See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/43527206

Performance analysis of multiple-input multiple-output orthogonal

frequency division multiplexing systems under the influence of antenna
mutual coupling effect

Article in IET Microwaves Antennas & Propagation · April 2009

Source: OAI

CITATIONS READS

0 1,183

4 authors, including:

Shiyang Lu Hon Tat Hui

The University of Sydney National University of Singapore
30 PUBLICATIONS 445 CITATIONS 63 PUBLICATIONS 875 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Shiyang Lu on 11 October 2014.

The user has requested enhancement of the downloaded file.

www.ietdl.org

Published in IET Microwaves, Antennas & Propagation

Received on 10th March 2007
Revised on 16th May 2008
doi: 10.1049/iet-map:20070302

ISSN 1751-8725

Performance analysis of multiple-input

multiple-output orthogonal frequency
division multiplexing systems under the
influence of antenna mutual coupling effect
S. Lu1 H.T. Hui2 M. Bialkowski1
1
School of Information Technology and Electrical Engineering, The University of Queensland, St. Lucia, Qld 4072, Australia
2
Department of Electrical and Computer Engineering, National University of Singapore, 4 Engineering Drive 3, Singapore
117576, Singapore
E-mail: shiyang@itee.uq.edu.au

Abstract: The effect of antenna mutual coupling on signal correlation and bit error rate (BER) of a multiple-input
multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) system has been modelled. A MIMO-
OFDM model with correlated channels is presented. The correlation matrices are determined with the antenna
mutual coupling taken into account. The effect of antenna mutual coupling is characterised using the receiving
and transmitting mutual impedances. A rigorous expression for the BER of the system employing quadrature
phase-shift keying digital modulation is derived. A 2  2 MIMO-OFDM system using the Alamouti space – time
block code for OFDM symbol transmission over 64 sub-channels spanning a 20 MHz band centred at 5.2 GHz
is simulated. The simulated results show that the antenna mutual coupling has a significant effect on the BER
performance, especially when the spacing between the antenna elements is small.

1 Introduction In turn, analyses of MIMO-OFDM systems taking into

account the signal correlation because of finite separation of
Recent years have witnessed an increasing demand for high antenna elements (spatial correlation) were presented in [5, 6].
data-rate wireless communication systems. In order to meet Considerations were limited to antenna elements treated as
this demand, the combination of the multiple-input independent and isolated sensors. Using this approach, the bit
multiple-output (MIMO) technique with the orthogonal error rate (BER) performance of MIMO-OFDM system was
frequency division multiplexing (OFDM), leading to the presented in [7, 8].
MIMO-OFDM system, has been proposed [1, 2]. The
MIMO-OFDM system is a spectrally efficient broadband In practice, the spatial correlation is affected not only by
wireless system which solves the frequency selective finite spacing in antenna arrays but also by mutual coupling
problem by dividing a wideband channel into multiple that exists between their elements. Taking this effect into
narrowband (frequency flat) channels using multiple consideration is of special importance especially in tightly
carriers. Its high spectral efficiency is obtained by applying spaced arrays. Only a limited number of works has dealt
the MIMO technique for individual sub-channels. This is with assessing BER performance of MIMO-OFDM
accomplished through the use of multiple element antennas system under the influence of antenna mutual coupling.
at both sides of the communication link and a suitable While dealing with the mutual coupling effect, correlation
signal transmission scheme. matrices based on the open-circuit definition of mutual
impedance were employed [9]. However, it has been shown
Models for multipath signal propagation for assessing in [10] that the approach of receiving and transmitting
MIMO system performances were presented in [2 – 4]. mutual impedances is more appropriate to model the

288 IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295
& The Institution of Engineering and Technology 2009 doi: 10.1049/iet-map:20070302

Authorized licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.
 www.ietdl.org

effects of mutual coupling in receiving and transmitting With respect to Fig. 1, the received signals at the kth
antennas. In particular, it has been shown that this subcarrier are expressed as
approach produces more accurate results for mutual
rffiffiffiffiffiffiffiffi
impedances than the one relying on the open-circuit s
r(k) ¼ H (k)c(k) þ w(k) (1)
definitions [11, 12]. MT f

In this paper, we apply the concept of receiving and where s is the average signal-to-noise ratio (SNR) at the
transmitting mutual impedances to determine the transmitter end and s/MT is the SNR for each transmitting
correlation matrices of the transmitting and receiving arrays antenna. c(k) is a MT  1 column vector given as
and derive a rigorous formula for calculating the average c(k) ¼ [c1 (k), c2 (k), . . . , cMT (k)]T , where ci (k) (i ¼ 1, 2, . . . ,
BER of a MIMO-OFDM system operating under the MT) represents the OFDM symbol transmitted over the kth
influence of antenna mutual coupling. Based on this subcarrier by the ith transmitting antenna. ci (k) is assumed
formula, we develop an algorithm to investigate the BER to be zero mean and of unit variance, that is
performance of this system. Simulation results on several ci (k)  CN (0, IMT ). w(k) ¼ [w1 (k), w2 (k), . . . , wMR (k)]T
typical scenarios will be presented. denotes the additive Gaussian noise vector with elements
being complex Gaussian random variables all having a zero
mean and unit variance. r(k) and w(k) are the MR  1
column vectors. H f (k) denotes the channel frequency
response for the kth subcarrier between the transmitting and
2 System model receiving antennas.
2.1 MIMO-OFDM system
We consider a MIMO-OFDM system with MT 2.2 Correlated MIMO-OFDM channel
transmitting antennas, MR receiving antennas and K models
subcarriers, as shown in Fig. 1. Each pair of transmitting
The statistical characteristics of MIMO-OFDM channels
and receiving antennas is assumed to have L
are described as follows. The channel frequency response
independent signal delay paths with the same delay
H f (k) in (1) is expressed as [13]
profile. In this system, the data to be transmitted are
first encoded by space– time or space– frequency encoders X
L
into MT encoded blocks. Then, each coded block forms H f (k) ¼ H l ej2pk=K (2)
an OFDM block to be transmitted by the array antenna. l¼1
At the receiving side, the signals received by multiple
receive antennas are decoded using the channel state where H l is a MR  MT gain matrix corresponding to the lth
information, which is provided by a suitable channel path used in the tapped delay line model. To simplify the
estimation algorithm. problem, the considered MIMO-OFDM channels are

Figure 1 MIMO-OFDM system model

IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295 289
doi: 10.1049/iet-map:20070302 & The Institution of Engineering and Technology 2009

Authorized licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.
www.ietdl.org

assumed to have only spatial correlation, whereas the other BER for QPSK modulation as given by
types of correlations such as temporal correlation [5], !
frequency correlation and tap gain correlation [14] are ð p=2
 2 sin2 p=4
temporarily ignored. According to [15], the correlation P(eg) ¼ exp g du (6)
p 0 sin2 u
structure of H l can be assumed as the Kronecker product of
the transmitting and receiving correlation matrices. This
assumption reasonably agrees with an experiment. The where u stands for the component of phase shift which is
covariance matrix RH of H l can be expressed as uniformly distributed over the range [0, 2p), as defined in
[8]. From this formula, it is apparent that the conditional
error probability is a function of the instantaneous SNR, g.
RH ¼ E{vec(H l )vec(H l ) } ¼ rT  rR (3)
By substituting (6) into (5), the following is obtained
where rT and rR denote the transmitting and receiving ð p=2 ð 1 !
correlation matrices, respectively. The expression given in 2 sin2 p=4
P e ¼ p(g) exp g dg du (7)
(3) is accurate when the transmitting and receiving arrays are p 0 0 sin2 u
well separated so that the average characteristics of the
channel response do not depend on the position of a In (7), the integrand is the moment-generating function
particular antenna. Using (3), the channel matrix H l can be (MGF) of g. In general, MGF is defined as [17]
generated using the following relation [13]
ð1
pffiffiffiffi Fx (s) ¼ E{esx } ¼ esx Px (x)dx (8)
vec(H l ) ¼ pl U L1=2 G (4) 1

where G is a vector containing independent and identically where E denotes the expectation operator and x is a real-
distributed (i.i.d.) complex Gaussian elements with a zero valued random variable. By substituting (8) into (7), the
mean and a unit variance. U and L are, respectively, the following is obtained
eigenvector matrix and eigenvalues matrix of the covariance !
matrix RH . pl is the average power distribution across the ð p=2
2 sin2 p=4
delay paths [15]. P e ¼ Fg du (9)
p 0 sin2 u

3 BER performance under the From (9), it is apparent that the average BER can be
influence of antenna mutual expressed as the integral of MGF Fg with the random
variable g being the instantaneous SNR. The MGF Fg
coupling can be rewritten as
The following section provides derivations for BER of a 
MIMO-OFDM system that takes into account the 
Fg (s) ¼ E{exp(sg)}s¼sin2 p=4=sin2 u (10)
antenna mutual coupling. This is done in two steps.
From (9) and (10), the problem of determining the average
3.1 Average BER performance under the BER under the influence of spatial correlation is simplified
influence of spatial correlation to calculating the output SNR at the receiving end in the
In a wireless multipath radio environment, the average BER presence of spatial correlation. The average SNR per
performance under spatial correlation fading is obtained symbol per each transmitting antenna has already been
by averaging the conditional error probability over the introduced in (1) and is given as s/MT. By defining
probability density function (PDF) of the instantaneous the channel gain as H l F , where k†kF denotes the
SNR, g, at the receiving end, as given by (5) Frobenius norm, the instantaneous output SNR, g, can be
calculated as
ð1
P e ¼ P(ejg)p(g)dg (5) s H 2

0
g¼ l F
MT
s
where P(ejg) denotes the conditional error probability. ¼ tr{H Hl Hl}
Expressions for the conditional BER for several coherent MT
and non-coherent modulation schemes have been reported s
¼ tr{vec(H l )vec(H H
l )} (11)
in many previous studies, for example [16, 17]. Here, we MT
assume the quadrature phase-shift keying (QPSK) digital
modulation scheme as a signal modulation scheme of the Next, substituting (4) into (11) leads to the following
system. Following the work concerning the conditional
error probability for multiple phase-shift keying digital s
g¼ p tr{LGGH } (12)
modulation schemes, [8], we can derive the conditional MT l

290 IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295
& The Institution of Engineering and Technology 2009 doi: 10.1049/iet-map:20070302

Authorized licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.
 www.ietdl.org

If the elements of G are denoted by gi, j and the elements of spatial correlation is obtained, as given by
the eigenvalues matrix L are given as li, j , (12) can be
ð p=2  1 
2Y
rewritten as M
T Y R M
s 
P e ¼ 1s p l s¼sin2 p=4=sin2 u du
p i¼1 j¼1 0 MT l i, j

s X
M
T XR M  
 
2 ð p=2 " #1
2Y T Y
M M
g¼ p l li, j gi, j  (13)
R li, j
MT i¼1 j¼1 ¼ 1þ du (16)
p i¼1 j¼1 0 2 sin2 u(MT =p l s)

where li, j are the eigenvalues of the correlation matrix RH .

By substituting (13) into (10), we obtain a new expression for
MGF Fg as From (16), one can see that P e is inversely related toli, j .
These eigenvalues become large when the spatial
( !) correlation becomes smaller. Therefore when the spatial
s X T XR  2
M M
Fg (s) ¼ E exp s p l
 
li, j gi, j  correlation is small, P e also becomes small. This is the case
MT i¼1 j¼1 when the performance of MIMO-OFDM system is
improved. On the other hand, when the spatial correlation
Y
MT YMR    2 
s    is strong, P e becomes large and the system performance
¼ E exp s 
pl li, j gi, j  s¼sin2 p=4=sin2 u
MT
becomes worse.
i¼1 j¼1

(14) 3.2 Average BER performance under the

influence of antenna mutual coupling
Here, the elements of G are assumed to be i.i.d. complex The next step concerns the modification of elements of the
Gaussian random variables with a zero mean and a unit correlation matrices due to the presence of antenna mutual
variance. Therefore (14) can be simplified to coupling. To this purpose, we will apply the concept of
receiving and transmitting mutual impedances, as described
in [10].
Y
MT YMR   
s
Fg (s) ¼ E exp s p l li, j 3.2.1 Receiving and transmitting mutual
MT
i¼1 j¼1 impedances: Fig. 2 shows a (MT , MR ) MIMO system.
MT Y
Y MR  In the presence of mutual coupling, the receiving antenna
1 
¼ s¼sin2 p=4=sin2 u (15) can be regarded as a coupled MR port network with MR
i1 j¼1
1  s s =M 
p l
T l i, j terminals. Let vR ¼ [vR1 , vR2 , . . . , vRMR ]T be the vector
of terminal voltages at the receiving antennas including the
mutual coupling effect. Let further the terminal voltages at
Now, by substituting (15) back into (9), the closed-form the receiving antennas without the mutual coupling effect
expression for the average BER under the influence of be represented by the vector uR ¼ [uR1 , uR2 , . . . , uRMR ]T .

Figure 2 MIMO system with antenna mutual coupling

IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295 291
doi: 10.1049/iet-map:20070302 & The Institution of Engineering and Technology 2009

Authorized licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.
www.ietdl.org

Using the concept of receiving mutual impedances, vR and than the corresponding formulation of the coupled received
uR are related by [11] voltages in [19]. Also, as shown in [10–12], the new concept
leads to more accurate results for determining mutual
1
vR ¼ Z R uR (17) coupling effects in transmitting and receiving antennas.

where
2 3 3.2.2 Total signal correlation and the average
R
Z1,2 R
Z1,MR
BER performance: In a rich multipath environment, the
6 1    7 propagation of multipath signals can be modelled using
6 ZL ZL 7
6 7 Clarke’s model [20]. In this case, the elements of the
6 ZR R
Z2,M 7
6  2,1
6 1   R 7
7 transmitting and receiving correlation matrices rT and rR
ZR ¼ 6
6
ZL ZL 7
7 (18) in (3) take the form of Bessel function [21]
6 .. .. .. .. 7
6 . . . . 7
6 7 rT(R)
6 R R 7 i, j ¼ J0 (kdi, j ) (22)
4 ZM ,1 ZM R ,2
5
 R   1
ZL ZL where di, j is the antenna separation between the ith and jth
antennas in the transmitting or receiving array. In an actual
ZL is the antenna terminal load impedance (assumed to be the multipath environment, the mean angle-of-arrival (AOA)
same for all antenna elements) and Zi,R j of the signal and AOA spread may also influence the signal
(i, j ¼ 1, 2, v . . . , MR ) is the receiving mutual impedance correlation [2]. Here, we assume Clarke’s signal model with
between the ith and jth receiving antennas, as defined in [10]. AOA from all directions.

The signal coupling at the transmitting end can be In order to include antenna mutual coupling effects in the
modelled in a similar way. Assuming vT and uT denote, channel correlation model, we need to recalculate the
respectively, the terminal voltage vectors at the transmitting correlation coefficients by replacing the terminal voltages
antennas with and without mutual coupling effect, vT has a without mutual coupling effect uR and uT by the terminal
similar relation to uT as that between vR and uR , namely voltages vR and vT [(17) and (19)] that take into account
1
the mutual coupling effect. For the case of two
v T ¼ Z T uT (19) transmitting and two receiving antennas (2  2 MIMO
case), the correlation coefficients with the antenna mutual
where coupling effect taken into account [22, 23] are given as
2 T T
3  2  2
Z1,2 Z1,M 
6 1    T
7 2Re{ZL Zi,Rj } þ (ZL  þZi,R j  )rRi, j
6 ZL ZL 7 rRMC(i, j) ¼    2 (23)
6 7 Z 2 þZ R  2Re{Z Z R }rR
6 T
Z2,1
T
Z2,M 7 L i, j L i, j i, j
6 T 7
6  1   7
ZT ¼ 6
6
ZL ZL 7
7 (20)  2  2
6 .. .. .. .. 7 2Re{ZL Zi,Tj } þ (ZL  þZi,Tj  )rTi, j
6 . . . . 7 rTMC(i, j) ¼    2 (24)
6 7
6
4 ZM T T
ZM
7
5 Z 2 þZ T  2Re{Z Z T }rT
,1 L i, j L i, j i, j
T ,2
 T   1
ZL ZL
The variations of rR T
MC(1,2) and rMC(1,2) with antenna
and Zi,T j
(i, j ¼ 1, 2, . . . , MT ) is now the transmitting mutual separation are shown in Fig. 3 for the case of two
impedance between the ith and jth transmitting antennas. monopole antennas with the length of 0.24l and assuming
the wire diameter of 0.01l where l is the wavelength
For the case of the two antenna arrays, the transmitting measured at the frequency of 5.26 GHz (the centre
T frequency of the IEEE 802.11a standard [24]). While
mutual impedance, Z1,2 between antennas 1 and 2, is
defined as the voltage V1 induced on the first (transmitting) calculating the correlation coefficients, the value of terminal
antenna terminal load to the exciting current I2 through the load impedance ZL was assumed as 50 V. All the
second (transmitting) antenna terminal load. That is transmitting and receiving mutual impedances between the
antennas were determined numerically using the method of
T V1 moments [25]. Note that the values of the correlation
Z12 ¼ (21)
I2 coefficients with antenna mutual coupling included are
significantly different from the spatial correlation [values of
Note that the definition in (21) is similar to the conventional Bessel function in (22)]. The comparisons of results
definition of mutual impedance [18] except that it now takes presented in Fig. 3 indicate that antenna mutual coupling
into account the antenna load ZL . The formulation of the cannot be ignored. Considerable differences are observed
coupled voltages vR and vT in (17) and (19) is much simpler for the antenna spacing of up to one wavelength (1l ).

292 IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295
& The Institution of Engineering and Technology 2009 doi: 10.1049/iet-map:20070302

Authorized licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.
 www.ietdl.org

are the same. The information bits are first modulated by a

QPSK modulator. The Alamouti structured space–time
block code (STBC) word [26] is then used to transmit the
OFDM symbol. A channel bandwidth of 20 MHZ centred
around 5.26 GHz divided into 64 subchannels is
considered. Each data block contains 500 information bits.
The results are obtained using 10 000 channel realisations.

Figs. 4 and 5 show the average BER performance against

SNR of a MIMO-OFDM system when antenna elements in
the transmitting and receiving arrays are separated either by
0.2l (stronger coupling condition) or 0.4l (weaker
coupling condition), where l is measured at 5.26 GHz.

Figure 3 Variation of the correlation coefficients with

antenna separation for two monopole antennas in a rich-
multipath indoor environment
The antenna dimensions are: length ¼ 13.7 mm, diameter ¼ 0.6 mm
and operating at 5.26 GHz

Once the antenna mutual coupling is correctly built into

the correlation coefficients, the covariance matrix in (3) can
be modified accordingly, that is

RMC ¼ rTMC  rRMC (25)

where rRMC and rTMC are now the respective correlation

matrices with entries given by rRMC(i, j) and rTMC(i, j) in (23)
and (24), respectively. Also by denoting the eigenvalues of
the covariance matrix RMC as lMC(i, j) , the average BER in
the presence of antenna mutual coupling effect can be
expressed as Figure 4 Average BER performance of an STBC-OFDM
system with two transmitting and two receiving antennas
ð p=2 !1 under SNR conditions when both transmitting and
2Y T Y
M M
R lMC(i, j)
P e ¼ 1þ du (26) receiving antenna separation are 0.2 carrier wavelength
p i¼1 j¼1 0 2 sin2 u(MT =p l s)

Equation (26) is the final result for the BER of MIMO-

OFDM system with QPSK modulation scheme in which
the antenna mutual coupling is taken into account. Note
that the required eigenvalues lMC(i, j) can be calculated
through (22) – (25).

4 Numerical results and

discussion
This section presents a number of results that are generated
using the derived formula for the average BER of MIMO-
OFDM system operating under the presence of antenna
mutual coupling effect. In simulations, the case of 2  2
MIMO-OFDM is considered, with all antennas
(transmitting and receiving) having the same dimensions as
the two monopoles shown in Fig. 3. As indicated earlier, Figure 5 Average BER performance of an STBC-OFDM
the load impedance required for calculating mutual system with two transmitting and two receiving antennas
impedances is fixed to 50 V. It is further assumed that the under SNR conditions when both transmitting and
antenna separations in the transmitting and receiving arrays receiving antenna separation are 0.4 carrier wavelength

IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295 293
doi: 10.1049/iet-map:20070302 & The Institution of Engineering and Technology 2009

Authorized licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.
www.ietdl.org

These two figures illustrate two cases, when the effect of

antenna mutual coupling is neglected or included in
considerations. In addition to these two cases, there is
another case for comparison, that is, the case with i.i.d.
channels (when channels are de-correlated). It can be
observed from these figures that at low values of SNR, the
results obtained in all the cases are almost the same, with
the i.i.d. case producing the best performance. An
inspection of results for other values of SNR indicates that
the signal correlation differentiates itself only at high values
of SNR. It is interesting to note from Fig. 4 that under the
stronger coupling condition (antenna separation ¼ 0.2l ),
the case of ignoring the effect of antenna mutual coupling
produces the largest bit errors compared with the cases of
antenna mutual coupling included or i.i.d. case. In turn,
Fig. 5 reveals that under the weaker coupling condition Figure 6 Average BER performance of 2  2 STBC-OFDM
(antenna separation ¼ 0.4l ), the case of ignoring the effect over correlated MIMO channels against antenna
of antenna mutual coupling produces a slightly better BER separation with antenna separation from 0.1 to 2.0
performance compared with the cases when the antenna carrier wavelength and at an SNR equal to 20 dB
mutual coupling is included in considerations. To explain
this difference, we recall from Fig. 3 and the expression in coupling effect). The results shown in Fig. 6 are clear that
(26) that at an antenna separation of 0.2l, the presence of when the antenna separation is smaller than about 0.8l,
antenna mutual coupling reduces signal correlation which ignoring antenna mutual coupling causes a significant error
in turn makes the eigenvalues of covariance matrix RMC in in determining the BER. However, when the separation is
(26) become larger. This leads to the average BER greater than 0.8l, this error becomes negligible.
becoming smaller.

When the antenna separation is 0.4l, the situation is 5 Conclusions

reversed. The antenna mutual coupling effect increases the The effect of antenna mutual coupling on signal correlation
signal correlation and hence the eigenvalues of RMC has been successfully modelled into the calculation of the
become smaller. This increases the BER when antenna average BER of MIMO-OFDM system. Using the
mutual coupling is taken into account. Looking further at concept of receiving and transmitting mutual impedances,
both Figs. 4 and 5, it is clear that ignoring the existence of the coupling matrices for the receiving and transmitting
antenna mutual coupling is not acceptable when evaluating arrays have been obtained in a new way. A rigorous formula
the performance of MIMO-OFDM systems. For example, for evaluating BER performance of this system under the
in Fig. 4, the average BER with antenna mutual coupling influence of antenna mutual coupling effect has been
effect at SNR ¼ 14 dB is about 5  1025 while that rigorously derived. Based on this formula, computer
without antenna mutual coupling effect is about 1024, a simulations have been performed. The results obtained
100% difference. This observation leads to the conclusion under different degrees of antenna mutual coupling have
that the previous analyses of BER that ignored antenna shown that the antenna mutual coupling has a significant
mutual coupling over-estimated the average BER by 100%. effect on the BER performance, especially when antenna
This is an unacceptably large error. separation is small (,0.8l ). The presented approach
should be found useful to investigate the BER performance
Fig. 6 shows the average BER performance as a function of of MIMO-OFDM systems that operate under correlated
antenna separation from 0.1l up to 2.0l at SNR of 20 dB. It channel conditions.
can be seen that antenna separation (and hence antenna
mutual coupling) has a very significant effect on the average
BER. At small antenna separations 0.1l – 0.3l, antenna 6 References
mutual coupling reduces the average BER tremendously
from the case of ignoring antenna mutual coupling. At [1] SAMPATH H., TALWAR S., TELLADO J., ERCEG V. , PAULRAJ A.: ‘A
medium antenna separations 0.3l – 0.5l, antenna mutual fourth-generation MIMO-OFDM broadband wireless
coupling increases the average BER compared with the case system: design, performance, and field trial results’, IEEE
of ignoring antenna mutual coupling. At still greater Commun. Mag., 2002, 40, pp. 143 – 149
antenna separations 0.5l – 0.8l, the effect of antenna
mutual coupling on BER reverses again. Therefore Fig. 6 [2] BOLCSKEI H., BORGMANN M., PAULRAJ A.J.: ‘Impact of the
shows how serious is the error of ignoring antenna mutual propagation environment on the performance of space –
coupling in calculating the BER. It depends on the frequency coded MIMO-OFDM’, IEEE J. Sel. Areas
antenna separation (i.e. the strength of antenna mutual Commun., 2003, 21, pp. 427 – 439

294 IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295
& The Institution of Engineering and Technology 2009 doi: 10.1049/iet-map:20070302

Authorized licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.
 www.ietdl.org

[3] SKLAR B.: ‘Rayleigh fading channels in mobile digital [14] INTARAPANICH A., KAFLE P.L., DAVIES R.J., SESAY A.B.: ‘Effect of tap
communication systems. Part 1: Characterization’, IEEE gain correlation on capacity of OFDM MIMO systems’,
Commun. Mag., 1997, 35, pp. 90– 100 Electron. Lett., 2004, 40, pp. 86 – 88

[4] BIALKOWSKI M.E., UTHANSAKUL P., BIALKOWSKI K., DURRANI S.: [15] KAI Y., BENGTSSON M., OTTERSTEN B., MCNAMARA D., KARLSSON P.,
‘Investigating the performance of MIMO systems from an BEACH M. :‘Modeling of wide-band MIMO radio channels
electromagnetic perspective’, Microw. Opt. Technol. Lett., based on NLoS indoor measurements’, IEEE Trans. Veh.
2006, 48, pp. 1233– 1238 Technol., 2004, 53, pp. 655 – 665

[5] SADEK A.K., SU W.F., LIU K.J.R. : ‘Diversity analysis for [16] AALO V.: ‘Performance of maximal-ratio diversity
frequency-selective MIMO-OFDM systems with general systems in a correlated Nakagami fading environment’,
spatial and temporal correlation model’, IEEE Trans. IEEE Trans. Commun., 1995, 43, pp. 2360 – 2369
Commun., 2006, 54, pp. 878 – 888
[17] PROAKIS J.G.: ‘Digital communications’ (McGraw-Hill,
[6] WEIFENG S., SAFAR Z., OLFAT M., LIU K.J.R.: ‘Obtaining full- New York, 1995, 3rd edn.)
diversity space – frequency codes from space – time codes
via mapping’, IEEE Trans. Signal Process., 2003, 51, [18] JORDAN E.C.: ‘Electromagnetic waves and radiating
pp. 2905 – 2916 systems’ (Prentice-Hall Inc., Englewood Cliffs, NJ, 1968), Ch. 11

[7] SIMON M.K., ALOUINI M.S.: ‘A unified performance [19] JANASWAMY R.: ‘Effect of element mutual coupling on the
analysis of digital communication with dual selective capacity of fixed length linear arrays’, IEEE Antennas Wirel.
combining diversity over correlated Rayleigh and Propag. Lett., 2002, 1, pp. 157 – 160
Nakagami-m fading channels’, IEEE Trans. Commun., 1999,
47, pp. 33 – 43 [20] CLARKE R.H. : ‘A statistical theory of mobile-radio
reception’, Bell Syst. Tech. J., 1968, 47, pp. 957 – 1000
[8] FEMENIAS G.: ‘BER performance of linear STBC from
orthogonal designs over MIMO correlated Nakagami-m [21] NARASIMHAN M.S., CHRISTOPHER S.: ‘On expansion of
fading channels’, IEEE Trans. Veh. Technol., 2004, 53, cylindrical Bessel function’, IEEE Trans. Antennas Propag.,
pp. 307– 317 1984, AP-32, pp. 1268– 1270

[9] WOLFF E.A.: ‘Antenna analysis’ (Wiley, New York, 1966) [22] HUI H.T., OW YONG W.T., TOH K.B.: ‘Signal correlation between
two normal-mode helical antennas for diversity reception
[10] HUI H.T., LOW H.P. , ZHANG T.T., LU Y.L. : ‘Receiving mutual in a multipath environment’, IEEE Trans. Antennas
impedance between two normal-mode helical antennas Propag., 2004, 52, (2), pp. 572 – 577
(NMHAs)’, IEEE Antennas Propag. Mag., 2006, 48,
pp. 92– 96 [23] HUI H.T.: ‘Reply to “Comment on “Signal correlation
between two normal-mode helical antennas for diversity
[11] HUI H.T.: ‘An effective compensation method for the reception in a multipath environment””’, IEEE Trans.
mutual coupling effect in phased arrays for magnetic Antennas Propag., 2005, 53, (8), p. 2777
resonance imaging’, IEEE Trans. Antennas Propag., 2005,
53, pp. 3576 – 3583 [24] Naftali Chayat Task Group A (5 GHz PHY) Chair:
‘Criteria for comparison of 5 GHz high speed PHY
[12] HUI H.T.: ‘Improved compensation for the proposals’. IEEE P802.11-97/96, September 1997
mutual coupling effect in a dipole array for direction
finding’, IEEE Trans. Antennas Propag., 2003, 51, [25] HARRINGTON R.F. : ‘Field computation by moment
pp. 2498 – 2503 methods’ (IEEE Press, New York, 1993)

[13] BOLCKEI H., GESBERT D., PAULRAJ A.J.: ‘On the capacity of [26] ALAMOUTI S.M.: ‘A simple transmit diversity technique for
OFDM-based spatial multiplexing systems’, IEEE Trans. wireless communications’, IEEE J. Sel. Areas Commun.,
Commun., 2002, 50, pp. 225 – 234 1998, 16, pp. 1451– 1458

IET Microw. Antennas Propag., 2009, Vol. 3, Iss. 2, pp. 288– 295 295
doi: 10.1049/iet-map:20070302 & The Institution of Engineering and Technology 2009

Authorized
View publication stats licensed use limited to: National University of Singapore. Downloaded on March 27, 2009 at 09:46 from IEEE Xplore. Restrictions apply.