I. INTRODUCTION

Recently, the high-rate data transmission has been one

of key issues in wireless nomadic and mobile

communications. Various classes of multimedia traffic

need to be supported under the wireless LAN (Local Area

Network) as well as cellular environments [1],[2]. A

number of approaches have been considered to improve

the performance of capacity and spectral efficiency in

wireless communication systems [3]~[16]. MIMO

(Multiple-Input Multiple-Output) is an emerging

technology offering high spectral efficiency with the

increased link reliability and interference suppression.

In mobile communication standards, MIMO techniques

have been proposed by different industrial groups. Major

leading standard bodies include WiBro/WiMAX (IEEE

802.16d/e) [3], WiFi (IEEE 802.11n) [4]~[6], and HSDPA

(3GPP) [7]. Their common target is focused on high

spectral efficiency, and hence the candidate schemes are

designed based on the closed-loop systems with feedback

signaling.

In this paper, we overview several candidate schemes

of MIMO in various standard groups, and propose a novel

MIMO solution, which is applicable to cellular systems as

well as wireless LAN. The paper is organized as follows.

In Section II, an overview of MIMO proposals is

described. Section III investigates a novel proposed

scheme which exploits QR decomposition and multi-

channel diversity (MCD). Performance analysis and

simulation results are presented in Section IV and V,

respectively. Section VI draws the conclusions.

II. MIMO PROPOSALS IN STANDARDS

1. WiBro/WiMAX (IEEE 802.16d/e)

WiMax is a wireless technology that provides

broadband data at rates over 3 bits/second/Hz [3]. In order

422

Recently, the industrial organizations have proposed various MIMO schemes in wireless communication standards.

Major standard bodies include WiMAX/WiBro (IEEE 802.16d/e), WiFi (IEEE 802.11n), and HSDPA (3GPP). In this

paper, we overview a number of selected MIMO techniques proposed by major industrial groups and investigate their

performance optimality. We also present our novel multi-user MIMO scheme, of which the sum-rate performance

approaches extremely close to the sum capacity of MIMO downlink channels when the number of users is larger than

the number of transmit antennas. Furthermore, multi-channel diversity (MCD) in the proposed solution greatly reduces

the amount of channel state information signaling, which is fed back from receivers to the transmitter in order to find

optimal precoding structure at the transmitter.

Keywords: MIMO, 3GPP, HSDPA, WiMax, WiFi, WiBro, Multi-user MIMO, Sum-rate.

Sungjin Kim, Hojin Kim, Kiho Kim: Samsung advanced institute of technology

Kwang Bok Lee: Seoul National University

Near-Optimal MIMO Solutions in

WiBro/WiFi/B3G Communication Standards

Sungjin Kim ·Hojin Kim ·Kiho Kim ·Kwang Bok Lee

to increase the range and reliability, the IEEE 802.16e

standard supports optional multiple-antenna techniques

such as space-time block coding, adaptive antenna systems

and MIMO. The closed-loop MIMO schemes in IEEE

802.16e have a common feature which is transmit

precoding. Multiple access scheme is based on orthogonal

frequency division multiple access (OFDMA). Each

transmit scheme uses different feedback signaling. Among

them, Intel proposed SVD based MIMO with multiplexing

transmission. There are two key features which are

compact feedback signaling and per-stream adaptive bit

loading. A compact feedback signaling is proposed to

reduce the overhead by a factor from 3.3 to 10 at the cost

of additional computations. The overhead reduction is

achieved by three means. First, the receiver feeds back

transmit beamforming vectors instead of the channel

matrix. This reduces the overhead by a factor of more

than 1.6 on average. Second, the elements of each

beamforming vector are jointly quantized by vector

quantization using three small codebooks of sizes 16, 32

and 64 respectively. The vector quantization reduces the

overhead by a factor of two compared to the scalar

quantization in current draft. Finally, the scheme feeds

back the beamforming vectors only for the active spatial

channels. This provides a significant overhead reduction

in the case of spatial channel puncture, where the spatial

channel corresponding to the weakest eigenmode is

usually punctured.

The codebook is employed in the feedback from

mobile user to base station. The mobile user learns the

channel state information from downlink and selects a

transmit beamforming matrix for the codebook. The index

of the matrix in the codebook is then fed back to the base

station. Each codebook corresponds to a combination of

Nt, Ns, and Ni, where Nt, Ns, and Ni are the numbers of BS

transmit antennas, available data streams, and bits for the

feedback index respectively. Once Nt, Ns, and Ni are

determined in the mobile user, the mobile user will feed

back the codebook indexes each of Ni bits. After receiving

a Ni bit index, the base station will look up the

corresponding codebook and select the matrix (or vector)

according to the index. There are several different types

of codebooks proposed by companies, which are antenna

grouping, Grasmmannian, Givens, Household, etc.

Feedback methods include channel matrix index,

transmit antenna index, quantized MIMO (sub) channels,

quantized SVD decomposed MIMO channel.

The difference between the greatest and the smallest

eigenvalues increases with the number of spatial streams,

and it is greater than 17 dB for 4x4. This large difference

is hard to be compensated by FEC coding and adaptive

bit/power loading is required. The exact adaptive bit (or

power) loading has the flexibility to put a different number

of bits (or amount of power) on each OFDM subcarrier

and each spatial channel. In order to reduce the overhead,

we propose per-stream adaptive bit loading as shown in

Figure 1. It assigns the same number of bits on each spatial

channel, where the i-th spatial channel is formed by the i-th

eigenmodes of each subcarrier. To further reduce the

feedback overhead, we define a set of modulation coding

schemes (MCSs), where each MCS specifies the modulations

on each stream and the FEC code rate (and suggested power

ratio across streams). The eigenvalue distributions of 4x1,

4x2, 4x3, and 4x4 are shown in Figure 2.

2. WiFi (IEEE 802.11n)

In IEEE 802.11n, there are two divided groups toward

the harmonized standardization which are TGn Sync

[4]~[5] and WWiSE [6]. Currently 802.11 Task Group n

(TGn) is in the process of standardizing the next-

generation WLAN technology to provide over 100 Mb/s

Nest Generation Mobile Communication: Near-Optimal MIMO Solutions in WiBro/WiFi/B3G Communication Standards 423

Spatial channel 1, 64 QAM, power 50%

Spatial channel 2, 16 QAM, power 40%

Spatial channel 3, BPSK, power 10%

Spatial channel 4, power 0%

Figure 1. Adaptive bit loading when #TX antennas are 4

424 Telecommunications Review·Vol. 15 No. 3·2005. 6

over 600Mbps. This complements the evolution of

modern technologies such as USB 2.0, IEEE 1394b, and

PCI Express to provide a dramatic performance upgrade

for users of current wireless designs. Adaptive Radio

Technology to intelligently use spectrum and adapt to its

expansion by worldwide regulatory bodies for unlicensed

and licensed applications. This allows products to remain

interoperable while adapting to different numbers of

spatial streams (2 to 4) as well as different amounts of

spectrum (10, 20, 40MHz). Adaptive radio is essential to

the mobile handsets, PC laptops, and other products that

only have two antennas, because it dramatically increases

their performance while functioning as an interoperable

good neighbor. Both Extended Modulation Coding

Scheme (MCS) and Basic Beamforming to increase the

speed and reliability of data links under conditions that

disrupt many MIMO networks. This enables the advanced

802.11n capabilities to be sustained over range and also

maintain full interoperability with existing 802.11a/b/g

devices. Timed Receive Mode Switching (TRMS) and

Multiple Receiver Address (MRA) Power Management

enables products to operate in extremely low power modes

and engage advanced capabilities on demand. This is

important for voice handsets, notebook computers and any

power-sensitive applications, because it lets them take full

[5]. The design of the next-generation WLAN is based on

MIMO and orthogonal frequency-division multiplexing

(OFDM). As in IEEE 802.16e, SVD based MIMO was

proposed by Qualcomm [4]. The MIMO WLAN uses

OFDM modulation in the 20 MHz band of operation as in

802.11a/g. The 802.11a/g OFDM symbol is composed of

64 subbands where a total of 48 subbands are used for data

and four as pilot.

The WWiSE technical proposal includes several

innovative techniques that enhance data rate, network

efficiency, operational range, and reliability [6]. One

unique aspect of the activities of TGn is that both MAC

and PHY changes are considered. Changes in the MAC

protocol in the WWiSE proposal are implemented

primarily to increase network efficiency and manage

network access when 40MHz optional channels are in use.

PHY enhancements are aimed primarily at increasing peak

data rates.

The TGn Sync proposal expands the appeal of 802.11n

beyond traditional Wi-Fi devices and high end products.

Important innovations include methods to reduce power

consumption for small mobile phones and increase the

user capacity of public networks. MIMO Spatial Division

Multiplexing to support data rates of up to 243Mbps in

standard two antenna designs, with extensions to support

4×1

4×2

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

0.80.60.40.2

0

4×3

0 1 2 3 4 5 6

1.5

1

0.5

0

4×4

0 1 2 3 4 5 6

1.5

1

0.5

0

0 1 2 3 4 5 6

Figure2. Eigenvalue distributions of spatial modes

2

1.51

0.5

0

2.2. Spatial Spreading (SS)

When full CSI is not available, it is desirable to

achieve maximum diversity while transmitting on some or

all spatial channels. Spatial Spreading (SS) is a

generalized space-frequency code over the OFDM

subbands. With SS, the transmitter forms the transmitted

vector x(k)=W(k)×s(k), where W(k) is the unitary SS

matrix used in OFDM subband k. The spatial spreading

matrices W(k), can be selected to provide many

independent ''looks'' at the channel over the set of OFDM

subbands. One effective set of spatial spreading matrices

that is simple to implement employs a fixed unitary

spreading matrix S followed by a linear phase shift per

transmitted stream. The transmitter ''spreads'' the NS data

streams across the N=min(NR, NT) spatial channels of the

MIMO channel using the columns of a unitary spreading

matrix S. For example, S may be a Hadamard matrix or a

Fourier matrix. The number of data streams is determined

based on SNR (or rate) feedback from the receiver. As an

example, consider the case NS=2 and NR=NT=4. Then

the SS is provided by the first two columns of the 4×4

Hadamard matrix, which ensures that both data streams

''see'' all four spatial channels. This is followed by a

''uniform'' phase shift steering matrix. Note that this

uniform phase shift can be trivially implemented by

introducing a fixed cyclic time shift in the OFDM symbol

per transmit antenna. The linear phase shift across the

OFDM subbands provides additional diversity in channels

with low dispersion.

3. WCDMA/HSDPA (3GPP)

3.1. Per Antenna Rate Control (PARC)

Lucent initially proposed their multiple antenna

solution, which is called the per-antenna rate control

advantage of high data rates to reduce the amount of time

their radios must operate. Fast radios extend battery life.

The MCS definitions and indexing for the Basic

MIMO set are found in Table 1. The same definitions are

used for both 20 and 40MHz channels. There is one

exception. MCS 32 (not listed in the table) is a BPSK rate

1/2 duplicate format transmission mode that provides a

6Mbps rate for 40MHz channels.

Qualcomm proposed transmit beamforming MIMO

schemes, which are Full CSI schemes imply the transmitter

has full knowledge of the MIMO channel (i.e., amplitude

and phase response of each OFDM subband) [4].

2.1. Eigenvector Steering (ES)

With full CSI available at the transmitter, the MIMO

channel can be decomposed into orthogonal spatial

channels commonly referred to as eigenmodes. Using the

example above, H(k) can be represented as H(k)=U(k)×

D(k)×VH(k), where U(k) and V(k) are unitary matrices

representing the left and right eigenvectors of H(k)

respectively, D(k) is a diagonal matrix of the singular

values of H(k), and VH(k) is the conjugate transpose of the

matrix. The matrices U(k), V(k), and D(k) can be

determined from H(k) using SVD. This scheme is called

the Eigenvector Spreading (ES). The larger eigenmodes

have substantially less frequency selectivity than the

smaller ones. This is significant, as it suggests that the

coded performance on the larger eigenmodes will be closer

to additive white Gaussian noise (AWGN) performance

than the underlying 1×1 channel that is highly frequency-

selective. Moreover, the code rate and modulation on the

eigenmodes that have low frequency selectivity can be

chosen based on the average SNR per wideband eigenmode

to achieve throughput performance comparable to that

achieved by adaptive bit loading (i.e., matching a code rate

and modulation per eigenmode per subband).

Nest Generation Mobile Communication: Near-Optimal MIMO Solutions in WiBro/WiFi/B3G Communication Standards 425

Modulation

BPSK

QPSK

QPSK

16 QAM

16 QAM

64 QAM

64 QAM

64 QAM

FEC rate

1/2

1/2

3/4

1/2

3/4

2/3

3/4

5/6

MCS indices: 1/2/3/4 streams

0 / 8 / 16 / 24

1 / 9 / 17 / 25

2 / 10 / 18 / 26

3 / 11 / 19 / 27

4 / 12 / 20 / 28

5 / 13 / 21 / 29

6 / 14 / 22 / 30

7 / 15 / 23 / 31

Table 1. MCS definition (TGn Sync)

was originally proposed by Texas Instruments in 3GPP,

which was compared with PARC for system performance.

DSTTD has no feedback signaling, resulting in capacity

degradation. Thus, Mitsubishi proposes the improved

version of DSTTD which is equipped with adaptive

modulations and feedback signaling for capacity

enhancement.

3.5. Multipath Diversity with Rate Control (MPD-RC)

MPD also uses spatial multiplexing with rate control

on each stream [13]. The difference is that each stream is

transmitted from two antennas with the spreading codes

differentiated by a delay of one chip interval. MPD also

uses space-time block coding as in DSTTD.

3.6. TxAA based Schemes

Nokia proposed TxAA based MIMO schemes, which

is an extension of the closed loop transmit diversity used

in Rel99 using receiver diversity [14].

3.7. TPRC for CD-SIC MIMO

Transmit power ratio control (TPRC) was proposed by

SNU & Samsung [15]. To cancel out the effect of time-

domain interference signal, the code-domain interference

canceller, e.g. the code-domain successive interference

canceller (CD-SIC), may be preferable to the time-domain

interference signal because of its good performance and

simplicity.

3.8. Multi-user MIMO Schemes

We propose a multi-user MIMO scheme [16], which is

the enhanced version of [17] where multi-user diversity

and scheduling techniques are exploited [18]~[25]. More

details are examined in the next section.

III. BLOCK MMSE-DP WITHGREEDY MCSD

1. System Model

Consider a K user wireless downlink communications

system with multiple transmit antennas at the base station,

as shown in Figure 3, and multiple receive antennas for

each user. We assume that the base station has t transmit

antennas, the user k has rk receive antennas, and the

426 Telecommunications Review·Vol. 15 No. 3·2005. 6

(PARC) [8], in 3GPP MIMO technical report (TR) [7].

The transmitter of PARC is similar to the structure shown

in Figure 1, in which separately encoded data streams are

transmitted from each antenna with equal power but

possibly with different data rates while spreading code is

reused through all streams. The data rates for each

antenna are controlled by adaptively allocating transmit

resources such as modulation order, code rate, and number

of spreading codes. The post-decoding signal-to-

interference-plus-noise ratio (SINR) of each transmit

antenna is estimated at the receiver and then fed back to

the transmitter, which is used to determine the data rate on

each antenna. The vector signaling with more feedback

overhead over the scalar signaling in conventional systems

is required for link adaptation.

3.2. Selective PARC (S-PARC)

The selective PARC (S-PARC) has been proposed by

Ericsson, which is conceptually based on PARC scheme in

the previous subsection [9]. In S-PARC, selection

diversity is combined together with PARC by controlling

transmit antenna configurations with adaptive resource

allocations. Recent results have shown that PARC

achieves the full open-loop capacity of the flat fading

MIMO channel. However, there is a significant gap

between the open-loop capacity and the closed-loop

capacity, when signal-to-noise ratio (SNR) is low and/or

the number of receive antennas is less than the number of

transmit antennas. An approach to achieve the near-

capacity of the closed-loop MIMO is S-PARC, which

compensates for the capacity loss by the gain of antenna

selection.

3.3. Per Stream Rate Control (PSRC)

To enhance the performance of PARC, the unitary

precoding based spatial multiplexing scheme has been

proposed, which is the combined technique of PARC and

transmit adaptive array (TxAA), called the per-stream rate

control (PSRC) [10]. Given a precoding matrix,

modulation size and code rate are selected to maximize the

total throughput. Note that when only one data stream is

transmitted, PSRC is reduced to TxAA.

3.4. Double STTD with Sub Group Rate Control (DSTTD-SGRC)

DSTTD with SGRC [11] was proposed by Mitsubishi

in 3GPP, noting that DSTTD without considering SGRC

had been proposed by TI [12]. On the other hand, DSTTD

number of all receive antennas in the system is r=ΣKk=1

rk. Also, we model the channel as a frequency-flat block

fading channel. Interference from neighboring cells is

modeled as additive Gaussian noise, as we concentrate on

the single cell model. The received signal of user k is

expressed as

yk=Hkx+nk

where the tx1 input signal vector x is transmitted by

the base station and is constrained to have power no

greater than a sum-power constraint P, i.e., tr(E[xxH])≤

P, and the tx1 vector zk represents the random additive

noise for user k where zk~CN (0,I). The channel Hk is a

rkxt matrix, whose entries are assumed to be independent

and identically distributed (i.i.d.) circularly symmetric

complex Gaussian random variables with zero-mean and

unit variance. Also, Hk is independent of Hj for all j≠k.

In general, it is difficult for the base station to have the

perfect knowledge of downlink channel state information

(CSI) because the feedback link has delayed lossy

feedback characteristics. Hence, the problem at hand is to

find the transmit and receive structure that minimizes the

feedback rate subject to the performance constraint such

that the data throughput is kept as close as possible to the

sum capacity.

2. Block QR Decomposition

We propose a multi-user MIMO scheme that is based

on unitary beamforming and user selection diversity. It is

Nest Generation Mobile Communication: Near-Optimal MIMO Solutions in WiBro/WiFi/B3G Communication Standards 427

assumed that t is the number of transmit antennas, r is the

number of receive antennas, and K is the number of users.

Beamforming using unitary transformation matrix W that

is a function of the channel unitary matrices fed back from

users is employed at the transmitter. The channel unitary

matrix for feedback denotes the right-most matrix Vkobtained by SVD of the kth user channel Hk=UkDkVk

H.

Each data stream for transmission is allocated to each

beam vector of the unitary transform matrix, and the

transmitter adjusts antenna rates independently. In the

proposed system the channel is rotated using the right

unitary matrix obtained by SVD of the each user channel,

so as to reduce feedback overhead at the transmitter.

MIMO channel is decomposed into multiple parallel

MISO channels Fk, which is referred to as the effective

channel

Fk=UkHHk=DkVk

H

The row of the effective channel matrix Fk is also

noted as the effective channel vector. In the transmitter,

controlled beamforming is implemented by applying QR

decomposition to the combination of the effective

channels F=[F1T, ..., FK

T]T. The effective BC F can then

be treated as the multi-user MISO channel matrix. As in

the algorithm of [6] for MISO, the QR decomposition is

obtained using the Gram-Schmidt orthogonalization

procedure to the rows of F. That is, geometrical

projection is performed based on SVD decomposition, and

then the finite dimensional subspace is determined by QR

process. Using QR decomposition, the effective BC is

Data streams

User1 Encoder/

Modulator

Encoder/

Modulator

Feedback

Controller

Figure 3. Schematic of the transmitter for the proposed scheme

Feedback

Information

Unitary

PrecoderUser/Rate

Selector

Userk

428 Telecommunications Review·Vol. 15 No. 3·2005. 6

represented as F=RW, where R is a r x t lower triangular

matrix and W is a t x t matrix with orthonormal rows. The

unitary matrix WH is used for beamforming, and hence is

applied to the transmitted signal

y=Fx+z

=RWWHs+z

=Rs+z

where y=[y1T, ..., yK

T]T and z=[z1T, ..., zK

T]T. The sum-rate

performance based on block QR decomposition is

maximized by adopting MCSD which is described in the

next subsection.

3. Multi-Channel Selection Diversity

Multi-user diversity is the promising solution to

improve capacity gain while Costa precoding is the

capacity-achieving strategy in MIMO BCs. In our

proposed scheme, multi-channel based selective diversity

(i.e., MCSD) is exploited in combination with Costa

precoding for known interference cancellation, which

means that the channel vectors of active users are selected

and ordered to achieve diversity gain with the increase of

the number of users and antennas therein, and interference

cancellation using Costa precoding is processed at the

transmitter to approach maximum sum-rate.

Let S⊂{1, ..., r} be a subset of the effective channel

vector indices that the BS selects for transmission using

MCSD, and F(S)=[f1T(S), ... , f|S|

T(S)]T be the

corresponding submatrix of F . The t x t unitary

beamforming matrix WH(S) is obtained by QR

decomposition of the submatrix such that F(S)=

R(S)W(S), where W(S)=[w1T(S), ..., w|S|

T(S)]T and wi(S)

is a 1 x t vector. Then, the achievable sum-rate of this

system by Costa precoding is given by

PR≅max Σ log(1+------------|fi(S)w1

H(S)|2)S i∈S |S| ,

K≤ max log|I+Σ Hk

HQkHk|Σk tr(Qk)≤P, Qk≥0 k=1

where each of the matrices Qk is an rk x rk positive semi-

definite covariance matrix. The selection process is partly

performed in mobile users such that they select and feed

back l active channels corresponding to the l largest

eigenmodes, which reduces the feedback amount by a

factor of l. The upper bound is the sum capacity of the

MIMO BC as described above and the bound is achievable

when the power P goes to infinity and the number of

receive antennas is one for all receivers.

4. Candidate Schemes for Comparison

The sum-rate maximization can be solved efficiently

by using SP-IWF, which achieves the sum capacity of a

MIMO BC. On the other hand, time-division multiple-

access (TDMA), where the BS transmits to only a single

user at a time by using all transmit antennas, is a

suboptimal solution when the BS has multiple transmit

antennas, called TDMA-MIMO, while it achieves the sum

capacity with only one transmit antenna. It is then shown

that the maximum sum-rate of TDMA-MIMO is the

largest single-user capacity of the K users, which is given

by

CTDMA-MIMO= max C(Hi, P)i=1, ..., K

where C(Hi, P) denotes the single-user capacity of the i-th

user subject to power constraint P.

IV. PERFORMANCE ANALYSIS

In this section, the performance analysis is presented.

We remind that the entries of {Hk} are assumed to be i.i.d.

zero-mean complex-Gaussian random variables. The

proofs of the following lemmas and theorems are

presented in [10].

Theorem 1 (Optimizing transmit covariance matrix)

The objective of the transmit covariance matrix design is

to find a covariance matrix set that maximizes the system

throughput, subject to the sum power constraint and the

unknown-interference free constraint. The transmit

covariance matrix satisfying this objective is obtained by

QR decomposition of F.

Lemma 1 We assume that user k is not allowed to

know CSI of all other users. That is, any information

related to this CSI is not delivered from the transmitter as

well as not exchanged between users. In this case, the

Nest Generation Mobile Communication: Near-Optimal MIMO Solutions in WiBro/WiFi/B3G Communication Standards 429

optimal processing for user k is SVD-based (single-user)

water-filling, in which the receive beamforming is

performed with the left unitary matrix of the user k's

channel.

Lemma 2 We consider a user that performs receive

beamforming by the left unitary matrix of the

corresponding channel. The average throughput of a

MIMO BC with the user is no worse than the performance

obtained based on non-cooperative reception across

antennas, e.g., MMSE-DP.

Theorem 2 Receive beamforming with the left

singular matrix offers the average throughput that is no

worse than any fixed unitary matrix beam scheme.

V. NUMERICAL RESULTS

In this section, numerical results are presented. In

18

16

14

12

10

8

6

4

Sum rate (bps/Hz)

1 2 3 4 5 6 7 8 9 10

Number of users

Figure 4. Ergodic sum-rate comparison when t=4 and r=2

SP-IWF (full HH)

Novel scheme (I=2)

Novel scheme (I=1)

MMSE-DP (full HH)

TDMA-MIMO(I=2)

TDMA-MIMO(I=1)

20

18

16

14

12

10

8

6

Sum rate (bps/Hz)

1 2 3 4 5 6 7 8 9 10

Number of users

Figure 5. Ergodic sum-rate comparison when t=4 and r=4

SP-IWF (full HH)

Novel scheme (I=4)

Novel scheme (I=1)

MMSE-DP (full HH)

TDMA-MIMO(I=4)

TDMA-MIMO(I=1)

VI. CONCLUSIONS

In this paper, we have proposed a multiuser MIMO

transmission scheme that is efficient in terms of

computational complexity and feedback overhead while

obtaining near the maximum sum-rate of BC. Our novel

scheme has employed the block QR decomposition at the

transmitter, which reduces the computational complexity

to design transmit covariance matrices. Using MCSD in

combination with known interference cancellation (Costa

precoding), the proposed scheme with partial channel

information at the transmitter has shown to still achieve

the near-optimal sum capacity, which was not observed in

TDMA-MIMO. Numerical results have shown that the

gain of sum-rate is 2bps/Hz over the conventional MMSE-

DP scheme with full channel feedback and the gap from

SP-IWF is 0.4bps/Hz.

ACKNOWLEDGMENTThe authors would like to thank Hyeon Woo Lee, Juho

Lee, and Jin-Kyu Han for their comments, as well as Lab

director, Seung-yong Park. This paper has been supported

in part by the Samsung Advanced Institute of Technology

(SAIT) and in part by National Research Laboratory

(NRL) program.

[REFERENCES][1] G. J. Foschini, ''Layered space-time architecture for

wireless communication in a fading environment when using multiple antennas,'' Bell Labs Syst. Tech. J., Vol. 1, Autumn 1996, pp. 41-59.

[2] I. E. Telatar, ''Capacity of multi antenna Gaussian channels,'' tech. rep., AT&T Bell Laboratories Internal Technical Memorandum 1995, published in the European Transactions on Telecommunications, Nov/Dec 1999.

[3] IEEE P802.16-REVd/D5-2004 Draft IEEE Standards for local and metropolitan area networks part 16: Air interface for fixed broadband wireless access systems

[4] S. Nanda, R. Walton, J. Ketchum, M. Wallace, and S. Howard, "A High-Performance MIMO OFDM Wireless LAN," IEEE Communications Magazine, Feb. 2005, pp. 101-109

[5] IEEE 802.11-04/891r5, TGnSync Proposal PHY Results

[6] IEEE 802.11-05/0149r1, High throughput extension to the 802.11 standard (WWiSE)

Figures 4 and 5, we compare the ergodic sum-rate

performance of different MIMO downlink strategies. The

signal-to-noise ratio (SNR) is assumed to be 10dB. Given

the number of users, TDMA-MIMO achieves the

maximum sum-rate corresponding to the largest single-

user capacity, which shows relatively a small gain in

proportion to the number of users. When the number of

the active channel vectors is equal to the number of the

effective channel vectors and one user is assumed, the

performance of the proposed novel scheme is the same as

that of TDMA-MIMO since in both cases receivers feed

back the effective channel matrix Fk=DkVkH, instead of

the full channel matrix Hk. As the number of users

becomes large enough, the performance of the novel

scheme approaches close to the sum capacity, which can

be driven by SP-IWF. Both figures show sum-rate

improvement of 2bps/Hz over MMSE dirty paper

(MMSE-DP) scheme with full channel feedback and a gap

of 0.4bps/Hz from SP-IWF, in which MMSE-DP scheme

exploits Costa precoding based on MMSE QR

decomposition modified slightly from Caire's zero forcing

dirty paper (ZF-DP) coding in [2].

In our proposed scheme, different feedback scenarios

are examined. In Figure 4, each user has two eigenmodes,

i.e., two effective channel vectors, available since four

transmit and two receive antennas are assumed. The sum-

rate of the novel scheme with feedback of one active

channel vector (one eigenvector multiplied by the

corresponding eigenvalue that is the largest one) gets

tightly close to the performance having feedback of two

active channel vectors when the number of users is five.

Contrastingly, TDMA-MIMO with one vector never gets

close to TDMA-MIMO with two vector. Four transmit

and four receive antennas are considered in Figure 5,

where two feedback signaling (i.e., one, four active

channel vectors) are examined for the novel and TDMA-

MIMO schemes. Both figures show that the novel scheme

with reduced feedback, i.e., with the fewer active channel

vectors, achieves slightly lower rate performance with

small number of users compared to the scheme with full

effective channel vector. However, the performance

approaches extremely close to the upper bound as the

number of users increases. Therefore, in the proposed

scheme feedback of active channel vectors is shown to

have the equivalent sum-rate performance with feedback

of full effective channel vectors, resulting in the

outstanding feedback robustness. That is, the feedback

signaling per user can be significantly reduced with the

increase of the number of users.

430 Telecommunications Review·Vol. 15 No. 3·2005. 6

A. Goldsmith, ''Sum power iterative water-filling for multi-antenna Gaussian broadcast channels,'' Submitted to IEEE Trans. on Information Theory, Jul., 2004.

[22] W. Yu, ''A dual decomposition approach to the sum power Gaussian vector multiple access channel sum capacity problem,'' in Conference on Information Sciences and Systems (CISS), Mar. 2003.

[23] Z. Tu and R. S. Blum, ''Multiuser diversity for a dirty paper approach,'' IEEE Commun. Lett., Vol. 7, Aug 2003, pp. 370-372.

[24] M. Airy, A. Forenza, R. W. Heath, Jr., and S. Shakkottai, ''Practical Costa precoding for the multiple antenna broadcast channel,'' Accepted to Proc. IEEE GLOBECOM 2004.

[25] D. J. Love, R. W. Heath Jr., W. Santipach, and M. L. Honig, ''What is the value of limited feedback for mimo channels?'' IEEE Commun. Mag., Vol. 42, No. 10, Oct. 2004, pp. 54-59.

[26] M. Sharif and B. Hassibi, ''On the capacity of mimo broadcast channel with partial CSI,'' in Proc. Asilomar conf., Pacific Grove, CA, Nov. 2003.

Nest Generation Mobile Communication: Near-Optimal MIMO Solutions in WiBro/WiFi/B3G Communication Standards 431

[7] 3GPP TR 25.876 V 1.7.0, 3GPP TSG RAN Multiple input multiple output in UTRA

[8] S. T. Chung, A. Lozano, H. C. Huang, ''Approaching eigenmode BLAST channel capacity using V-BLAST with rate and power feedback,'' in Proc. of VTC, Atlantic City, NJ USA, Oct. 2001, pp. 915-919.

[9] Ericsson, ''Selective Per Antenna Rate Control (S-PARC),'' 3GPP TSG-R WG1, R1#36 (04)0307, Malaga, Spain, 16th - 20th Feb. 2004.

[10] Lucent, "Per stream rate control (PSRC) with Code Reuse TxAA and APP Decoding for HSDPA," in 3GPP R1(02)0570, Apr. 2002.

[11] Mitsubishi, ''DSTTD-SGRC text proposal for TR 25.876,'' 3GPP TSG-R WG1, R1#36 (04)0290, Malaga, Spain, 16th - 20th Feb. 2004.

[12] Texas Instruments (TI), ''Double-STTD scheme for HSDPA systems with four transmit antennas: Link Level Simulation Results,'' 3GPP TSG-R WG1, TSGR1#20(01)0458, Busan, Korea, 21st - 24th May 2001. LG, ''Double TxAA for MIMO,'' 3GPP TSG-R WG1, R1#36 (04)0222, Malaga, Spain, 16th - 20th Feb, 2004.

[13] Nortel, ''System level simulations for RC-MPD,'' 3GPP TSG-R WG1, R1#36 (04)0186, Malaga, Spain, 16th - 20th Feb. 2004.

[14] Nokia, ''Closed Loop MIMO with 4 Tx and 2 Rx antennas,'' 3GPP TSG-R WG1, R1#36 (04)0206, Malaga, Spain, 16th - 20th Feb. 2004..

[15] SNU and Samsung, ''Text proposal for TPRC for CD-SIC MIMO,'' 3GPP TSG-R WG1, R1#36 (04)04255, Malaga, Spain, 16th - 20th Feb. 2004.

[16] James Sungjin Kim, Hojin Kim, K. B. Lee ''Efficient Feedback Transmission for Multi-User MIMO Systems'' to appear in IST 2005 summit conf.

[17] Samsung and SNU, ''PU2RC Simulation Considering SPARC and 4TxAA mode1 Signalling,'' 3GPP TSG-R WG1, R1#36 (04)0362, Malaga, Spain, 16th - 20th Feb. 2004.

[18] H. Sato, ''An outer bound on the capacity region of the broadcast channel,'' IEEE Trans. Inform. Theory, Vol. 24, May. 1978, pp. 374-3778.

[19] G. Caire and S. Shamai, ''On the achievable throughput of a multiantenna gaussian broadcast channel,'' IEEE Trans. Inform. Theory, Vol. 49, No. 7, Jul. 2003, pp. 1691-1706.

[20] W. Yu, W. Rhee, S. Boyd, and J. Cioffi, ''Iterative water-filling for gaussian vector multiple access channels,'' IEEE Trans. Inform. Theory, Vol. 50, No. 1, Jan. 2004, pp. 145-151.

[21] N. Jindal, W. Rhee, S. Vishwanath, S. Jafar, and

432 Telecommunications Review·Vol. 15 No. 3·2005. 6

Kiho Kim

Kiho Kim obtained his bachelor’s degree in Electronics

and Communications Engineering from the College of

Engineering, Hanyang University, Korea in 1980, his

master’s degree from KAIST in 1982 and his PhD from

University of Texas at Austin in 1991. He worked at KBS

technology center from 1982 to 1987. Since 1991, he has

been with Samsung advanced institute of technology as

vice president. His interests include signal processing and

wireless communications.

E-mail: kihokim@samsung.com

Fax.: 82-31-280-9569

Tel:+82-31-280-9220

Sungjin Kim

Sungjin (James) Kim was born in Korea in 1969. He

obtained his Bachelor and Master of Engineering degree in

Electronics and Communications Engineering from the

College of Engineering, Hanyang University, Korea in

1994 and in 2000, respectively. He is now pursuing his

Doctor of Philosophy in Electrical and Computer

Engineering from the College of Engineering, Seoul

National University. In February 1994 he joined Samsung

Advanced Institute of Technology, and he is now a senior

member of technical research staff. Since 1999, he has

been the Editor-in-Chief of 3GPP (WCDMA standard)

Transmit Diversity TR. His research interests include the

areas of transmit diversity (TxD), multiple-input and

multiple-output (MIMO), wireless scheduling and

adaptive signal processing for 3G+/4G wireless

communications.

E-mail: kimsj@mobile.snu.ac.kr

Tel.:+82-31-280-9222

Fax.:+82-31-280-9569

Hojin Kim

Hojin Kim was born in Korea in 1973. He obtained his

Bachelor of Science in Electrical and Computer

Engineering from Purdue University, Indiana in 1997. He

received his Master of Science from the Electrical and

Computer Engineering at the University of Florida, Florida in

2000. In 2000, he was with LG electronics institute of

technology as a research engineer. Since 2001, he has been a

research engineer at Samsung advanced institute of

technology. His research interests include MIMO, OFDM,

Ad-hoc network, and 3GPP standardization.

E-mail: hkim73@samsung.com

Tel.:+82-31-280-9222

Fax.:+82-31-280-9569

Nest Generation Mobile Communication: Near-Optimal MIMO Solutions in WiBro/WiFi/B3G Communication Standards 433

Kwang Bok Lee

Kwang Bok Lee received the B.A.Sc. and M.Eng. degrees

from the University of Toronto, Toronto, Ont., Canada, in

1982 and 1986, respectively, and the Ph.D. degree from

McMaster University, Canada in 1990. He was with

Motorola Canada from 1982 to 1985, and Motorola USA

from 1990 to 1996 as a Senior Staff Engineer. At

Motorola, he was involved in the research and

development of wireless communication systems. He was

with Bell-Northern Research, Canada, from 1989 to 1990.

In March 1996, he joined the School of Electrical

Engineering, Seoul National University, Seoul, Korea.

Currently he is an Associate Professor in the School of

Electrical Engineering. He was a Vice Chair of the School

of Electrical Engineering from 2000 to 2002. He has been

serving as a Consultant to a number of wireless industries.

Since 2003, he has been a senior member of the IEEE. His

research interests include mobile communications,

communication technique covering physical layer and

upper layer. He holds ten U.S. patents and four Korean

patents, and has a number of patents pending.

Dr. Lee was an Editor of the IEEE JOURNAL ON

SELECTED AREAS IN COMMUNICATIONS, Wireless

Series in 2001, and has been an Editor of the IEEE

TRANSACTIONS ON WIRELESS COMMUNICATIONS

since 2002. And he is a co-chair of the ICC2005 Wireless

Communication Symposium. He received the Best Paper

Award from CDMA International Conference 2000 (CIC

2000), and the Best Teacher Award in 2003 from College

of engineering, Seoul National University.

E-mail: klee@snu.ac.kr

Tel.:+82-31-880-8415

Fax.:+82-31-880-8215