(i.e.,after despreading).Other-cell interference is suppressed by the basic processing gain,since UEs are assumed to be synchronized only to their serving cell. Thus,the potential capacity gain of synchronous WCDMA strongly depends on the other-to-own-cell interference ratio, i.e.,the gain in the single cell is significantly larger than that in multicell environments.It is well known that in flat fading radio channels the use of dual antenna reception with maximal ratio combining(MRC)reduces the other-to-own-cell interference ratio compared to the single-antenna reception[11].In the later sections,results for both the one and the two antenna reception will be presented.Naturally,the capacity gain also depends on the degree of synchronization.In this study,we will assume that a simple time-delay tracking loop is applied with a finite resolution of timing adjustments in the UE.It is assumed that the timing error between the received signals at the BS is periodically estimated and,subsequently,the UEs are informed via the downlink to increase or decrease their transmit timing. This is a very simple approach that has proven to be sufficiently accurate for UEs with low velocities,assuming a radio channel with marginal time dispersion[8].In environments with large temporal dispersion,the orthogonality of the signals at the BS is known to be degraded due to the misalignment of signals carried by different multipath components.However,in radio channels with low time dispersion,such as the ITU power-delay profile Pedestrian A[12],the synchronization loss is marginal [13].

The fact that a UE can be synchronized to only a single cell has an impact on the potential uplink SHO gain.In order to illustrate this,let us consider the following example,where a UE is in SHO with cell number one and cell number two with the UE synchronized to cell number one.The path loss toward both cells is the same,assuming an equal wide-band interference level at the two cells.The signal-to-interference ratio(SIR)after despreading will,therefore,be higher at cell number one as compared to cell number two,because the own-cell interference in cell number two is nonorthogonal.In an asynchronous WCDMA system,the signal-to-noise ratio (SNR)after despreading would have been equal in the two cells,indicating a significant SHO persity gain by using either selection persity combining or MRC.The loss of SHO gain for synchronous WCDMA is addressed in this study.Here,it is assumed that a UE that enters SHO remains synchronized with its original serving cell during the SHO period.When the UE drops the connection with the original cell,it synchronizes with the new serving cell.

In the following,the potential capacity gain of uplink-syn-chronous WCDMA will be evaluated.Scenarios where only a certain percentage of UEs are operated in synchronous mode are also addressed.The latter is of interest for an existing system such as UMTS,where USTS will most likely be introduced gradually(i.e.,in the initial phase there will be a mixture of synchronous and asynchronous UEs)based on UE capability classes.

III.T HEORETICAL A NALYSIS OF THE C APACITY OF

U PLINK-S YNCHRONOUS WCDMA

Let us derive a simple expression for the expected uplink ca-pacity gain of a synchronous WCDMA system,where we as-sume that all UEs are operated with the same bit rate and are perfectly power controlled.In this context,perfect power con-trol refers to the case where the transmit power of each UE is adjusted so that the energy-per-bit-to-noise ratio(Eb/No)at the BS equals a target value corresponding to a certain block error rate.For the sake of simplicity,we will not include cases where UEs are in SHO mode in this simple theoretical analysis.SHO cases are instead studied by means of dynamic system level sim-ulations in Sections IV and V.We will define the maximum cell capacity as the number of UEs that can be supported at a given noise rise(NR)at the BS.The NR at the BS is known to be a robust measure of the uplink load of a WCDMA system,which is often used by radio resource management(RRM)algorithms to control the uplink load[6],[14].The NR is defined as

NR(1)

where is total average received wideband power at the BS

and is the power of the thermal noise at the BS.The NR is related to the uplink load factor as

[6]

NR

NR

(2)

where is the uplink load factor.The system has reached its pole capacity

when approaches unity.Hence,the NR can be used to control how close the system is operated to the pole capacity.In deriving an expression for the NR,we will assume that there

are UEs in the cell of interest,which are transmitting asynchronously.In addition,there

are

UEs in synchronous mode transmitting under scrambling code

number.Let us furthermore assume that the required Eb/No is identical for all UEs.Under these assumptions,we can approxi-mate the Eb/No for the UEs operating in asynchronous mode

as

(3)

984IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY ,VOL.53,NO.4,JULY

2004Fig.1.NR as a function of the number of UEs.

where is the effective processing gain (ratio between the chip

rate and the bit

rate),is the received power level at the BS

from a UE in asynchronous mode,

and

is the total received power.Similarly,we can express the Eb/No for synchronous

UEs under scrambling code

number

as

(4)

where is the received power level at the BS from a syn-

chronous UE under scrambling code

number

and

expresses the degree of orthogonality between the signals

received under the same scrambling code.Assuming a radio

channel with marginal time dispersion and perfect synchro-

nization at the

BS,

equals unity.On the contrary,if the synchronization of the signals received at the BS completely

fails due to erroneous adjustment of timing of the transmitted

signals or excessive time dispersion in the radio channel,

then

.Hence,assuming a perfectly synchronized WCDMA

system where the average power delay profile of the radio

channel between all UEs and their serving BS is the same,(4)

is valid.The

factor is often referred to as the orthogonality

factor [15]–[17].From (3)and (4),the following expressions

are

obtained:

(5)

(6)The total received power at the BS can be expressed

as (7)

where is the own-cell

power,is the other-cell

power,is the noise level,

and is the other-to-own-cell interference ratio.The own-cell power

equals (8)

where is the number of enabled scrambling code groups within the cell of 2dd8c08fb0717fd5360cdc50bining (5),(6),and (8)

yields (9)An expression for the NR at the BS is subsequently obtained by combining (7)and (9),i.e.,

NR (10)For systems where UEs are not necessarily assumed to be trans-

mitting all the time,we will introduce an activity

factor

OUTES et al.:CAPACITY GAIN OF AN UPLINK-SYNCHRONOUS WCDMA SYSTEM

985

Fig.2.Capacity gain of uplink-synchronous WCDMA for a NR target of4.0

dB.

.

Hence,for UEs with constant transmission,

while

is equivalent to transmission50%of the time.As-

suming that the average activity factor is identical for all the

UEs,it can easily be shown that the expression in(10)general-

izes to

NR

(11)

The expression for the NR is plotted in Fig.1versus the number

of UEs,conditioned

on

,(3.84Mc/s/12.2

kb/s),

dB,,

and.These parameter

settings correspond to a typical microcellular environment with

12.2kb/s speech users[6],assuming a chip rate of3.84Mc/s.

Results are presented for cases where all UEs are either in asyn-

chronous or synchronous mode.The curve labeled no code limit

refers to the case

where,independent of the number of

synchronous UEs,i.e.,corresponding to an infinite number of

channelization codes under a single scrambling code.The other

curve for UEs in synchronous mode assumes a maximum of

50UEs under each scrambling code[8].For this scenario,UEs

are first allocated under scrambling code number one.Once the

number of UEs exceeds the maximum number of channelization

TABLE I

M ODIFIED ITU P EDESTRIAN A P OWER D ELAY P

ROFILE

codes,the second scrambling code is enabled,and so forth.As

an example,for65synchronized

UEs,

with

and.It is observed from Fig.1that the NR increases

rapidly for the case where all UEs are operated in asynchronous

mode,while the NR increases much more slowly for the cases

with synchronous UEs.This is equivalent to a capacity gain of

uplink-synchronous WCDMA compared to conventional asyn-

chronous systems.The two NR curves for the synchronous cases

are identical up to50UEs,whereafter the curve conditioned on a

maximum number of UEs per scrambling code starts to increase

much more quickly.This behavior is observed because synchro-

nized UEs under different scrambling codes are nonorthogonal.

Let us define the capacity in a cell as the sum of the

throughput transmitted by every UE in the cell.Hence,the

capacity per cell in a synchronous and an asynchronous system

equals

(12)

(13)

986IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY ,VOL.53,NO.4,JULY

2004Fig.3.Mapping table of FER versus average Eb/No for Pedestrian A and a UE speed of 3.0km/h.

respectively,

where is the bit rate for a single UE.The

capacity gain of uplink-synchronous WCDMA can then be

expressed

as

(14)

By carrying out some manipulations,from (11)–(14)it is pos-

sible to derive the following expression for the capacity gain of

uplink

synchronization:

(15)

where

is the maximum number of UEs that can be allo-cated under the same scrambling code

and

is the load factor increase associated to the UEs in scrambling

code

.

and

can be calculated

as

(16)

(17)

where denotes the smallest integer value that is bigger

than

.

The capacity gain of synchronous WCDMA for a NR target of

4.0dB is plotted in Fig.2versus for different values

of ,with

and without constraints on the maximum number of synchro-

nized UEs under a single scrambling code.According to (2),

NR dB corresponds

to ,which indicates that we are quantifying the capacity gain at 60%of the systems pole ca-pacity.The remaining parameters are identical to the ones used in Fig.1.It is observed that the capacity gain decreases for in-creasing .This is due to the fact that only the own-cell inter-ference is reduced by introducing synchronous WCDMA and,hence,the gain decreases when the other-cell interference starts to become dominant.Similarly,it is observed that the capacity gain decreases for decreasing values

of .For all the consid-ered cases in Fig.2,the capacity gain is larger than or equal to 20%.The marginal variations in the capacity gain for the cases with a limited number of UEs per scrambling code is caused by a varying number of active scrambling

codes versus .IV .S YSTEM M ODEL FOR D YNAMIC S IMULATIONS A.Overall Description Dynamic Monte Carlo network simulations are conducted for both single-and multicell environments,using the general simu-lation methodology outlined in [18].As a reference,simulation parameters are selected according to the UMTS specifications,assuming that the synchronous WCDMA is implemented in co-herence with the preliminary specifications of USTS [8].The multicell simulations consider a network with 24cells,created using three sector BSs with a 3-dB beamwidth antennas of 72.UEs are uniformly distributed in the network,traveling with a fixed speed of 3km/h in a random direction.A new UE en-tering the system automatically gets connected to the cell with a minimum path loss.The distance-dependent path loss is mod-eled according to the pedestrian model in [12],i.e.,the path loss

OUTES et al.:CAPACITY GAIN OF AN UPLINK-SYNCHRONOUS WCDMA SYSTEM987

grows with the distance to the power of four.Shadow fading is modeled with a log-normal distributed random variable,with an exponentially decaying spatial autocorrelation function[19]. Multipath fading is modeled in compliance with the modified ITU power delay profile,Pedestrian A[12].The relative average power of each multipath component for the modified Pedestrian A model is summarized in Table I.The amplitude of each path is assumed to be Rayleigh distributed,independent for each tap, with a classical Doppler spectrum.Each simulation run corre-sponds to200s of real time.

B.Interference and Block Error Rate Calculations

The time resolution of the dynamic network simulator is slot level(i.e.,0.66ms).A conventional rake receiver is assumed to be implemented in the BSs,which tracks all the multipath com-ponents within a window of15dB,compared to the strongest path.Signals tracked by different rake fingers are combined using MRC.In every slot period the Eb/No is first computed for each inpidual rake finger and UE,taking into account that the time-aligned interference from synchronized UEs within the cell is orthogonal.Subsequently,the Eb/No after rake finger combining is obtained as the sum of the Eb/Nos per rake finger.In cases where dual antenna reception is used at the BS, the Eb/No is obtained as the sum of Eb/Nos per rake finger, computed for each antenna branch.This approach is valid provided that the noise and interference at the two antennas are uncorrelated.

The frame error rate(FER)for each connection is computed via lookup tables,which map the geometrical average of Eb/Nos corresponding to one frame length into the FER.This is a well-known approach that is often applied in network simulators[20].The aforementioned mapping table is obtained from link-level simulations of a12.2kb/s speech service with a20-ms frame length.The mapping table is plotted in Fig.3;the results in this figure include a dedicated physical control channel(DPCCH)with spreading factor(SF)256and

a dedicated physical data channel(DPDCH)with SF64or128

[6].The DPCCH and DPDCH are code multiplexed,so each UE uses two channelization codes.Convolutional coding with a coding rate of1/3and an interleaver with either repetition or puncturing,depending on the SF of the DPDCH,is assumed. The overhead of the logical control channel,transmitted at 2.4kb/s,is included in the mapping table.Note that the config-uration with a DPDCH with SF128allows more UEs under the same scrambling code compared to a DPDCH with SF64.This is obtained at the expense of a0.5-dB higher required Eb/No for a FER of1%.

C.Radio Resource-Management Algorithms

A simplified admission-control(AC)algorithm determines whether a new UE is allowed access to a cell.The average NR is computed over a period of100ms(corresponding to150slots) and if the NR is below a predefined NR target,the UE is allowed access;otherwise,the call is rejected.Only one UE is allowed to enter a cell per a20-ms period.A UE in synchronous mode, which is granted capacity in a given cell,is assigned a unique channelization code.If there are no free channelization codes, an additional scrambling code is enabled for the cell,so that

TABLE II

S UMMARY OF THE M AIN S IMULATION P

ARAMETERS

more channelization code resources are made available.This implies that a maximum number of50and84UEs can be sup-ported under one scrambling code for DPDCHs with SF64and SF128,respectively[8].

All the UEs are power controlled by their serving BS.A con-ventional closed-loop power-control(PC)algorithm is imple-mented,where the UE transmit power is adjusted in steps of 1dB every slot period in coherence with the current Eb/No, compared to the Eb/No target[6].The reception of transmit power control commands at the UEs is assumed to be error free and subject to no delays.The Eb/No target is updated every frame period to obtain a FER target of1%according to the outer loop PC algorithm in[21].The maximum transmit power from each UE is limited to24dBm.

SHO is modeled in the simulator,assuming selection per-sity combining between cells on different BSs and MRC be-tween cells on the same BS(i.e.,softer handover).For the sake of simplicity,the active set size is limited to two.A UE that en-ters SHO remains synchronized to its serving BS.Decisions on adding,replacing,or dropping a SHO leg are based on simple average path-loss measurements from the UE to the BSs in the candidate set and/or the active set[22].This is a well-known ap-proach for implementation of a robust handover algorithm.All the relevant simulation parameters are summarized in Table II.

V.S IMULATION R ESULTS

A.Single-Cell Results

Simulation results from a single-cell scenario are presented in Fig.4for various numbers of receiver antennas and activity fac-tors.It is assumed that the DPDCHs are transmitted at SF64for all UEs.The average number of UEs is reported for synchronous and asynchronous uplink WCDMA scenarios.The capacity gain of the synchronous WCDMA is observed to be equal to45.7%, assuming one receiver antenna and an activity factor of one.De-creasing the activity factor to0.5results in a reduced capacity gain,i.e.,the capacity gain drops to26.1%.This is caused by

988IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY,VOL.53,NO.4,JULY

2004

Fig.4.Average number of UEs per cell in single-cell environment(DPDCHs with SF64). the increase in the number of UEs,which requires additional

active scrambling codes in order to avoid channelization code shortage.Let us recall that a maximum of50UEs can be sup-

ported under each scrambling code.The capacity gain equals

13.0%for the case with two antennas at the BS,assuming an ac-

tivity factor of0.5.For this particular scenario,five scrambling

codes are required on an average,where all the channelization

codes under four of the scrambling codes are fully utilized.Only

a few channelization codes under the fifth scrambling code are

in use.Hence,for the demodulation of the received signal from

each UE,only approximately25%of the total received interfer-

ence is orthogonal.This results in the rather low capacity gain of

13.0%,compared to the capacity gain of45.7%for the case with

a single BS antenna and an activity factor of one.For systems

with no constraints on the number of available channelization codes,the capacity gain would approximately equal100%for

all four of the cases considered(for the particular case of two an-

tennas and an activity factor of0.5,the capacity gain increases

from13.0%to102.1%).The simulation results in Fig.4and

the corresponding theoretical results in Table III are observed

to match.The theoretical results are computed according to the

expressions in Section III

for dB(the value that cor-

responds to1%FER in Fig.

3),(obtained by using

the expression in[15]with the multipath model in Table I),a

NR target of4dB,

and.For the case with two receive an-

tennas,dB(6.1–3.0).

B.Multicell Results Without Soft Handover

Multicell simulation results are presented in Fig.5,assuming

that SHO is disabled.The capacity gain for two antennas at

the BS and an activity factor of0.5is reported to equal10.8%

under the channelization code constraint.Hence,for this par-

ticular setup,the capacity gain is slightly lower than the ca-

pacity gain for the single-cell scenario.This can be explained

as follows:In the multicell environment,both the own-cell and

other-cell interference are received at the BS.For the consid-

ered scenario,an approximately equal amount of own-cell and

other-cell interference is

received.From Fig.5,it

is observed that three scrambling codes are required per cell

(i.e.,

ceil

UEs UEs per scrambling

code),where

the load of the third code equals33%(17/50)on an average,

TABLE III

T HEORETICAL S INGLE-C ELL R ESULTS(DPDCH S W ITH SF

64)

TABLE IV

T HEORETICAL M ULTICELL R ESULTS(DPDCH S W ITH SF

64)

while all the channelization codes under the first two scrambling

codes are in use.Hence,a UE transmitting under scrambling

code number one or two is subject to approximately43%or-

thogonal and57%nonorthogonal own-cell interference,plus the

nonorthogonal other-cell interference.This is equivalent to the

single-cell case,where approximately24%of the total received

interference could be regarded as orthogonal.However,for the

UEs transmitting under the third scrambling code,only8%of

the total received interference is orthogonal.For comparison,

the corresponding theoretical findings are reported in Table IV.

It is observed that the simulation results match the theoretical

findings.The theoretical results are calculated for equal to1.6

and0.8for one and two receive antennas,respectively.

C.Multicell Results With Soft Handover Enabled

In order to illustrate the impact of SHO on the performance

of synchronous WCDMA,multicell results with SHO enabled

are presented in Fig.6.The used SHO parameter settings re-

sult in a scenario where27%of the UEs are in two-way SHO.

Comparing the average number of UEs per cell in Figs.5and

6,it is observed that SHO provides a capacity gain for all the

considered cases.However,due to the increase in the absolute

number of UEs per cell,the load of the third scrambling code

OUTES et al.:CAPACITY GAIN OF AN UPLINK-SYNCHRONOUS WCDMA SYSTEM

989Fig.5.Average number of UEs per cell in multicell environment without SHO (DPDCHs with SF

64).

Fig.6.Average number of UEs per cell in a multicell environment with SHO (DPDCHs with SF 64).

slightly increases for the case with SHO,which consequently

maps to a marginal loss of the capacity gain of synchronous

WCDMA.For the case with two BS antennas and an activity

factor of 0.5,the capacity gain drops from 10.8%to 9.6%,for

the case where SHO is enabled.This minor reduction of the ca-

pacity gain leads us to conclude that SHO does not provide a

major obstacle for the capacity gain of synchronous WCDMA.

Notice that a UE in SHO cannot synchronize with more than

one cell simultaneously.On the other hand,the use of SHO re-

duces other-cell interference,which increases the performance

of uplink-synchronous WCDMA.To give an example,for two-

branch reception persity,the average other-cell-to-own-cell

interference ratio is 0.8without SHO and 0.4with SHO.

The main limitation is the relatively small number of available

channelization codes under a single scrambling code.Lifting

the channelization code constraint,our simulation results show

that the capacity gain increases from 9.6%to 35.8%.This re-

sult clearly demonstrates that the capacity gain of synchronous

WCDMA is limited by channelization code shortage.

The number of available channelization codes per scrambling

code can be increased by the use of a higher SF for the DPDCHs,

as discussed in Section IV .Changing the SF of the DPDCHs

from 64to 128,the maximum number of UEs per scrambling

code increases from 50to 84[8],while maintaining the same

effective bit rate.However,this is obtained at the expense of a 0.3dB higher required Eb/No for a FER of 1%(see Fig.3).The results pictured in Fig.7correspond to the case where UEs in synchronous mode are operated with SF 128and asynchronous UEs use SF 2dd8c08fb0717fd5360cdc50paring the results in Figs.6and 7,it is observed that this scheme only performs better than the case with SF 64if the number of UEs per cell is higher than 50and smaller than 84(i.e.,the maximum number of UEs under the same scrambling code for the SF 128case).Hence,for the case of two receiver antennas and an activity factor of 0.5,there is no significant capacity gain from varying the SF of the DPDCHs from 64to 128.D.Capacity Gain Versus the Penetration Rate of Synchronous UEs The previously presented results have compared the capacity gain of a system with asynchronous UEs to a system where all the UEs are operated in synchronous mode.In the following,we will present results for a system with a mixture of UEs operated in either synchronous or asynchronous mode.The ratio between the number of synchronous UEs and the total number of UEs will be referred to as the penetration rate of synchronous UEs,i.e.,a penetration rate of 50%is equivalent to an equal number of synchronous and asynchronous UEs.The presented results are obtained for a multicell scenario with

990IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY,VOL.53,NO.4,JULY

2004

Fig.7.Average number of UEs per cell in multicell environment with SHO(DPDCHs with SF128for synchronous UEs and SF64for asynchronous

UEs).

Fig.8.Capacity gain versus the penetration rate of synchronous UEs in a multicell environment with SHO(DPDCHs with SF64).

SHO enabled and DPDCHs with SF64.The capacity gain is observed to increase gradually with the penetration rate.The dependency between the capacity gain and the penetration rate is nonlinear,so a rather large penetration rate of synchronous UEs is required before a significant capacity gain is observed. The discontinuities on the presented curves are caused by introduction of additional scrambling codes.

VI.C ONCLUDING R EMARKS AND D ISCUSSION

The performance of uplink-synchronous WCDMA has been assessed in terms of the capacity gain relative to an asynchronous system.The capacity gain is evaluated theoreti-cally and by means of dynamic system-level simulations.The simulation results in the case where SHO is not considered accurately match the theoretical predictions.The maximum number of available channelization codes turns out to be a major restriction for the capacity gain of an uplink-synchronous WCDMA,since the signals received from UEs allocated to different scrambling codes are nonorthogonal.For the most realistic conditions,with a noise rise target of4dB(75%of the pole capacity)the capacity gain of an uplink-synchronous WCDMA equals9.6%,whereas in an ideal case without channelization code restrictions,the capacity gain grows to

OUTES et al.:CAPACITY GAIN OF AN UPLINK-SYNCHRONOUS WCDMA SYSTEM 991

35.8%.The use of dual-antenna reception at the BS,soft handover,and voice-activity detection is found to decrease the capacity gain of an uplink-synchronous WCDMA system,since the channelization code limits are reached earlier when these capacity-enhancing techniques are deployed.However,provided that the problems associated with the channelization code shortage can be mitigated (e.g.,by the use of higher order modulation),uplink-synchronous WCDMA has been demonstrated to have a significant potential capacity gain.However,it should be noticed that uplink synchronization in WCDMA require changes to current 3GPP specifications [8]and,hence,it may take years before a high penetration rate of USTS capable UEs is reached.This reduces the potential gain of USTS in the short term (see Fig.8).As mentioned in Sec-tion

I,-branch receive antenna persity [1],[2]and/or mul-tiuser detection at the BS [3]–[5]are alternative methods to im-prove the uplink capacity in WCDMA systems.The latter up-link capacity-enhancing techniques do not require changes to the UEs,i.e.,no legacy UE problems.Thus,despite of the po-tential gain of uplink synchronization,implementation of USTS in WCDMA at this stage should be considered carefully.

R EFERENCES

[1]J.H.Winters,“Smart antennas for wireless systems,”IEEE Pers.

Commun.,vol.5,pp.23–27,Feb.1998.

[2]R.Kohno,“Spatial and temporal communication theory using adaptive

antenna array,”IEEE 2dd8c08fb0717fd5360cdc50mun.,pp.28–35,Feb.1998.

[3] A.Duel-Hallen,J.Holtzman,and Z.Zvonar,“Multiuser detection for

CDMA systems,”IEEE 2dd8c08fb0717fd5360cdc50mun.,vol.2,pp.46–58,Apr.1995.[4]S.Moshavi,“Multi-user detection for DS-CDMA communications,”

IEEE Commun.Mag.,vol.34,pp.124–136,Oct.1996.

[5]S.Verdu,Multiuser Detection .Cambridge,U.K.:Cambridge Univ.

Press,1998.

[6]H.Holma and A.Toskala,Eds.,WCDMA for UMTS.Radio Access for

Third Generation Mobile Communications .New York:Wiley,2000.[7]Y .Wei,J.Krogmeier,and S.Gelfand,“Reliable uplink code-timing

synchronization for cellular DS-CDMA,”in Proc.IEEE Int.Symp.Spread Spectrum Techniques and Applications ,vol.2,NJ,Sept.2000,pp.628–632.

[8](2001)Study Report for Uplink Synchronous Transmission Scheme

(USTS),TR 25.854,Version 5.0.0,Release 5.3rd Generation Partner-ship Project.[Online].Available:2dd8c08fb0717fd5360cdc50

[9]R. D.J.van Nee,“Timing aspects of synchronous CDMA,”in

Proc.IEEE Personal,Indoor and Mobile Radio Communications (PIMRC’94),vol.2,The Hague,The Netherlands,Sept.1994,pp.439–443.

[10] D.K.Kim,S.H.Hwang,E.K.Hong,and S.Y .Lee,“Capacity esti-mation for an uplink synchronised CDMA system with fast TPC and two-antenna persity reception,”IEICE 2dd8c08fb0717fd5360cdc50mun.,vol.E84-B,no.8,pp.2309–2312,Aug.2001.

[11]2dd8c08fb0717fd5360cdc50iho,A.Wacker,and T.Novosad,Eds.,Radio Network Planning and

Optimization for UMTS .New York:Wiley,2002.

[12](1998)Selection Procedures for the Choice of Radio Transmission

Technologies of the UMTS,TR 101112(UMTS 30.03),Version 3.2.0.Eur.Telecommun.Standards Inst.[Online].Available:2dd8c08fb0717fd5360cdc50 [13](2001)System Level Performance of USTS,TSG RAN WG1#21-01-0892.3rd Generation Partnership Project,Turin,Italy.[Online].Avail-able:2dd8c08fb0717fd5360cdc50

[14] E.Dahlman,P.Beming,J.Knutsson,F.Ovesjo,M.Persson,and C.

Roobol,“WCDMA-the radio interface for future mobile multimedia communications,”IEEE Trans.Veh.Technol.,vol.47,pp.1105–1118,Nov.1998.

[15] C.Passerini and G.Falciasecca,“Correlation between delay-spread and

orthogonality factor in urban environments,”Electron.Lett.,vol.37,no.6,pp.384–386,2001.

[16]N.B.Mehta,L.J.Greenstein,T.M.Willis,and Z.Kostic,“Analysis

and results for the orthogonality factor in WCDMA downlinks,”in Proc.IEEE Veh.Technol.Conf.(VTC’02),AL,May 2002,pp.100–104.

[17]K.I.Pedersen and P.E.Mogensen,“The downlink orthogonality

factors influence on WCDMA system performance,”in Proc.IEEE Veh.Technol.Conf.(VTC’02),Vancouver,BC,Canada,Sept.2002,pp.2061–2065.

[18]S.H ?m ?l ?inen,H.Holma,and K.Sipil ?,“Advanced WCDMA radio

network simulator,”presented at the IEEE Personal,Indoor and Mobile Radio Communications (PIMRC ’99),Osaka,Japan,Sept.1999.

[19]M.M.Zonoozi and P.Dassanayake,“Shadow fading in mobile radio

channel,”in Proc.IEEE Personal,Indoor and Mobile Radio Communi-cations (PIMRC’96),vol.2,Taipei,Taiwan,Oct.1996,pp.291–295.[20]S.H ?m ?l ?inen,P.Slanina,M.Hartman,2dd8c08fb0717fd5360cdc50ppetel ?inen,H.Holma,

and O.Salonaho,“A novel interface between link and system level sim-ulations,”in Proc.Advanced Communications Technology and Services (ACTS)Summit ,Aalborg,Denmark,Oct.1997,pp.509–604.

[21] A.Sampath,P.S.Kumar,and J.M.Holtzman,“On setting reverse link

target SIR in a CDMA system,”in Proc.IEEE Veh.Technol.Conf.(VTC’97),vol.2,Phoenix,AZ,May 1997,pp.929–933.

[22]N.Binucci,K.Hiltunen,and M.Caselli,“Soft handover gain in

WCDMA,”in Proc.IEEE Veh.Technol.Conf.(VTC’00),vol.3,Boston,MA,Sept.2000,pp.1467–

1472.

JoséOutes Carnero received the M.Sc.degree in telecommunication engineering from the University of M álaga,M álaga,Spain,in 2000.He is currently working toward the Ph.D.degree at Aalborg Univer-sity,Aalborg,Denmark,in close cooperation with Nokia Networks.

His current research interests include uplink syn-chronization and enhanced uplink packet access for third-generation cellular

systems.

Klaus Ingemann Pedersen (S ’97–A ’00–M ’00)re-ceived the M.Sc.degree in electrical engineering and the Ph.D.degree from Aalborg University,Aalborg,Denmark,in 1996and 2000,respectively.

He is with Nokia Networks,Aalborg,Aalborg,Denmark,where he is a Wireless Network Specialist on various topics related to wide-band code-pision multiple-access (WCDMA)and its evolution.His current research interests include radio resource management for WCDMA systems,high-speed downlink packet access (HSDPA),and adaptive

antenna array

systems.

Preben E.Mogensen (M ’95)received the M.Sc.E.E.and Ph.D.degrees from Aalborg University (AAU),Aalborg,Denmark,in 1988and 1996,respectively,where he has been employed since.

Since 1999,he has been a part-time Research Professor with the Center for Personkommunikation,AAU,where he is leading the Cellular Systems Research (CSYS)group.Since 1999,he has also been with Nokia Networks,Aalborg,Denmark,on a part-time basis,where he is a Principal Engineer in third-generation radio systems research.

本文来源：https://www.bwwdw.com/article/7ual.html