10 th MCM - Novi Sad, 23-24 March 2006 Joaquim Bastos ( [email protected] ) 1 The information in this document is provided as is and no guarantee or warranty

10th MCM - Novi Sad, 23-24 March 2006 Joaquim Bastos ( [email protected] ) 1

The information in this document is provided as is and no guarantee or warranty is given that the information is fit for any particular purpose. The user thereof uses the information as its sole risk and liability.

Equalization Schemes for Extended Alamouti Codes in

MC-CDMA Systems

J. Bastos and A. Gameiro

Instituto de Telecomunicações



IST 4MORE Project

The main objective of IST 4MORE project is to implement a cost effective low power System on Chip (SoC) solution for a 4G mobile terminal employing multiple antennas, based on MC-CDMA techniques. 4MORE is a joint European project involving:

IETR

http://www.ist-4more.org



Overview

Introduction

Double Alamouti

System model

Decoding and Equalization

Simulations setup

Results

Conclusion



Introduction (I)

CDMA benefits from SS and reusable single frequency.

OFDM robustness against frequency selective fading.

MC-CDMA considered strong candidate for B3G systems.

Space-Time coding enhances high capacity of MIMO systems.

MIMO schemes are key solution for capacity limitation imposed by MAI on CDMA based systems. The popular Alamouti scheme

benefits both from space and time diversity, at the expense of moderate complexity.

Promising combination for emerging mobile

communication systems.



Space-Time coding is used to combat fading by exploiting diversity.

The popular Alamouti STBC (2 Tx, 1 Rx) can be extended through replication, allowing to reduce the constellation size of the modulation scheme used by the system while keeping the same data throughput:

– Lower constellation sizes will allow better tolerance to phase errors and automatic gain compensation;

– The data throughput of a system using Double Alamouti (4 Tx, 2 Rx) and QPSK modulation is the same as for a system implementing standard Alamouti, using 16-QAM modulation;

– Spectral efficiency can be kept while reducing constellation size, at the expense of a higher number of elements in both Tx and Rx antenna arrays;

– On the other hand, keeping the same constellation size allows to have twice the data throughput, without needing further bandwidth.

Introduction (II)



Double Alamouti scheme (4x2)

STC 1 Alamouti

s1 s2

STC 2 Alamouti

s3 s4

s1 -s2*

s2 s1*

t0 t1

s3 -s4*

s4 s3*

ATx1

ATx2

ATx3

ATx4

ARx2

Double Alamouti Decoding w/ ZF or

MMSE criteria

ARx1

s1 s2 s3 s4r1,t0 r1,t1

r2,t0 r2,t1

si

r1,t0 =s1h1,1(t)+s2h2,1(t)+s3h3,1(t)+s4h4,1(t)+n1(t)

r1,t1 =-s2*h1,1(t+1)+s1*h2,1(t+1)-s4*h3,1(t+1)+s3*h4,1(t+1)+n1(t+1)

r2,t0 =s1h1,2(t)+s2h2,2(t)+s3h3,2(t)+s4h4,2(t)+n2(t)

r2,t1 =-s2*h1,2(t+1)+s1*h2,2(t+1)-s4*h3,2(t+1)+s3*h4,2(t+1)+n2(t+1)

HTx,Rx

Received signals after MIMO channel



System model

This simulation chain represents the system model for the Downlink:– Convolutional encoding using UMTS code (Rc = ½);

– Channel interleaving according to UMTS procedures;

– MIMO channel implemented according 3GPP Spatial Channel Model;

– STBC is achieved by MIMO Encoding, Decoding and Equalization.



Double Alamouti decoding (I)

At the receiver, the two signals need to be decoded, and for that, they must be combined taking into account the Double Alamouti scheme.

Considering the following expression, we implemented two ways to obtain the appropriate estimates of the transmitted symbols. These were based on ZF and MMSE criteria, respectively.

ZF:

MMSE:

RCS .~ 1

RICCCS cHH ..

~ 12

NSCR .

4

3

2

1

4

3

2

1

*2,3

*2,4

*2,1

*2,2

*1,3

*1,4

*1,1

*1,2

2,42,32,22,1

1,41,31,21,1

*

*.

1,2

1,1

0,2

0,1

n

n

n

n

s

s

s

s

hhhh

hhhh

hhhh

hhhh

r

r

r

r

t

t

t

t



Double Alamouti decoding (II)

After decoding operation it is still necessary to perform symbol equalization given that the chips, referring to each transmitted symbol, were each exposed to dissimilar conditions as the propagation channel is not flat-fading.

Filtering is necessary to equalize the received OFDM signal, demodulated and STBC decoded. There are several choices for non-linear equalization. We propose in this work EGC and MRC:

– EGC: gi = wii* / |wii| ; MRC: gi = wii

* /σceqTi .

– W = T · C(T is the transformation matrix used earlier in decoding:

• ZF: T = C-1 ; MMSE: T = CH

(C . CH + σc

2 I)-1 .)

– Total variance per subcarrier: (assuming K is high)

4

1

224

1

222

ijj

ijaj

ijcc wKteqTi



Simulations setup

Carrier Frequency 5.2 GHz

Bandwidth 61.44 MHz

Spreading Factor 32 (Walsh-Hadamard codes)

Number of Subcarriers 1024 (672 used)

OFDM Symbol Duration 21.5 s (24 OFDM Symbols per frame)

Guard Period 4.2 s

Antenna Spacing BS: 10 MT: /2

Number of Antenna Elements DAl.: 4 Tx, 2 Rx SAl.: 2 Tx, 2 Rx

Modulation (constellation size) DAl.: QPSK (2) SAl.: 16-QAM (4)

Coding Rate 0.5

Mobile Terminal Speed 60 km/h

ChannelModel

Spatial 3GPP Urban Macro

Time ETSI BRAN E



Results



Conclusion

An extension of the Alamouti scheme was proposed in this work, and its performance was compared with the original scheme. In order to compare them properly, similar spectral efficiencies were considered.

The attained results show that Double Alamouti scheme associated with QPSK, MMSE based decoding and EGC or MRC equalization provide quite similar performance as Single Alamouti with 16-QAM and MMSE equalization.

Using ZF based decoding associated with any proposed equalization techniques provides better performance than Single Alamouti with MRC equalization, only when the existing SNR is higher than 5 dB.

The proposed scheme should be an interesting option when its extra complexity isn’t a setback for system accomplishment. It can allow the system to use more tolerant modulation schemes, like QPSK, while preserving data throughput, bandwidth and performance.



Documents

10 th MCM - Novi Sad, 23-24 March 2006 Joaquim Bastos ( [email protected] ) 1 The information in this document is provided as is and no guarantee or warranty