Tsoulos George (ред.) MIMO System Technology for Wireless Communications

Подождите немного. Документ загружается.

204 MIMO System Technology for Wireless Communications

15. D. Gerlach and A. Paulraj, “Adaptive transmitting antenna arrays with feed-

back,” IEEE Signal Processing Letters, Vol. 1, No. 10, October 1994, pp. 150–152.

16. T. Haustein, C. von Helmolt, E. Jorwieck, V. Jungnickel, and V. Pohl. 2002.

“Performance of MIMO systems with channel inversion,” in Proceedings of the

IEEE 55th Vehicular Technology Conference, Vol. 1, Birmingham, AL, May 2002,

pp. 35–39.

17. C.B. Peel, B.M. Hochwald, and A.L. Swindlehurst. 2005. “A vector-perturbation

technique for near-capacity multiantenna multiuser communication-part I:

channel inversion and regularization,” IEEE Transactions on Communications,

Vol. 53, No. 1, January 2005, pp. 195–202.

18. M. Stojnic, H. Vikalo, and B. Hassibi. 2004. “Rate maximization in multi-anten-

na broadcast channels with linear preprocessing,” in Proceedings of IEEE Globe-

com, November 2004, pp. 3957–3961.

19. F. Rashid-Farrokhi, K.R. Liu, and L. Tassiulas. 1998. “Transmit beamforming

and power control for cellular wireless systems,” IEEE Journal on Selected Areas

in Communications, Vol. 16, No. 8, October 1998, pp. 1437–1450.

20. E. Visotsky and U. Madhow. 1999. “Optimum beamforming using transmit

antenna arrays,” in Proceedings of the IEEE Vehicular Technology Conference, Vol. 1.

Houston, TX: IEEE, May 16–20, 1999, pp. 851–856.

21. M. Bengtsson and B. Ottersten. 2001. “Optimal and suboptimal beamforming,”

in Handbook of Antennas in Wireless Communications, L. Godara, ed., Boca Raton,

FL: CRC Press.

22. H. Boche and M. Schubert. 2002. “Multi-antenna downlink transmission with

individual SINR receiver constraints for cellular wireless systems,” in Proceed-

ings of the 4th International ITG Conference on Source and Channel Coding, Infor-

mationstechnische Gesellschaft im VDE (ITG). Berlin: VDE Verlag GmbH,

January 2002, pp. 159–166.

23. M. Schubert and H. Boche. 2004. “Solution of the multiuser downlink beam-

forming problem with individual SINR constraints,” IEEE Transactions on

Vehicular Technology, Vol. 53, No. 1, January 2004, pp. 18–28.

24. Q.H. Spencer, A.L. Swindlehurst, and M. Haardt. 2004. “Zero-forcing methods

for downlink spatial multiplexing in multi-user MIMO channels,” IEEE Trans-

actions on Signal Processing, Vol. 52, No. 2, February 2004.

25. J.-H. Chang, L. Tassiulas, and F. Rashid-Farrokhi. 2002. “Joint transmitter re-

ceiver diversity for efﬁcient space division multiaccess,” IEEE Transactions on

Wireless Communications, Vol. 1, No. 1, January 2002, pp. 16–27.

26. Z. Pan, K.-K. Wong, and T. Ng. 2003. “MIMO antenna system for multi-user

multi-stream orthogonal space division multiplexing,” in Proceedings of the IEEE

International Conference on Communications, Vol. 5, Anchorage, AK: IEEE, May

2003, pp. 3220–3224.

27. K.-K. Wong, R. Murch, and K.B. Letaief. 2003. “A joint-channel diagonalization

for multiuser MIMO antenna systems,” IEEE Transactions on Wireless Commu-

nications, Vol. 2, No. 4, July 2003, pp. 773–786.

28. Q.H. Spencer, A.L. Swindlehurst, and M. Haardt. 2003. “Fast power minimi-

zation with QoS constraints in multi-user MIMO downlinks,” in Proceedings of

the IEEE International Conference on Acoustics, Speech, and Signal Processing, IEEE,

April 2003.

4190_book.fm Page 204 Tuesday, February 21, 2006 9:14 AM

Performance of Multi-User Spatial Multiplexing with Measured Channel Data 205

29. Q.H. Spencer and A.L. Swindlehurst. 2004. “A hybrid approach to spatial

multiplexing in multi-user MIMO downlinks,” EURASIP Journal on Wireless

Communications and Networking, Vol. 2004, No. 2, December 15, 2004, pp.

236–247. Available: http://wcn.hindawi.com/volume-2004/issue-2.html.

30. R.L.-U. Choi and R.D. Murch. 2002. “A downlink decomposition transmit pre-

processing technique for multi-user MIMO systems,” in Proceedings of IST

Mobile & Wireless Telecommunications Summit, June 2002.

31. A. Bourdoux and N. Khaled. 2002. “Joint Tx-Rx optimization for MIMO-SDMA

based on a null-space constraint,” in Proceedings of the IEEE Vehicular Technology

Conference, September 2002, pp. 171–174.

32. Q.H. Spencer and M. Haardt. 2002. “Capacity and downlink transmission

algorithms for a multi-user MIMO channel,” in Conference Record of the 36th

Asilomar Conference on Signals, Systems and Computers. IEEE, November 2002.

33. M. Rim. 2002. “Multi-user downlink beamforming with multiple transmit and

receive antennas,” Electronics Letters, Vol. 38, No. 25, December 5, 2002,

pp. 1725–1726.

34. R. Choi and R. Murch, “A transmit preprocessing technique for multiuser

MIMO systems using a decomposition approach,” IEEE Transactions on Wireless

Communications, Vol. 3, No. 1, January 2004, pp. 20–24.

35. V. Stankovic, M. Haardt, and M. Fuchs. 2004. “Combination of block diagonal-

ization and THP transmit ﬁltering for downlink beamforming in multi-user

MIMO systems,” in Proc. European Conference on Wireless Technology (ECWT

2004), Amsterdam, the Netherlands, Oct. 2004, pp. 145–148.

36. J.W. Wallace, M.A. Jensen, A.L. Swindlehurst, and B.D. Jeffs. 2003. “Experimen-

tal characterization of the MIMO wireless channel: data acquisition and anal-

ysis,” IEEE Transactions on Wireless Communications, Vol. 2, No. 2, March 2003,

pp. 335–343.

37. Q.H. Spencer, C.B. Peel, A.L. Swindlehurst, and M. Haardt. 2004. “An intro-

duction to the multi-user MIMO downlink,” IEEE Communications Magazine,

Vol. 42, No. 10, October 2004, pp. 60–67.

38. Q. Spencer and T. Svantesson. 2003. “MIMO downlink spatial multiplexing

algorithms applied to channel measurements,” in Proceedings of the IEEE 58th

Vehicular Technology Conference, Orlando, FL: IEEE, October 6–9 2003.

39. Q.H. Spencer, T. Svantesson, and A.L. Swindlehurst. 2004. “Performance of

MIMO spatial multiplexing algorithms using indoor channel measurements

and models,” Wireless Communications and Mobile Computing, Vol. 4, No. 7,

November 2004, pp. 739–754,

40. P. Zetterberg, M. Bengtsson, D. McNamara, P. Karlsson, and M.A. Beach. 2002.

“Performance of multiple-receive multiple-transmit beamforming in WLAN-

type systems under power or EIRP constraints with delayed channel esti-

mates,” in Proceedings of the 55th Vehicular Technology Conference, Vol. 4, IEEE,

pp. 1906–1910.

4190_book.fm Page 205 Tuesday, February 21, 2006 9:14 AM

207

Multiuser MIMO for UTRA FDD

Jyri Hämäläinen, Risto Wichman, Markku Kuusela, Esa Tiirola,

and Kari Pajukoski

CONTENTS

Abstract.................................................................................................................208

8.1 Introduction................................................................................................208

8.2 UTRA Framework .....................................................................................210

8.2.1 UTRA FDD Downlink .................................................................. 211

8.2.1.1 High-Speed Downlink Packet Access .........................212

8.2.2 UTRA FDD Uplink .......................................................................214

8.3 Current Multiantenna Methods in UTRA FDD....................................214

8.3.1 Beamforming..................................................................................215

8.3.2 Transmit Diversity.........................................................................216

8.4 MIMO in UTRA FDD Uplink..................................................................217

8.4.1 MIMO Algorithms ........................................................................218

8.4.1.1 Diversity MIMO..............................................................218

8.4.1.2 Information MIMO.........................................................219

8.4.1.3 Advanced Receiver Techniques....................................220

8.4.2 Uplink Load Equations ................................................................221

8.4.2.1 Basic Load Equations .....................................................221

8.4.2.2 Load Equations for MIMO............................................223

8.4.3 Performance Comparisons...........................................................225

8.5 MIMO in UTRA FDD Downlink ............................................................229

8.5.1 Theoretical Background ...............................................................230

8.5.1.1 Single-User MIMO..........................................................230

8.5.1.2 Multiuser MIMO.............................................................233

8.5.2 Proposed MIMO Algorithms within 3GPP Standardization ...234

8.5.3 Practical Considerations...............................................................237

8.5.4 Multiuser Beamforming...............................................................239

8.6 Conclusions and Discussion ....................................................................244

References.............................................................................................................245

4190_book.fm Page 207 Tuesday, February 21, 2006 9:14 AM

208 MIMO System Technology for Wireless Communications

Abstract

Link capacity of cellular networks has become an issue with forecasted

growing public demand for high data rate services. Recently, multiple-input

multiple-output (MIMO) antenna techniques have been received enthusias-

tically within wireless communications due to their potential to increase link

capacities in a spectrally efﬁcient manner. Despite their great theoretical

promise, the introduction of MIMO transceivers to commercial wireless com-

munication systems faces several challenges.

In this chapter, we introduce recent developments of MIMO techniques

within a third generation cellular system known as universal terrestrial radio

access (UTRA), which is standardized by the third generation partnership

project (3GPP). Rather than discussing general MIMO algorithms, we will

concentrate on MIMO techniques that can be implemented without major

revisions to the current UTRA air interface, wideband code division multiple

access (WCDMA). The ﬁrst full speciﬁcation of WCDMA was already com-

pleted at the end of 1999, and backward compatibility to the legacy system

is one of the most important design issues when considering future enhance-

ments to WCDMA. Thus, small modiﬁcations to air interface can be accepted

more easily in standardization than schemes that would require large rede-

sign efforts. We will concentrate on UTRA frequency division duplex (FDD)

mode, because it is a more widely deployed system than UTRA time division

duplex mode.

8.1 Introduction

Multiple-input multiple-output (MIMO) transceivers have great potential to

improve the performance of wireless systems. Compared to wireline systems,

the inferior capacity of wireless cellular systems is caused by several different

physical constraints like co-channel and adjacent channel interference, chan-

nel propagation loss, and ﬂat or multipath fading channels. Multiantenna

transmission and reception techniques are currently regarded as one of the

most promising approaches for signiﬁcantly increasing the coverage, capac-

ity, and spectral efﬁciency of wireless systems.

The multiple degrees of freedom offered by multiple transmit and receive

antennas can be used for diversity or for spatial multiplexing. In the former

case, a single data stream is transmitted and multiple antennas are used to

decrease the variance of the received signal and thereby improve the quality

of the radio link. In the latter case, multiple transceiver antennas are used

for parallel multiplexing, i.e., to transmit several data streams simultaneously

4190_book.fm Page 208 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD

209

to a user to increase peak data rates. In this study, these two approaches will

be referred to as diversity MIMO and information MIMO, respectively.

In terms of channel capacity, information MIMO is a much more impres-

sive concept than diversity MIMO. With parallel multiplexing, the capacity

increases linearly with the number of transmit and receive antennas [1,2]

when the number of transmit antennas equals the number of receive anten-

nas and when the channels between the antennas are independent and

identically distributed. This observation has stimulated great interest in

MIMO transceivers, and in addition to academic research, MIMO is actively

being studied in standardization for different

wireless systems, such as

UTRA, IEEE 802.11n and IEEE 802.16e.

Multiple antenna transceivers promise high spectral efﬁciencies to wireless

systems, and spectral efﬁciencies as large as 20 to 40 bps/Hz have been

reported in laboratory environments [3]. This would present a signiﬁcant

improvement to current cellular systems that typically operate with 0.05 to

2 bps/Hz range. However, the introduction of MIMO to commercial systems

still faces several challenges, especially when a system has not been designed

for MIMO from the beginning. For example, channel estimation in GSM

relies on seven different midamble sequences that have been carefully

selected to possess optimal autocorrelation properties. If one base station

(BS) uses multiple midamble sequences, the seven sequences quickly run

out, because frequency reuse limits the usage of the same sequences in

nearby base stations. Furthermore, cross-correlation properties of the GSM

training sequences are poor, making their use in the same base station sub-

optimal. Thus, the introduction of coherent information MIMO techniques

would require the redesign of the entire GSM channel estimation scheme.

At the same time hundreds of millions of legacy GSM handsets should be

compatible with the redesigned system. In principle, differential modulation

or blind detection and estimation could be applied to relax the requirement

of multiple training sequences in the case of MIMO. However, blind algo-

rithms are not considered viable in commercial cellular systems due to their

complexity of implementation. Differential modulation suffers from 3 dB

SNR penalty when compared to coherent signaling, and the use of differen-

tial modulation and increased power levels would increase the interference

in wireless networks.

The ﬁrst release of universal terrestrial radio access (UTRA) frequency

division duplex (FDD) air interface, referred to as wideband code division

multiple access (WCDMA), was completed in 1999 by the third generation

partnership project (3GPP). At that time, research on information MIMO

techniques was in the initial stage, but the release already speciﬁed two-

antenna open-loop and closed-loop transmit diversity modes that are part

of the most recent Release 6 [4] as well. Still, the introduction of more

advanced MIMO techniques to WCDMA has not been straightforward. A

great deal of standardization effort has been dedicated to extend the speci-

ﬁcation of diversity MIMO to more than two transmit antennas [5], and

4190_book.fm Page 209 Tuesday, February 21, 2006 9:14 AM

210 MIMO System Technology for Wireless Communications

information MIMO techniques to WCDMA high-speed downlink packet

access (HSDPA) [6] have been extensively studied as well [7]. However, these

techniques were not accepted before Release 6 and the standardization for

Release 7 is still ongoing.

In this chapter, we concentrate on MIMO transceiver techniques that do

not require major modiﬁcations to the current UTRA FDD speciﬁcations.

Instead of assessing the performance of single user MIMO systems we dis-

cuss MIMO transceivers in system level within UTRA FDD. This is because

the gains of a single link do not necessarily translate into gains in a cellular

system where users are subject to intracell and intercell interference. To this

end, it is ﬁrst necessary to understand the limitations and characteristics of

UTRA FDD air interface. Section 8.2 highlights the important features in

WCDMA uplink and downlink, and Section 8.3 presents multiantenna trans-

ceiver techniques that can be applied within the current WCDMA framework.

The salient difference between the uplink and downlink in UTRA FDD is

that the uplink is interference limited while the downlink is code limited.

Thus, code reuse and spectrally efﬁcient MIMO techniques are in great

demand in downlink, while in uplink each user can employ the entire chan-

nelization code set and MIMO schemes relying on well-known multiuser

detection algorithms [8,9]. MIMO in UTRA FDD uplink is discussed in

Section 8.4.

Section 8.5 presents various information MIMO algorithms [10] that have

been proposed for WCDMA HSDPA mode. Requirements for MIMO per-

formance according to [10] are discussed and their meaning to the develop-

ment of MIMO algorithms is outlined. It is interesting to note that while the

discussion on MIMO transceivers usually focuses on high data rates, increas-

ing peak data rates are only considered of secondary importance in 3GPP,

the most important factors being coverage and cell throughput. At the end

of this section we present a multiuser MIMO scheme for WCDMA downlink

that increases cell throughput without requiring any modiﬁcations to the

present UTRA FDD speciﬁcation. Finally, discussion and visions on future

developments of MIMO transceiver techniques within UTRA FDD frame-

work are presented in Section 8.6.

8.2 UTRA Framework

In WCDMA, narrowband user information is spread over a wide bandwidth

by modulating a low-rate user data sequence with a high-rate spreading

code (channelization code). The length of the spreading codes varies from

4 to 512 chips (in the uplink the maximum code length is 256 chips) and the

chip rate is 3.84 Mcps (Mega chips per second), which together with transmit

pulse shape ﬁlter leads to the carrier bandwidth of approximately 5 MHz.

4190_book.fm Page 210 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD

211

Variable spreading factors and multicode transmission are used in order to

support a wide scale of different data rates.

Scrambling codes of length 38400 chips are employed on top of channel-

ization codes. While channelization codes are used in the downlink to sep-

arate different intracell users and separation of intercell users is based on

different scrambling codes, in the uplink the scrambling codes separate

different users and channelization codes separate different physical data and

control channels of a user. This difference between the code usage in the

downlink and uplink is essential: The limited number of orthogonal chan-

nelization codes imposes a strict upper bound to the achievable downlink

capacity unless information MIMO techniques

reusing the channelization

codes are used. The limitation is not strict in the uplink, because one user

may use all channelization codes and the set of long scrambling codes

contains several million codes.

The basic UTRA FDD mode supports user data rates up to 2.3 Mbps both

in uplink and downlink. In downlink, transmission on the peak data rate

requires the use of three parallel channelization codes with spreading factor

4, which allocates 75% of code resources to a single user. In uplink, terminal

transmitting with the peak data rate is seen as a large interference source in

the network but available code resources of other intracell users are not

affected. High peak-to-average ratio of multicode signals sets stringent

requirements to transmitters, which are particularly critical to mobile stations.

8.2.1 UTRA FDD Downlink

Figure 8.1 presents a simpliﬁed downlink channel and frame structure. Each

frame is divided into 15 time slots, with the frame length being 10 ms. The

radio frame carries a dedicated physical channel (DPCH) that further divides

into dedicated physical data channel (DPDCH) and dedicated physical con-

trol channel (DPCCH). The control channel contains transmit power control

(TPC) bits for fast power control, transport format combination indicator

FIGURE 8.1

Downlink frame structure for dedicated physical channel.

DPDCH

Data bits

Power

1 Radio frame = 15 time slots: 10 ms

Other control channels

S-CPICH(0−15)

P-CPICH (1−2)

Data bits Pilot bits TPC TFCI

DPCCH DPDCH DPCCH

4190_book.fm Page 211 Tuesday, February 21, 2006 9:14 AM

212 MIMO System Technology for Wireless Communications

(TFCI) bits that inform the receiver which transport channels are active and

the pilot bits enabling the channel and signal to interference and noise ratio

(SINR) estimation from DPCH.

Figure 8.1 further shows the primary common pilot channel (P-CPICH)

and the secondary common pilot channel (S-CPICH). The numbers in paren-

theses indicate that one sector in downlink may contain 1–2 parallel

P-CPICHs and 0–15 parallel S-CPICHs. The downlink pilot structure is intro-

duced here, because it is needed to understand system speciﬁc problems of

different multiantenna processing methods. While the receiver may estimate

the channel from the dedicated channel, a better result is usually obtained

when employing P-CPICH or S-CPICH, because these common pilot chan-

nels, especially P-CPICH, have higher transmit power. In practice, P-CPICH

usually takes 10% of transmit power resources of the base station, while

power on S-CPICHs is lower. This high transmit power on P-CPICH is due

to the fact that P-CPICH is always transmitted to the whole cell enabling

handover and cell selection/reselection measurements. By adjusting power

differences between P-CPICHs of adjacent cells, the network load can be

balanced. The power of S-CPICH can be lower because it has to cover only

part of the cell. Usually S-CPICH is applied in connection with beamforming.

According to present UTRA FDD speciﬁcation only two P-CPICHs are

available at maximum while multiple S-CPICHs can be introduced, for exam-

ple, for beamforming purposes. This introduces a serious backward compat-

ibility problem for the design of MIMO algorithms. If the power in P-CPICH

is kept ﬁxed but the power is divided to more than two transmit antennas,

the channel estimation in legacy mobile terminals is deteriorated. If the

power of P-CPICH is increased, interference to network increases as well.

8.2.1.1 High-Speed Downlink Packet Access

Starting from Release 5, WCDMA contains a highly optimized downlink

data transmission concept referred to as high speed downlink packet access

(HSDPA). The HSDPA concept introduces a new type of transport channel,

referred to as the high speed downlink shared channel (HS-DSCH), with a

ﬁxed spreading factor of length 16. HSDPA applies several advanced phys-

ical layer solutions to enable high throughput and reduced service delays.

The new physical layer techniques in HSDPA are outlined below.

• Adaptive modulation and coding (link adaptation)

• High order modulation

• Hybrid automatic repeat request (ARQ) solution

• Reduced length (2 ms) transport time interval (TTI)

• Fast cell selection (FCS)

• Physical layer scheduler located in base station

4190_book.fm Page 212 Tuesday, February 21, 2006 9:14 AM

Multiuser MIMO for UTRA FDD

213

The control of HS-DSCH is terminated in the base station to adapt rapidly

to changing channel conditions. Adaptation rate is increased by shortening

transport time interval TTI from 10 ms in Release ’99 and Release 4 to 2 ms.

The same TTI can be used to transmit with multiple channelization codes

to the same user equipment (UE), depending on terminal capability. In

addition, it is possible to multiplex multiple UEs in the code domain within

one HS-DSCH TTI.

Link adaptation is implemented with a large set of possible transport

format combinations, each associated with a unique combination of modu-

lation, coding, and block size parameters. Release ’99 and Release 4 support

only QPSK modulation, but an HSDPA terminal also supports 16-QAM. This

increases the peak date rates in good channel conditions. Fast power control,

which aims to keep the received SINR constant, is not applied. This is because

with non-real time data transmission it is more efﬁcient to allow the received

SINR to vary and change modulation and coding according to the channel

state. Furthermore, fast power control used together with a high date rate

and high power transport channel would generate large interference peaks

to neighboring cells.

Data rates are assigned to a UE based on channel state information. Chan-

nel state information is measured at the UE and signaled to the base station

in the form of a channel quality indicator (CQI). Thus, CQI feedback provides

the necessary information for efﬁcient real-time link adaptation, and this

enables a form of multiuser diversity when applied together with physical

layer scheduling. In a single-input single-output (SISO) scenario, throughput

is maximized by maximum SNR scheduler transmitting to the user that

reports the best SNR to the base station [11]. When the distances of the users

from the serving base station are different, the scheduler is not fair, because

the users on cell edge rarely get any transmission. Therefore, different fair

schedulers have been developed, which aim to equalize transmission periods

among the users. This kind of physical layer scheduling mechanism is part

of the 3GPP WCDMA HSDPA speciﬁcation, but its implementation is a

vendor-speciﬁc option. Instead of SNR, it is also possible to signal data rates

or capacities by mapping the SNR values appropriately.

Fast cell selection is used to select the transmitting base station from the

set of active base stations. Hybrid ARQ (HARQ) combines retransmission

with the previous transmissions to improve reliability, while ARQ ignores

erroneously received packets. HARQ provides implicit rate matching, while

adaptive modulation and coding (AMC) is used to maximize instantaneous

throughput given the instantaneous CQI.

A number of contributions have emerged within 3GPP that propose dif-

ferent MIMO and MISO transceivers to further increase instantaneous data

rates or spectral efﬁciency of HSDPA. These proposals are presented in more

detail in Section 8.2. In addition, the enhancements to the physical layer

techniques already presented are considered as well [13].

4190_book.fm Page 213 Tuesday, February 21, 2006 9:14 AM