You are on page 1of 4

Kuldeep Kumar* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES Vol No. 7, Issue No.

2, 301 - 304

A comparison of different detection algorithms in a MIMO system


Kuldeep Kumar Rayat Institutes Of Engg. and IT, Ropar, Punjab Email: kuldeep_heer@yahoo.com Manwinder Singh Rayat Institutes Of Engg. and IT, Ropar. Punjab Email: Singh.manwinder@gmail.com Multiple-input multiple-output (MIMO) systems in known to offer both high data rates and high link reliability for a wireless communications system due to the space diversity. Spatial multiplexing MIMO systems also provide a higher data rates as each transmit antenna can emits an independent information symbol in the different time slots given , i.e with Nt be the transmit antennas, the data rate can has been increased by a factor of Nt. On the other hand, the achievable diversity order (which is the negative exponent of error rate at high SNR) in MIMO depends on the different detection technique used at the MIMO receiver. The better detector that minimizes the bit error probability is the maximumlikelihood (ML) detector [1]. But, the ML-detector is practically difficult as it has computational complexity is exponential. Different algorithms that are generally known as sphere decoders has been developed to get near ML performance with polynomial complexity [2]. On the other end this complexity spectrum the equalizer based on the MIMO detection schemes can be found. We discuss here the zero-forcing (ZF) detector [3] and the minimum-mean-square error (MMSE) detector [4]. The ZF-detector and the MMSE detector have lesser computational calculations as they require only a matrix operations to be carried out , e.g. pseudo-inverse. However, the error performance in case of both ZF and MMSE detectors are greatly lower than the optimal ML detector. It is to be noted that both ZF-detector and MMSE-detector have the diversity of the order of Nr Nt +1, and the optimal ML-detector has the diversity order of Nr, where Nr is given as the number of receive antennas. Even the integrated-circuit technology are available in today world to implement the most complex MIMO detection detectors such as ML-detector (It uses sphere decoding) in a chip, but the actual integration cost of these detectors is not considered and is out of scope for most of the expected applications of MIMO systems.

1. INTRODUCTION

IJ
ISSN: 2230-7818

It has been shown that Multiple-Input Multiple-Output (MIMO) systems is used to support higher data rate in the MIMO system we keep the transmit power budget and performance requirement same as compared to a SingleInput Single-Output (SISO) systems. The MIMO version of the ZF receiver acts same as to a ZF equalizer channels. The MIMO channel at the receiver is inverted in order to totally minimize the interference from other transmitted signals. The output of the ZF filter is only the function of the data to be detected and the noise at the receiver . It is then fed to a ML detector which try to approximates the transmitted symbol. The complexity of ZF decoder is same as of SISO ML decoding, but the step to invert is responsible for the noise enhancement.

Bit Error Rate(BER) ,Zero Forcing, Maximum likelihood ,minimum mean square error, Multi Input Multi Output (MIMO).
Key Words

ES

In this paper, we propose a the different detection scheme for MIMO systems such as maximum-likelihood (ML) detection of MIMO signals which requires computational resources that are above from the capabilities of most practical systems. Alternative in order to reduce the complexity MIMO detection techniques are proposed, but the complexity of algorithmic schemes are in higher than that of the equalizer-based techniques such as zero-forcing (ZF) and MMSE. The equalizer based solution for detection have a relatively poor performance in terms of error rate. In this paper, we compare the BER for the following detection scheme MMSE,ML and ZF detection technique for a 4x4 MIMO system. Also we compare these detection schemes for a 1x4,2x4,3x4 ,4x4MIMO system.

Abstract

1
@ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 301

Kuldeep Kumar* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES Vol No. 7, Issue No. 2, 301 - 304

Thus the race is still on to get a MIMO detector system with a near ML error performance while having a very low complexity close to ZF-detector. In this context, we will discuss the performance of three detectors namely MMSE , ML and ZF detectors . We would focus our discussion to the experimental results carried out to different MIMO systems and then try to analyze which of the detectors have a better performance in terms of BER for a given SNR. The article is organized as follows. In section 2 MIMO we describe the details the based MIMO detector. section 3, we discuss the results and performance comparisons of different receivers and section 4, we conclude our discussion.

channel is inverted when the signal is at the receiver in order to totally suppress the signal interference from other transmitted symbols data. The output of the ZF filter is the combination of function of the symbol to be detected and the noise. The signal is then fed into a ML decoder which estimates the transmitted symbol data. The ZF decoding is similar to SISO ML decoding, but there is a inversion step responsible for the noise enhancement. Assuming that a symbol vector is given by

C 1/ nt [c1...cnt ]T

(1)

2. MIMO DETECTORS
The reception of data that is transmitted serially over the dispersive medium is complex as the presence of channel-induced intersymbol interference (ISI) is there .There are many techniques for reducing the complexity for the optimal reception of the receiver .The most Common schemes that we employ this technique include the orthogonal frequency division multiplexing (OFDM) and the discrete multi tone (DMT) modulation, both employ linear block-by-block transmission of data and block-by-block reception of the data .Now a days a general class of linear block-by-block transmission schemes have been found to have considerable attention [1, 2]. The, linear block precoders is used which minimize the mean square error (MSE) of the data .Zero forcing (ZF) , minimum mean square error (MMSE) equalization and Maximum likely hood strategies are discussed . We have seen that the designs that result in minimum MSE, they do not have necessarily result in having the minimum bit error rate (BER). Following are the results of the BER discussed for the different equalizers using ZF equalization, we also discuss herein linear block precoders for systems with (linear) MMSE equalizers and the threshold detection system which then minimize the BER as the size of the transmitted block of data grows.

T
Gyk [c1...cnt ]T
2

If C is the transmitted signal then the output of the ZF detector is given by

Z k GZF yk [c1...cnt ]T GZF nk

(2)

ES
GMMSE

where GZF function is to invert the channel .The ZF receiver also removes the interference but enhances noise. This is not be significant at high SNR, but at low SNR, it is both used and practical to design a filter in order to maximizing the global signal to noise plus interference ratio (SNIR). One of the possibility is to minimize the total resulting noise, we need to find a value of G so as (3)

Is minimum . The result for the solution is

nt n ( H kH H k t I nt ) 1 H kH Es

(4)

IJ
ISSN: 2230-7818

At low SNR, the MMSE filtering reduces to matched filtering model .This in turn reduce the noise and outperforms ZF filtering. It is also optimal from a capacity perspective. Also the interference is much smaller than the noise in MMSE. The maximum rate is naturally achieved by matched filtering. At high SNR, the MMSE filter is also equivalent to ZF and the diversity that can be achieved is thus limited to n r nt 1 . In Maximum-likelihood (ML) decoding scheme is assumed when we the deriving spacetime code designs. With the instantaneous channel realizations are perfectly known at the receive side of the signal, the ML decoder calculates an estimate of the transmitted signal according to

3. MIMO BER FOR DIFFERENT DETECTORS


The MIMO version of a ZF receiver is similar to a ZF equalizer in frequency selective channels. In this MIMO

2
@ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 302

Kuldeep Kumar* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES Vol No. 7, Issue No. 2, 301 - 304

C arg min yk Es H k ck
k 0

T 1

(5)

10

BER of MMSE in a MIMO system 1x4 MIMO 2x4 MIMO 3x4 MIMO 4x4 MIMO

10

T
-5

Where the minimization is done over all possible codeword vectors C. When we transmit the codeword C through nt antennas, we are interested in the probability that the ML decoder decodes the codeword properly. This probability is called the pairwise error probability (PEP) and is studied as a measure of error performance.After having discussed in brief the different detection models we discuss the result of these detection models graphically. As per the results in Fig.1,2, 3 the BER of a MIMO system decreases and is better for ML detector model. Following figures are the result that we get for the BER of different MIMO system then we specifically compare the result of different detectors for a 4X4 MIMO system by varying the SNR.
0

10

-1

10
BER

-2

10

-3

10

-4

10 12 Eb/No(db)

14

16

18

20

BER of ZF in a MIMO system

10

10

-1

ES
1x4 MIMO 2x4 MIMO 3x4 MIMO 4x4 MIMO
10
0

Figure 2. BER for different MIMO system with ZF detection.


BER of ML in a MIMO system 1x4 MIMO 2x4 MIMO 3x4 MIMO 4x4 MIMO

10
BER

-2

10

-1

10

A
0 2 4 6 8 10 12 Eb/No(db) 14 16 18 20

-3

10
BER

-2

10

-4

10

-3

IJ
10
-5

10

-4

Figure 1. BER for different MIMO system with ZF detection.

10

-5

10 12 Eb/No(db)

14

16

18

20

Figure 3. BER for different MIMO system with ML detection.

3
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 303

Kuldeep Kumar* et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES Vol No. 7, Issue No. 2, 301 - 304

10

BER of a 4x4 MIMO system 4x4 MIMO ZF 4x4 MIMO MMSE 4x4 MIMO ML

[3]

G. J. Foschini, K. Karakayali, and R. A. Valenzuela, Coordinating multiple antenna cellular networks to achieve enormous spectral efficiency, IEE Proceedings: communications, vol. 153, no. 4, pp. 548555, 2006. Z. Guo and P. Nilsson, Algorithm and implementation of the k-best sphere decoding for MIMO detection, IEEE J. Sel. Areas Common., vol. 24, no. 3, pp. 491C503, Mar. 2006. S. Simon and A. Moustakas. Optimizing MIMO antenna systems with channel covariance feedback. IEEE J. Select. Areas Common., 21(3):406417, April 2003.

10

-1

[4]

BER

10

-2

[5]

[6]
10
-3

10 12 Eb/No(db)

14

16

18

20

4. CONCLUSION

IJ
5. REFERENCES
[1] [2]
ISSN: 2230-7818

This paper we have developed an understanding and discussed the optimal receiver that can be used for a MIMO system for the detection of the symbol with minimum BER .As we know that we need a receiver that can detect the signal under the different fading conditions We discussed the 4x4 MIMO system for MMSE, ZF and ML equalizers. We studied that the MIMO receiver with ML (maximum likelihood) detection has the least BER for a given SNR.We see that the ZF (zero Forcing) detector has a greater BER

J. Zhang, R. Chen, J. G. Andrews, A. Ghosh, and W.Heath, Networked MIMO with clustered linear precoding, IEEE Transactions on Wireless Communications, vol. 8, no. 4, pp. 19101921, 2009.

H. Dai, A. F. Molisch, and H. V. Poor, Downlink capacity of interference-limited MIMO systems with joint detection, IEEE Transactions on Wireless Communications, vol. 3, no. 2, pp. 442453, 2004.

ES
[7] [8] [9] [10] [11] [12]

Figure 4. BER for 4x4 MIMO system with different type of detection.

M. O. Damen, H. E. Gamal, and G. Caire. On maximum-likelihood detection and the search for the closest lattice point. IEEE Transactions on Information Theory, 49(10):23892402, October 2003. D. Seethaler, and F. Hlawatsch, .Efficient detection algorithms for MIMO channels: A geometrical approach to approximate ML detection,. IEEE Trans. Signal Processing, Special Issue on MIMO Communications Systems, vol. 51, no. 11, pp. 2808.2820, Nov. 2003. A. Tarighat and A.H. Sayed. Joint compensation of transmitter and receiver impairments in OFDM systems. IEEE Trans. Wireless Communications, 6(1):240247, January 2007.

K. Karakayali, R. Yates, G. Foschini, and R. Valenzuela, Optimum zero-forcing beamforming with per-antenna power constraints,in Proceedings of the IEEE International Symposiumon Information Theory (ISIT 07), pp. 101105, Nice, France,June 2007. G.J. Foschini, K.Karakayali, and R.Valenzuela, Coordinating multiple antenna cellular networks to achieve enormous spectral efficiency, IEE Proceedings Communications, vol. 153, no. 4, pp. 548555, Aug. 2006. A. Del Coso, S. Simoens Distributed Compression for MIMO Coordinated Networks with a Backhaul Constraint Submitted to IEEE Trans. On Wireless Communications, Sep.2008. B. Hassibi and B. M. Hochwald. How much training is needed in multiple-antenna wireless links. IEEE Transactions on Information Theory, 48(4):951 963, April 2003.

4
@ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 304