Performance Analysis of MIMO Receiver Techniques

Performance Analysis of MIMO Receiver
Techniques
Mr. V. Ayyem Pillai,
Dr. R. Sudhakar,
Assistant Professor / Dept of ECE,

Karpagam College of Engineering,
Coimbatore, INDIA.
Professor and Head / Dept of ECE,

Dr. Mahalingam College of Engineering & Technology,
Coimbatore, INDIA.
Abstract Multiple antenna systems provide an enormous

increase in spectral efficiency compared with single antenna
systems.Various Transceiver and receiver techniques have
been developed over the years to detect the symbols with as far
as possible small error form the mixed symbols. In this paper,
an attempt has been made to analyse various receive
techniques and compare their performance. Simulations have
been done for comparison purpose.
Section VII deals with simulation and comparision. Finally

conclusion has been given in section VIII.
Keywords-component; Zero Forcing, Minimum mean square

error, ZF SIC, MMSE SIC, SVD, QRD, GMD
I.INTRODUCTION
Multiple antenna systems provide an enormous increase
in spectral efficiency compared with single antenna
systems[1]. The first practical multiple input multiple output
system that proved this potential is the BLAST(Bell Labs
Layered Space Time) structure. The original BLAST
structure was D-BLAST(Diagonal BLAST) architecture, in
which code blocks are dispersed diagonally in spacetime[2]. D-BLAST was difficult to be implemented due to
its diagonal nature. A simplified version, called V-BLAST
(Vertical BLAST) was proposed in [3]. The V-BLAST
signals can be decoded in a parallel way by zero forcing
(ZF) and minimum mean square error(MMSE) principles.
In [2] and [3] Foschini et al proposed a ZF SIC
(successive interference cancellation) technique. MMSE
criterion was proposed for the original V-BLAST in [4] and
its lower complexity version was proposed in [5]. A very
efficient detection algorithm, which uses QR decomposition
of the channel matrix and called ZF-sorted QR
decomposition [SQRD] was proposed in [6] and [7]. An
extension to ZF SQRD with MMSE was given in[8]
MMSE-SQRD finds optimal order, not always. Singular
value decomposition with ZF was proposed in [13]. MIMO
transceivers based on geometric mean decomposition
(GMD) with ZF-BLAST was proposed in [9].
In this paper, various receiver techniques have been
reviewed , simulations have been done and the results have
been compared. Section I deals with linear receivers doing
parallel reception. Section II deals with VBLAST ZF and
MMSE
recivers. Section IV
discusses QR based
VBLAST reception. Section V and VI have been
respectively devoted for SVD and GMD based reception.
Linear Rceivers
A. ZF Receivers
ZF receiver is one of the linear receivers. ZF detection is
based on detecting the desired signal by inverting the
channel and multiplying the receiver antenna output.
where
Where
is the pseudoinverse of H
We have
The problems are,

1. The noise is enhanced
2. The noise is correlated
Each scalar channel is decoded independently
ignoring noise correlation. ZF receivers have less
complexity, but are sub-optimal.
II.LINEAR MMSE DETECTOR
Here, the detector design is based on minimum mean
square error criterion, where the overall error will be less
than that of ZF detectors. The matrix satisfying the linear
relationship and MMSE is
Here, the detected signal is
where y is the receiver antenna signal.

In this case, the noise enhancement is less than that
of ZF, however, the symbol interference is more. However,
the overall BER is less than that of ZF detectors.
First some i is made 1 with actual H, estimation for

sk+ k is made , then the
column of H is made zero, then
new G is found. From the new G, the min norm which
corresponding to max norm of H ) row is formed. For i=1
we find min norm row of which corresponds to max norm
column of H. which in turn corresponds to the strongest
transmitting antenna signal. For example, if 5 th antenna
signal is the strongest the decoded signal will be of
transmitting antenna 5.
Let the ordered set in which components of the
transmitted symbol vector s are extracted be
{
III
A
VBLAST RECEIVERS
ZF - VBLAST
In [1], ZF based successive interference cancellation

technique was proposed. In this, instead of parallel
detection, serious detection is done. For example, symbol
corresponding to layer I is detected and the corresponding
column in H is removed, leading to MT-1 transmit antennas.
This procedure of nulling and cancelling is followed until all
signals are detected.
k1 may correspond to any column of H. After

extracting the symbol corresponding to k1 (k1 may be any
from 1 to MT ), that column has to be nulled.
We g
The statistical properties of QHn is same as n

Now,
ZF BLAST WITH QR
II.
In this case, the channel matrix is QR decomposed,

where Q is a unitary
matrix, and R is an
upper triangular matrix
H = QR
(
(
)
)(
independent of
the upper layer signals. The lowest layer signal is
where is antenna-received signal vector and is the

modified received signal vector.
Here,
Q stands for quantization.
The term with

is an interference term, and will
be removed.
`Assuming all previous decisions are
correct,
where
()
where
is the interference due to signals from
, where n is ZMCSW noise.
Now, this signal is subtracted from y and ith

column is made zero in the channel matrix, and now it will
be viewed as if the system has MT-1 transmitter antennas.
If we remove
, We get
Q takes care of the signal detected from AWGN.

Q is the decision made in accordance with the
constellation.
IV SINGULAR VALUE DECOMPOSITION BASED MIMO *
TANSCEIVER SYSTEM
Unlike the original VBLAST system, the SVD-based

transceiver system is a closed loop one. For spatial
multiplexing (SM), the channel information is used for
orienting beams in eigen vector directions.
The channel matrix is decomposed as,
where
,
are unitary
matrixes and is a diagonal matrix consisting of singular
values of H.
B.
SVD with MMSE

This technique uses GMD of the MIMO channel matrix
for precoding and receiver processing.
GMD of H is given as follow
H = QRPH
where, Q and P are unitary P is the linear precoder
and
is an upper triangular matrix. The diagonal
elements of R are geometric mean of the K positive
singular values of H.
V. GMD WITH V-BLAST
The transmitted symbol vector,
unprecoded
, where is
The received signal y is processed as follows,
y =
A particular symbol is
,where
is an eigen value
of H
the problem has been reduced to QR V blast .

i = 1,2,3,,r-1,he is the rank of
H.
A. SVD with ZF
The modified antenna received vector
For MMSE case,
[(
) (
G is given by
]
VI.SIMULATION & RESULTS

Simulation was carried out for a 4x4 MIMO SM system
using 1000 symbols and 4-qam modulation. The channel
selected was i.i.d.(independent and identically distributed).
The detections were done using ZF, MMSE, V-BLAST ZF
SIC and VBLAST MMSE SIC. The SNR(dB) selected was
in the range of 0 to 20 dB. It can be seen from figure that
the successive interference techniques outperform the
parallel techniques. The MMSE SIC outperforms the ZF
SIC especially in high SNR range.
The simulatons were also carried out for two other
modulations, Bpsk and QPSK. Form figure 2 and 3 it is
observed that the BER performance is poor for both types

for all techniques.
-1
10
BER
Simulations were done for VBLAST

with QR
decomposion based receivers. BPSK and QPSK
modulations were used. It can be seen that the QPSK
outperforms BPSK.
Comparison of BER performance between SVD based ZF and MMSE Systems

0
10
mmselinear
zflinear
Finally simulations were done for SVD based ZF and

SVDbased .MMSE receivers. As expected
MMSE
performs better than ZF receivers.
-2
10
VBLAST with MMSE,MMSE-ZIC,ZF,ZF-SIC
10
-3
-1
10
BER
10
MMSE
MMSE-SIC
ZF
ZF-SIC
8
10
12
symbol SNR(dB)
14
16
18
20
Figure 3. Simulated plots illustrating BER vs SNR(dB) for four different

receiver decoding techniques, ZF, ZF SIC, MMSE and MMSE SIC for
BPSK modulation.
-2
10
10
MMSE
MMSE-SIC
ZF
ZF-SIC
-3
10
-4
10
12
SNR in dB
14
16
18
20
BER
10
-1
10

receiver decoding techniques, ZF, ZF SIC, MMSE and MMSE SIC for 4QAM modulation.
-0.21
10
MMSE
MMSE-SIC
ZF
ZF-SIC
-0.22
10
-2
10
10
12
14
16
18
20
22
SNR in dB
24
26
28
30
-0.23
10
-0.24
Figure 4. Simulated plots illustrating BER vs SNR(dB) SVD based

receivers
BER
10
-0.25
10
-0.26
10
V Blast detection with QR decompsition with QPSK
10
-0.27
10
-0.28
10
-1
10
12
SNR in dB
14
16
18
10
20

receiver decoding techniques, ZF, ZF SIC, MMSE and MMSE SIC for
QPSK modulation.
BER
-2
10
-3
10
-4
10
10
15
20
snr in dB
25
30
35
40
Figure 5. Simulated plots illustrating BER vs SNR(dB) for QR based

recivers.
VIII CONCLUSION
In this paper, a detailed analysis was made on various
techniques that have been developed so far. Some receiver
techniques are of closed loop MIMO systems and some are
of open loop systems.It is seen that generally MMSE
techniques with SIC do better than other techniques.
Simulations carried outsupport this statement.
REFERENCES
[1] I. E. Telatar, Capacity of multi-antenna Gaussian
channels, Eur. Tran. Telecommun., vol. 10, no. 6, pp.
585595, 1999.
[2] G. J. Foschini, Layered Space-Time architecture for
wireless communication in fading environment when
using multiple antennas,Bell Labs Technical
Journal,Vol.1, no.2 pp. 41-59, Autumn 1996.
[3] P. W. Walniansky G.J. Foschini G.D. Golden R. A.
Valenzuela, V-BLAST: An architecture for realizing
very high data rate in rich scatteing wireless channel,in
IEEE Proceedings of ISSSE-98 Pisa, Italy 29
September1998.
[4].A. Benjebbour, H. Murata, and S. Yoshida, .Comparison
of Ordered Successive Receivers for Space-Time
Transmission,. in Proc. IEEE Vehicular Technology
Conference (VTC), USA, Fall 2001.
[5] B. Hassibi, .An Ef_cient Square-Root Algorithm for
blast,. in Proc. IEEE Intl. Conf. Acoustic, Speech, Signal
Processing, Istanbul, Turkey,June 2000, pp. 5.9.
[6]. D. Wubben, J. Rinas, R. Bohnke, V. Kuhn, and K. D.
Kammeyer, .Ef_cient Algorithm for Detecting Layered
Space-Time Codes,. in Proc. ITG Conference on Source
and Channel Coding, Berlin, Germany, January 2002,
pp. 399.405.
[7]. D. Wubben, R. Bohnke, J. Rinas, V. Kuhn, and K. D.
Kammeyer, .Ef_cient Algorithm for Decoding Layered
Space-Time Codes,. IEE Electronic Letters, vol. 37, no.
22, pp. 1348.1350, October 2001.
[8]. Ronald Bohnke, Dirk Wubben, Volker Kuhn, and
Karl-Dirk Kammeyer Reduced Complexity MMSE
Detection for BLAST Architecture
[9] G. Lebrun, T.Y ing,'M. Faulher MIMO Transmission
over a Time-Varying Channel Using SVD
[10]. Yi Jiang Jian Li William W. Hager
Transceiver
Design
Using
Geometric
Decomposition.
MIMO
Mean

Performance Analysis of MIMO Receiver Techniques

Uploaded by

Copyright:

Available Formats

Performance Analysis of MIMO Receiver Techniques

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Performance Analysis of MIMO Receiver Techniques

Uploaded by

Copyright:

Available Formats

Performance Analysis of MIMO Receiver

Assistant Professor / Dept of ECE,

Professor and Head / Dept of ECE,

Abstract Multiple antenna systems provide an enormous

Section VII deals with simulation and comparision. Finally

Keywords-component; Zero Forcing, Minimum mean square

The problems are,

Here, the detected signal is

where y is the receiver antenna signal.

First some i is made 1 with actual H, estimation for

In [1], ZF based successive interference cancellation

k1 may correspond to any column of H. After

The statistical properties of QHn is same as n

In this case, the channel matrix is QR decomposed,

where is antenna-received signal vector and is the

Q stands for quantization.

The term with

is the interference due to signals from

, where n is ZMCSW noise.

Now, this signal is subtracted from y and ith

Q takes care of the signal detected from AWGN.

Unlike the original VBLAST system, the SVD-based

SVD with MMSE

The received signal y is processed as follows,

the problem has been reduced to QR V blast .

VI.SIMULATION & RESULTS

observed that the BER performance is poor for both types

Simulations were done for VBLAST

Comparison of BER performance between SVD based ZF and MMSE Systems

Finally simulations were done for SVD based ZF and

VBLAST with MMSE,MMSE-ZIC,ZF,ZF-SIC

Figure 3. Simulated plots illustrating BER vs SNR(dB) for four different

VBLAST with MMSE,MMSE-ZIC,ZF,ZF-SIC

Figure 1. Simulated plots illustrating BER vs SNR(dB) for four different

Figure 4. Simulated plots illustrating BER vs SNR(dB) SVD based

V Blast detection with QR decompsition with QPSK

Figure 2. Simulated plots illustrating BER vs SNR(dB) for four different

Figure 5. Simulated plots illustrating BER vs SNR(dB) for QR based

You might also like