Paper

Compensation of Fading Channels

Using Partial Combining Equalizer
in MC-CDMA Systems
Mohammed Zidane1 , Said Safi2 , and Mohamed Sabri1
1Department of Physics, Faculty of Sciences and Technology, Sultan Moulay Slimane University, Morocco
2 Department of Mathematics and Informatics, Polydisciplinary Faculty, Sultan Moulay Slimane University, Morocco

Abstract—In this paper the performance of a partial combin- obstacles in the channels. Reflections from these obstacles
ing equalizer for Multi-Carrier Code Division Multiple Access degrade the transmitted signal before it reaches the receiver.
(MC-CDMA) systems is analytically and numerically evalu- Hence, channel equalization is required to reduce the Bit
ated. In the part of channel identification, authors propose Error Rate (BER) of the receiver as small as possible. In
a blind algorithm based on Higher Order Cumulants (HOC) fact, the goal of the equalization techniques is to reduce
for identifying the parameters representing the indoor sce- the effect of the fading and the interference while not en-
nario of Broadband Radio Access Networks (BRAN A) chan-
hancing the effect of the noise on the decision of what data
nel model normalized for MC-CDMA systems. Theoretical
analysis and numerical simulation results, in noisy environ-
symbol was transmitted. Therefore, the problem of channel
ment and for different Signal to Noise Ratio (SNR), are pre- identification appears.
sented to illustrate the performance of the proposed algo- In this paper, authors propose an algorithm for blind chan-
rithm in the one hand, and the other hand the impact of par- nel identification, using higher order cumulants. There are
tial combining equalizer on the performance of MC-CDMA several motivations behind this interest [6]. Firstly, higher
systems. order cumulants are blind to all kinds of Gaussian noise that
is, the additive noise Gaussian will vanish in the higher or-
Keywords—bit error rate, blind channel identification, higher
order cumulants, MC-CDMA systems, partial combining equal- der cumulants domain. Secondly, cumulants are useful in
izer. identifying non-minimum phase channels when the input is
non-Gaussian and is contaminated by Gaussian noise.
The problem of the blind identification of the Broad-
1. Introduction band Radio Access Network (BRAN) channels, normal-
ized by the European Telecommunications Standards Insti-
Fourth generation (4G) technology allows user to efficiently tute (ETSI) [7]–[8], and downlink MC-CDMA equaliza-
share common resources. However, the exponential growth tion using higher order cumulants was proposed by several
of multimedia users request fast data rates and reliable authors [9]–[12].
transmission. The 4G wireless systems utilizing available
In this contribution, authors present a partial combining
limited bandwidth in a spectrally efficient manner. To at-
equalizer for downlink MC-CDMA systems equalization,
tain these aims, there are two principle contending tech-
for that is in the one hand, the problem of the blind identi-
nologies, i.e. Orthogonal Frequency Division Multiplexing
fication of (BRAN A) channel using the proposed algorithm
(OFDM) and Code Division Multiple Access (CDMA).
is considered, and in the other hand the presented equal-
Therefore, OFDM-CDMA and MC-CDMA gain a lot of
izer after the channel identification to correct the chan-
attention for wireless mobile communication [1]–[2]. The
nel’s distortion is used. The numerical simulation results,
principles of MC-CDMA [3] is that a single data symbol
in noisy environment, are presented to illustrate the accu-
is transmitted on multiple narrow band subcarriers. In-
racy of the proposed algorithm, and the performance of
deed, in MC-CDMA systems, spreading codes are applied
equalization.
in the frequency domain and transmitted over independent
subcarriers. However, multicarrier systems are very sen-
sitive to synchronization errors such as carrier frequency
offset and phase noise. Synchronization errors cause loss 2. Channel Identification using Higher
of orthogonality among subcarriers and considerably de- Order Cumulants
grade the performance especially when large number of
subcarriers presents [4]–[5]. The MC-CDMA modulator 2.1. Channel Model
is an effective technique for combating multipath fading
over highly dispersive wireless channels. The problem en- The term channel refers to the transmitting space (medium)
countered in digital communication is the synchronization between the transmitter and the receiver antennas as shown
between the transmitter and the receiver; there are many in Fig. 1.

5
Mohammed Zidane, Said Safi, and Mohamed Sabri

As the cumulants are expressed in function of moments,

the estimates cumulants of order two, three and four are
defined respectively by:
N
b2,y (τ1 ) = 1 ∑ y(i)y(i + τ1 ),
Cb2,y (τ1 ) = M (5)
N i=1

Cb3,y (τ1 , τ2 ) = b3,y (τ1 , τ2 ) =

M
1 N
= ∑ y(i)y(i + τ1)y(i + τ2 ) ,
N i=1
(6)

Cb4,y (τ1 , τ2 , τ3 ) = b4,y (τ1 , τ2 , τ3 ) −

M
Fig. 1. Channel model. − b2,y (τ1 )M
M b2,y (τ3 − τ2 ) −
− M b2,y (τ3 − τ1 ) −
b2,y (τ2 )M
The characteristics of wireless signal changes as it travels
− M b2,y (τ2 − τ1 ) .
b2,y (τ3 )M (7)
from the transmitter antenna to the receiver antenna. These
characteristics depend upon the distance between the two
antennas, the path(s) taken by the signal, and the environ- 2.3. Basic Relationships
ment (buildings and other objects) around the path. The
In this section, the main general relationships between cu-
profile of received signal can be obtained from that of the
mulants of the output signal and impulse response coeffi-
transmitted signal if we have a model of the medium be-
cients are described. The starting point for all algorithms
tween the two. This model of the medium is called channel
based on higher order cumulants is Brillinger and Rosen-
model (Fig. 1).
blatt relation shows that the m-th order cumulants of y(k)
In this paper it is assumed that the channel is time invariant
can be expressed as a function of impulse response coeffi-
and its impulse response is characterized by P paths of
cients h(i) as follows [13]:
magnitudes β p and phases θ p . The impulse response is
given by: Cm,y (τ1 , τ2 , . . . , τm−1 ) =
P−1
θp
h(τ ) = ∑ β pe δ (τ − τ p ) . (1) q
= ξm,x ∑ h(i)h(i + τ1 ) . . . h(i + τm−1 ), (8)
p=0
i=0
In general, the received signal can be obtained by convolv-
ing the transmitted signal with the impulse response of the where ξm,x represents the m-th order cumulants of the ex-
channel: citation signal x(k) at origin.
r(t) = h(t) ∗ x(t) + n(t) , (2) Peyre et al. presents the relationship between different m-th
Z+∞P−1 and n-th order cumulants of the output signal, y(k), and the
coefficients h(i), where n > m and (n, m) ∈ N ∗ − {1}, are
r(t) = ∑ β peθ p δ (τ − τ p)x(t − τ )d τ + n(t) = linked by the following relationship [14]:
−∞ p=0
P−1 q
= ∑ β p eθ p x(t − τ p) + n(t), (3) ∑ h( j)Cn,y ( j + τ1 , j + τ2, . . . , j + τm−1 , τm , . . . , τn−1 ) =
p=0 j=0
q h n−1 i
where x(t) is the input sequence, h(t) is the impulse re- = µ ∑ h(i) ∏ h(i + τk ) Cm,y (i + τ1 , i + τ2 , . . . , i + τm−1 ),
sponse coefficients, τ p is the time delay of the p-th path, i=0 k=m
and n(t) is the additive noise sequence. (9)
ξn,x
where µ = ξm,x .
2.2. Moments and Cumulants In order to simplify the construction of the proposed algo-
In this Subsection, a mathematical definitions of the esti- rithm we assume that:
mated moments and cumulants needed to identifying the • the input sequence, x(k), is independent and identi-
impulse response parameters of finite impulse response cally distributed (i.i.d.) zero mean, and non-Gaus-
(FIR) systems are presented. sian;
Let us consider a random non-Gaussian variable y(k). The
sample estimates are given by: • the system is causal and bounded, i.e. h(i) = 0 for
N i < 0 and i > q, where h(0) = 1,
bm,y (τ1 , . . . , τm−1 ) = 1 ∑ y(i)y(i + τ1 ) . . . y(i + τm−1 ) . (4)
M
N i=1 • the system order q is known,

6
 Compensation of Fading Channels Using Partial Combining Equalizer in MC-CDMA Systems

• the measurement noise sequence n(k) is assumed The system of Eq. (15) can be written in matrix form as:
zero mean, i.i.d., Gaussian and independent of x(k)  
with unknown variance. C4,y (q + 1, 0, 0) ... C4,y (2q, 0, 0)
 C4,y (q + 1, 1)−α ... C4,y (2q, 1, 0) 
 
The problem statement is to identify the parameters of the  . . . 
 ×
system h(i)(i=1,...,q) using the cumulants of the measured  . . . 
 . . . 
output signal y(k).
C4,y (q + 1, q, 0) ... C4,y (2q, q, 0)−α
 
h(1)  
3. Proposed Algorithm  .  0
   −C4,y (q, 1, 0) 
 .   
By substituting n = 4 and m = 2 into Eq. (9) the following    . 
 .   
equation can be obtained:    . 
×  h(i) = , (16)
   . 
 .   
q  .   . 
∑ h( j)C4,y ( j + τ1 , τ2 , τ3 ) =    . 
 . 
j=0 −C4,y (q, q, 0)
q h 3 i h(q)
= µ ∑ h(i) ∏ h(i + τk ) C2,y (i + τ1 ) , (10) where α = C4,y (q, 0, 0). Or, in more compact form, the
i=0 k=2
Eq. (16) can be written as follows:

q
Mhe = d, (17)
∑ h( j)C4,y ( j + τ1 , τ2 , τ3 ) = where M is the matrix of size (q + 1) × (q) elements, he
j=0
is a column vector constituted by the unknown impulse re-
q
sponse parameters h(i)i=1,...,q and d is a column vector of
= µ ∑ h(i)h(i + τ2)h(i + τ3 )C2,y (i + τ1 ), (11)
i=0
size (q + 1) as indicated in the Eq. (16). The least squares
solution of the system of Eq. (17), permits blindly identifi-
ξ4,x cation of the parameters h(i) and without any information
where µ = ξ2,x .
of the input selective channel. Thus, the solution will be
The autocorrelation function of the (FIR) systems vanishes written under the following form:
for all values of |τ | > q, equivalently:
 he = (M T M)−1 M T d .
b (18)
6= 0, |τ | ≤ q;
C2,y (τ ) =
0 otherwise.
If we suppose that τ1 = q the Eq. (11) becomes:
4. MC-CDMA Model
q
The multicarrier code division multiple access (MC-
CDMA) system is based on the combination of code di-
∑ h( j)C4,y ( j + q, τ2, τ3 ) = µ h(0)h(τ2)h(τ3 )C2,y (q), (12)
vision multiple access (CDMA) and orthogonal frequency
j=0
division multiplexing (OFDM), which is potentially robust
and for τ3 = 0 the Eq. (12) becomes: to channel frequency selectivity. However, the complex
q
symbol ai of each user i is, firstly, multiplied by each chip
ci,k of spreading code, and then applied to the modulator
∑ h( j)C4,y ( j + q, τ2 , 0) = µ h2(0)h(τ2 )C2,y (q) . (13)
j=0

The considered system is causal and bounded, thus, the

interval of the τ2 is τ2 = 0, . . . , q.
Else if we suppose that τ2 = 0, and using the cumulants
properties Cm,y (τ1 , τ2 , . . . , τm−1 ) = 0 , if one of the variables
τk > q, where k = 1, . . . , m − 1, the Eq. (13) becomes:

C4,y (q, 0, 0) = µ h3 (0)C2,y (q) . (14)

Thus, we are based on Eq. (14) for eliminating C2y (q) in

Eq. (13), we obtain the equation constituted of only the
fourth order cumulants, this equation describe the proposed
algorithm:
q
∑ h( j)C4,y ( j + q, τ2, 0) = h(τ2 )C4,y (q, 0, 0) . (15)
Fig. 2. MC-CDMA modulator principle.
j=0

7
Mohammed Zidane, Said Safi, and Mohamed Sabri

of multicarriers. Each subcarrier transmits an element of possible to express G – the diagonal matrix composed of
information multiply by a code chip of that subcarrier. the coefficients gk equalization:
Figure 2 explains the principle of the MC-CDMA systems.  
g0 0 ... 0
 0 g1 ... 0 
4.1. MC-CDMA Transmitter  
 . . . . 
G=  .
.
 (26)
The MC-CDMA signal is given by:  . . . 
 . . . . 
N p −1 0 0 ... gNp −1
ai
x(t) = p ∑ ci,k e2 jπ fkt , (19)
Np k=0

where fk = f0 + Tkc , N p is the number of subcarriers, and

we consider Lc = N p .
Fig. 3. Principe of the single user-detection.
4.2. MC-CDMA Receiver
After the equalization and the spreading operation, the es-
The downlink received MC-CDMA signal at the input re- timation abi of the emitted user symbol ai , of the i-th user
ceiver is given by the following equation: can be written by the following equations:
N p −1 Nu −1
1 P−1 abi = cTi Gr
r(t) = p ∑ ∑ ∑ ×
N p p=0 k=0 i=0 (20) = cTi G(HCa + n)
n o
×ℜ β p e jθ p ai ci,k e2 jπ ( f0 +k/Tc )(t−τ p ) + n(t) = cTi GHCa + cTi Gn . (27)

The Eq. (20) can be written as follows: Nu −1 N p −1

abi = ∑ ∑ ci,k (gk hk cq,k aq + gk nk ) =
r = HCa + n, (21) q=0 k=0
N p −1 Nu −1 N p −1
where r denotes a vector composed of the values received c2i,k gk hk ai +
on each subcarrier:
= ∑ ∑ ∑ ci,k cq,k gk hk aq +
k=0 q=0 k=0
| {z } | {z }
r = [r0 , . . . , rNp −1 ]T . (22) I (i=q) II (i6=q)
N p −1
The matrix H represents the matrix of complex coefficients
of channel with size N p × N p :
+ ∑ ci,k gk nk , (28)
k=0
| {z }
  III
h0 0 ... 0
 0 h1 ... 0  where the term I, II and III of Eq. (28) are, respectively, the
 
 . . . .  signal of the considered user, a signals of the others users
H = . (23)
 . . . .  (multiple access interferences) and the noise pondered by
 . . . . 
the equalization coeffcient and by spreading code of the
0 0 ... hNp −1 chip.
The matrix C represent the spreading codes: If we suppose that the spreading code are orthogonal, i.e.
  N p −1
c0,0 c0,1 ... c0,Nu −1
 c1,0 c1,1 ... c1,Nu −1  ∑ ci,k cq,k = 0 ∀i 6= q , (29)
  k=0
 . . . . 
C=  , (24) Eq. (28) will become:
 . . . . 
 . . . . 
N p −1 N p −1
cLc −1,0 cLc −1,1 ... cLc −1,Nu −1 c2i,k gk hk ai +
abi = ∑ ∑ ci,k gk nk . (30)
k=0 k=0
where | {z } | {z }
I (i=q) III
ci = [c0,i , c1,i , . . . , cLc −1,i ]T ,
a = [a0 , . . . , aNu −1 ]T , 5. Partial Combining Equalizer
n = [n0 , . . . , nNp −1 ]T . (25)
In [15] a partial combining (PC) technique was introduced,
At the reception, we demodulate the signal according the with coefficient gk function of a PC parameter, −1 ≤ β ≤ 1,
N p subcarriers, and then we multiply the received sequence as given by:
by the code of the user. Figure 3 explains the single user h∗k
detection principle. Using the above matrix notation, it is gk = . (31)
|hk |1+β

8
 Compensation of Fading Channels Using Partial Combining Equalizer in MC-CDMA Systems

The estimated received symbol, abi of symbol ai of the user i • all subchannel parameters are added, to construct the
is described by: full BRAN A channel impulse response.
N p −1 N p −1 The simulation is performed with Matlab software and for
|hk |2 h∗k
abi = ∑ c2i,k ai + ∑ ci,k nk . (32) different SNR.
k=0 |hk |1+β k=0 |hk |1+β

6.1. Identification of BRAN A Channel using the

5.1. Particular Case: β = 1 – Zero Forcing Equalizer
Proposed Algorithm
The gain factor of the zero forcing (ZF) equalizer, is given
In this subsection the BRAN A channel model is consid-
by the equation:
1 ered. Figure 4 show the impulse response estimation for
gk = . (33) this channel using the proposed algorithm for different SNR
hk
and an data length N = 5400.
The estimated received symbol, abi of symbol ai of the user i
is described by:
N p −1 N p −1
1
abi = ∑ c2i,k ai + ∑ ci,k nk . (34)
k=0 k=0 hk

6. Simulation Results
In this section the numerical results for blind identifica-
tion and equalization in MC-CDMA systems are presented.
For that we consider the BRAN A model representing the
propagation in an indoor case. The Eq. (35) describe the
impulse response of BRAN A channel:
NT
h(τ ) = ∑ Ai δ (τ − τi ) . (35)
i=0 Fig. 4. Estimated of the BRAN A channel impulse response, for
different SNR and a data length N = 5400.
In the Table 1 the measured values corresponding the
BRAN A radio channel impulse response are summarized.
This figure shows clearly the influence of Gaussian noise
Table 1 on parameter estimation of the BRAN A impulse response.
Delay and magnitudes of 18 targets of BRAN A channel This influence is clear principally for the last five values,
where the estimated parameters do not follow those mea-
Delay τi [ns] Mag. Ai [dB] Delay τi [ns] Mag. Ai [dB] sured. But, before the last fifth values, the 13th first es-
0 0 90 −7.8 timated values are closed to those measured are observed.
10 −0.9 110 −4.7 This due that the additive Gaussian noise vanished in the
20 −1.7 140 −7.3 higher order cumulants domain.
30 −2.6 170 −9.9
40 −3.5 200 −12.5
50 −4.3 240 −13.7
60 −5.2 290 −18
70 −6.1 340 −22.4
80 −6.9 390 −26.7

Although, the BRAN A channel is constituted by NT = 18

parameters and seeing that the latest parameters are very
small, for that we have taking the following procedure:
• the BRAN A channel impulse response is decom-
posed into three subchannels:
3
h(i) = ∑ h j (i) ; (36)
j=1

• the parameters of each subchannel are estimated in- Fig. 5. Estimated of the BRAN A channel impulse response
dependently, using the proposed algorithm; using all target, for different SNR and a data length N = 5400.

9
Mohammed Zidane, Said Safi, and Mohamed Sabri

6.2. Magnitude and Phase Estimation of BRAN A A model was used representing the propagation in an indoor
Channel using the Proposed Algorithm case normalized for MC-CDMA systems. To estimate the
coefficients of this equalizer, the authors have proposed an
In Fig. 5 we represent the estimation magnitude and phase algorithm based on fourth order cumulants. The proposed
of the impulse response of the BRAN A channel using the algorithm shows their efficiency in the impulse response
proposed algorithm, for an SNR varying between 12 dB channel identification with very good precision. In the part
and 20 dB, the data length is 5400. of the equalization for the MC-CDMA systems using the
The estimated magnitude and phase have the same form. presented equalizer, it has been demonstrated that the par-
In the fact, one can see a low influence of the noise on the tial combining equalizer is very adequate for correcting the
estimation of the magnitude and phase principally when the channel distortion for 0.7 ≤ β ≤ 0.9.
noise is SNR > 12 dB, and we have not more difference
between the estimated and the true ones.
References
6.3. Compensation of Fading Channels using Partial [1] Z. Wang and G. B. Giannakis, “Wireless multicarrier communica-
Combining Equalizer tions: where Fourier meets Shannon”, IEEE Sig. Process. Mag.,
vol. 17, no. 3, pp. 29–48, 2000.
In order to evaluate the performance of the MC-CDMA [2] S. Hara and R. Prasad, “Overview of multicarrier CDMA”, IEEE
system, using the presented equalizer. These performances Commun. Mag., vol. 35, pp. 126–133, 1997.
are evaluated by calculation of the BER, for different values [3] N. Yee, J.-P. M. G. Linnartz, and G. Fettweis, “Multi-Carrier-CDMA
of β , using the measured and estimated, using proposed in indoor wireless networks”, in Proc. 4th Int. Symp. on Personal,
Indoor and Mob. Radio Commun. PIMRC’93, Yokohama, Japan,
algorithm of the BRAN A channel impulse response. 1993, pp. 109–113.
[4] M. Frikel, B. Targui, F. Hamon, and M. M’Saad, “Adaptive equal-
ization using controlled equal gain combining for uplink/downlink
MC-CDMA systems”, Int. J. of Signal Process., vol. 4, no. 3,
pp. 230–237, 2008.
[5] M. Frikel, S. Safi, B. Targui, and M. M’Saad, “Channel Identifi-
cation using chaos for an uplink/downlink multicarrier code divi-
sion multiple access system”, J. Telecommun. and Inform. Technol.,
no. 1, pp. 48–54, 2010.
[6] M. Boulouird and M. M’Rabet Hassani, “Blind identification of MA
models using cumulants”, Int. J. of Electrical, Electronic Sci. &
Engin., vol. 1, no. 6, pp. 619–623, 2007.
[7] ETSI, “Broadband Radio Access Networks (BRAN); High Perfor-
mance Radio Logical Area Network (HIPERLAN) Type 2; Require-
ments and architectures for wireless broadband access”, Jan. 1999.
[8] ETSI, “Broadband Radio Access Networks (BRAN); HIPERLAN
Type 2; Physical Layer”, Dec. 2001.
[9] M. Zidane, S. Safi, M. Sabri, A. Boumezzough, and M. Frikel,
“Broadband radio access network channel identification and down-
link MC-CDMA equalization”, Int. J. of Energy, Inform. and Com-
mun., vol. 5, no. 2, pp. 13–34, 2014.
[10] M. Zidane, S. Safi, M. Sabri, and A. Boumezzough, “Blind Iden-
tification Channel Using Higher Order Cumulants with Application
to Equalization for MC-CDMA System”, Int. J. of Elec., Robot.,
Electron. and Commun. Engin., vol. 8, no. 2, pp. 369–375, 2014.
Fig. 6. BER of the estimated and measured BRAN A channel,
[11] S. Safi, M. Frikel, A. Zeroual, and M. M’Saad, “Higher order cu-
for different SNR, using the presented equalizers.
mulants for identification and equalization of multicarrier spreading
spectrum systems”, J. of Telecommun. and Inform. Technol., no. 1,
Figure 6 shows the simulation results of BER estimation, pp. 74–84, 2011.
[12] M. Zidane, S. Safi, M. Sabri, A. Boumezzough, and M. Frikel,
for different SNR, using presented equalizer of the BRAN
“Adaptive algorithms versus higher order cumulants for identification
A channel impulse response when 0.7 ≤ β ≤ 0.9 the partial and equalization of MC-CDMA”, J. of Telecom. & Inform. Technol.,
combining equalizer is more precise and gives good results no. 3, pp. 53–62, 2014.
than those obtained by ZF equalizer. Also, the ZF equal- [13] D. Brillinger and M. Rosenblatt, “Computation and interpretation of
izer is best compared to the results obtained using partial k-th order spectra”, in Spectral Analysis of Time Signals, B. Harris,
Ed. New York: Wiley, 1967, pp. 189–232.
combining equalizer for β ≤ 0.6.
[14] J. L. Peyre, D. Dembélé, and G. Favier, “Identification of MA mod-
els using higher order statistics: A general formulation”, in Proc.
2nd ATHOS Worksh. on System Identif. and Higher Order Statist.,
7. Conclusion Sophia-Antipolis, France, 1993, pp. 20–21.
[15] A. Conti, “MC-CDMA bit error probability and outage minimization
In this contribution, a partial combining equalizer has been through partial combining”, IEEE Commun. Let., vol. 9, no. 12,
pp. 1055–1057, 2005.
analytically and numerically investigated in MC-CDMA
systems. In the part of the channel identification, the BRAN

10
 Compensation of Fading Channels Using Partial Combining Equalizer in MC-CDMA Systems

Mohammed Zidane received tional School for Applied Sciences, Tangier Morocco, from
the M.Sc. in Electronic En- 2003 to 2005. Since 2006, he is a professor of applied
gineering from Sultan Moulay mathematics and programming at the Faculty of Science
Slimane University, Beni Mel- and Technic, Beni Mellal Morocco. In 2008 he received the
lal, Morocco, and M.Sc. in Ph.D. degree in telecommunication and informatics from
Optoelectronics and Laser In- the Cadi Ayyad University. His general interests span the
strumentation from Faculty of areas of communications and signal processing, estimation,
Science and Techniques Has- time-series analysis, and system identification-subjects on
san first University, Settat, Mo- which he has published 10 journal papers and more than
rocco. He has Ph.D. in Digital 40 conference papers. Current research topics focus on
Telecommunications and Signal transmitter and receiver diversity techniques for single- and
Processing from Sultan Moulay Slimane University, Beni multi-user fading communication channels, and wide-band
Mellal, Morocco. His research interest includes digital wireless communication systems.
communications and signal processing, linear and non- E-mail: safi.said@gmail.com
linear Broadband Radio Access Network (BRAN) chan- Department of Mathematic and Informatic
nels identification, higher order statistics, blind identifica- Polydisciplinary Faculty
tion and equalization in MC-CDMA systems, subjects on Sultan Moulay Slimane University
which he has published 8 journal papers and 4 conference PO box 592, Beni Mellal, Morocco
papers.
E-mail: zidane.ilco@gmail.com Mohamed Sabri received the
Department of Physics Ph.D. degree in Signal Proc-
Faculty of Sciences and Techniques essing and Telecommunica-
Sultan Moulay Slimane University tions, from Rennes I University,
PO box 523, Beni Mellal, Morocco France. His current research
interests are communication
networks evolution and human
Said Safi received the B.Sc. face detection and recognition.
degree in Physics from Cadi He is currently working as
Ayyad University, Marrakech, a Professor in Department of
Morocco, in 1995, M.Sc. and Physics, Faculty of Sciences
Ph.D. degrees from Chouaib and Techniques, University of Sultan Moulay Slimane,
Doukkali University and Cadi Beni Mellal, Morocco.
Ayyad University, Morocco, in E-mail: sipt03@yahoo.fr
1997 and 2002, respectively. Department of Physics
He has been a professor of in- Faculty of Sciences and Techniques
formation theory and telecom- Sultan Moulay Slimane University
munication systems at the Na- PO box 523 Beni Mellal, Morocco

11