You are on page 1of 7

4110

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 8, OCTOBER 2013

Dual-Hop Relaying Communications with Cochannel


Interference Over Fading Channels
Kostas P. Peppas, Member, IEEE

AbstractThis paper provides an analytical framework for the performance evaluation of dual-hop decode-and-forward (DF) cooperative
systems operating over fading channels in the presence of cochannel
interference (CCI) and additive white Gaussian noise (AWGN). The interferers are subject to independent but not necessarily identically distributed
fading. For the considered system, assuming multiple interferers at
both the relay and the destination, simple accurate analytical expressions
for the outage probability (OP), average bit error probability (ABEP),
and average capacity are derived. Various numerically evaluated results
accompanied with Monte Carlo simulations are presented to corroborate
the accuracy of the proposed approximations.
Index TermsApproximation methods, cochannel interference (CCI),
decode-and-forward (DF), moment-based estimators, fading.

I. I NTRODUCTION
Multihop relaying has been shown to improve the performance and
extend the coverage of many wireless communication systems due to
its ability to offer spatial diversity while still satisfying the size and
power constraints of mobile devices [1]. On the other hand, with the
rapidly growing demand for increased capacity broadband wireless
networks, frequency reuse, which is employed for better spectrum
efficiency, is considered an essential trend. However, cochannel interference (CCI) due to the deliberate reuse of radio channels could
severely limit the reception quality and therefore is regarded as a
major limiting factor on the design and implementation of relaying
systems. Although many papers have been published in the open
technical literature concerning the performance of relaying systems
under multipath fading, relatively few works have investigated these
kinds of systems when CCI is further being considered. Representative
past examples can be found in [2][10] and the references therein.
Recently, the so-called distribution has been proposed as a
versatile fading model, which can accurately model the small-scale
variations of the fading signal in non-line-of-sight conditions and
incorporates both Hoyt and Nakagami-m fading models as special
cases [11]. Because of its ability to characterize so many different
fading channel models, this generalized fading model has recently
gained interest in the context of the performance analysis of digital
communications over fading channels [12][20]. Nevertheless, to the
best of our knowledge, there are still very few works in the open
technical literature dealing with performance analysis issues of relaying systems operating in an fading environment. Some recent
representative examples can be found in [21][23].
In this paper, the performance of a dual-hop system employing the
decode-and-forward (DF) protocol and operating in an fading
environment is investigated. In our analysis, we assume the presence of
-faded multiple cochannel interferers with arbitrary fading parameters at both the relay and the destination. Because of the mathematical

intractability inherent to the exact analysis, simple precise approximations for the outage probability (OP), the average bit error probability
(ABEP) of digital modulation schemes, and the average capacity
are presented. For interference-limited environments, the proposed
expressions are valid for arbitrary values of the fading parameter
of the desired signal, whereas for interference-plus-noise-limited environments, they are valid for integer values of . However, there is
no such restriction on the fading severity of the interfering signals.
The remainder of this paper is structured as follows. Section II
shows an overview of the fading model. Section III outlines the
considered system model. In Section IV, the outage and error rate
performance and the average capacity of dual-hop systems in the
presence of CCI are investigated. Numerical results and their
interpretations are given in Section V, whereas Section VI concludes
this paper. The mathematical notations used in this paper are as
follows. () is the gamma function [24, Eq. (8.310/1)], Ia () is the
modified Bessel function of the first kind and order a [24, Eq. (8.406)],
2 F1 () is the Gauss hypergeometric function [24, Eq. (9.100)], () is
the Tricomi hypergeometric function [24, Eq. (9.210/2)], F1 is the first
Appell hypergeometric function of two variables [24, Eq. (9.180/1)],
and 2 () is the confluent Lauricella hypergeometric function of two
variables [25]. The probability density function (pdf) of a random variable (RV) X is fX (x), and its corresponding cumulative distribution
function (cdf) is FX (x). Pr{} is the probability operator, and E
denotes expectation.
II. OVERVIEW OF THE FADING M ODEL
Let be the instantaneous SNR of a link subject to fading. The
pdf of is given by [11]



2 +0.5 h 0.5
2h
exp
f () =

()H 0.5 +0.5

I0.5

2H

(1)

where = E, > 0 is related to the fading severity, h = (2 +


1 + )/4, and H = ( 1 )/4 with 0 < < .1 For arbitrary
values of , the cdf of can be obtained as [17]
F () =

(A1 A2 ) 2
2 (, ; 1 + 2; A1 , A2 )
(1 + 2)

(2)

where (A1 = 2(h H))/, and (A2 = 2(h + H))/. For integer values of and with the help of [15, Eq. (7)] and [24, Eq. (8.467)],
F () can be greatly simplified as
F () = 1

1
()

h
H

 
1 k1
2 

n=1 k=0

r=0

Arn r

r!

an, k exp(An )

(3)

where (a1, k = (1)k (+k1)!H k )/(2+k k!(H h)k ), (a2, k =


(1) ( + k 1)!H k )/(2+k k!(H + h)k ). Finally, using
Manuscript received February 9, 2012; revised March 17, 2013; accepted March 19, 2013. Date of publication April 16, 2013; date of current
version October 12, 2013. The review of this paper was coordinated by
Prof. M. D. Yacoub.
The author is with the Laboratory of Mobile Communications, Institute of
Informatics and Telecommunications, National Center for Scientific ResearchDemokritos, Athens 15310, Greece (e-mail: kpeppas@iit.demokritos.gr).
Digital Object Identifier 10.1109/TVT.2013.2258415

1 The pdf of can be expressed in two formats, namely, formats 1 and 2, for
which two fading models are associated [11]. In [11], it is shown that Format 1
can be converted into Format 2 using a bilinear transformation. Therefore,
without loss of generality, throughout this paper, we only consider Format 1
-faded links.

0018-9545 2013 IEEE

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 8, OCTOBER 2013

[11, Eq. (21)], the th moment of can be expressed in closed


form as
E  =

(2 + )
h+ (2) (2)

2 F1

+1
1 H2

, +
; + ; 2
2
2
2 h


.

(4)

III. S YSTEM AND C HANNEL M ODEL


Let us consider a dual-hop relaying system with multiple interferers
over fading channels, where source node S communicates using
two time slots with destination node D through the help of a DF
relay R. In the first time slot, the relay receives the source data and
the interfering signals from I1 interfering terminals. Consequently,
the received signal at R from S and I1 interfering terminals is
given by
y1 =

P 1 h1 s 1 +

I1 


1,  s1,  + Cn1 .
P1,  h

(5)

=1

In (5), P1 stands for the transmit power; h1 represents the fading


1
denotes the interfering
coefficient of the S R link; {
s1,  }I=1
I1
I1

symbols; {h1,  }=1 and {P1,  }=1 are the fading coefficients and the
average power of the interfering signals, respectively; n1 is the additive
white Gaussian noise (AWGN); and C {0, 1}. In addition, when
C = 1, both interference and thermal noise are taken into account. The
case C = 0 corresponds to interference-limited transmissions, i.e., the
effect of thermal noise is ignored. Such an assumption is valid in small
cell systems, namely in pico- and microcellular radio systems, where
radio link performance is usually limited by interference rather than
noise. In the second time slot, the relay decodes the source symbol s1 ,
and the re-encoded symbol s2 is transmitted to D with power P2 . At
the destination, the received signal from the relay and I2 interfering
terminals is given by
y2 =

P 2 h2 s 2 +

I2 


2,  s2,  + Cn2
P2,  h

(6)

=1

where h2 represents the fading coefficient of the R D link;


2
2,  }I2 and {P2,  }I2
denotes the interfering symbols; {h
{
s2,  }I=1
=1
=1
are the fading coefficients and the average power of the interfering
signals, respectively; and n2 is the AWGN. The corresponding signalto-interference-plus-noise ratio (SINR) at the relay and destination
nodes can be expressed as2
i =

Pi |hi |2
ai
=
 Ii
2
2

Ci + =1 Pi,  |hi,  |


C + =1 Yi, 

 Ii

4111

form expressions for these statistics have been derived in [19], assuming interference-limited environments and positive integer fading
parameters for the interfering signals. However, in the practical case of
arbitrary fading parameters, the exact computation of these statistics is
rather intricate, particularly when both interference and thermal noise
are considered. To circumvent
this problem, approximate expressions
I i
Yi,  can be used to obtain simple yet
to the pdf of Zi  =1
accurate formulations for the statistics of i . A possible solution
to this problem was given in [20], where an approximation to
the pdf of Zi was used to obtain highly accurate approximations
to the cdf of i in an interference-limited / fading environment. The resulting cdf expression is, however, in the form of a
single integral; consequently, the evaluation of performance metrics
of interest, such as the ABEP and the average capacity, involves the
computation of twofold integrals. Therefore, a simpler yet accurate
and efficient performance analysis of wireless systems over fading
channels with CCI is desirable. Motivated by this need, within the
context of this paper, the gamma distribution has been chosen as
the convenient approximation to the sum of squared RVs, for
which the parameters are estimated from moments of the sum of
the squared envelopes. The motivation behind the choice of the
gamma distribution is twofold: First, as it was shown in [11] and [16],
a squared RV can be expressed as the sum of two gamma RVs
with suitably defined parameters. Second, in [6], [7], [10], [26], and
[27], the gamma distribution was successfully used to obtain highly
accurate approximations for the sum of nonidentical gamma RVs.
Relying on the ideas presented in the previously cited works, we feel
that, although the distribution yields a better approximation to the
pdf of Zi , due to its additional degree of freedom, the proposed approximation will yield accurate results for the problem under consideration,
and the underlying mathematical analysis will be also significantly
simplified.
A. Gamma Approximation to the Sum of i.n.i.d. RVs
We propose to approximate the pdf of Zi , i.e., fZi (z), with a gamma
i , namely
distribution with parameters m
i and

fZi (z)  fSi (s) =

i = EZi 

To obtain analytical expressions for important performance metrics


of the considered system, the statistics of i are required. Closed2 Throughout

this paper and unless otherwise stated, index i = 1 and 2.

i 1
sm
m
is
exp
.
i
(m
i)

(8)

m
i =

i
EZi2 

(9)

The required moments E{Zi } and E{Zi2 } can be evaluated in terms


of the individual moments of Y, i using the multinomial identity as
[7, Eq. (11)]
E {Zi }

j1



  j1 

jIi 2

j1 =0 j2 =0

jIi 1 =0

IV. P ERFORMANCE A NALYSIS

m
i

To render fS (s) as a good approximation to the pdf of Zi , we use


moment-based estimators for the computation of the parameters m
i
i of fS (s) from the exact moments of Zi . Using the analysis
and
i
i are obtained in terms of the
presented in [7], the parameters Zi and
moments of Yi as

(7)

where ai = Pi |hi |2 /i2 , Yi,  is the interference-to-noise power ratio


of the th interferer, and i2 is the noise power. RVs ai are
distributed with parameters i , i , and i = E|hi |2 Pi /i2 . On the
other hand, Yi,  are assumed to be independent nonidentically distributed (i.n.i.d.) RVs with parameters i,  , i,  , and i,  =
i,  |2 Pi,  / 2 .
E|h
i

m
i
i

j1

j2

jIi 2

jIi 1

jI 1

1
E Yi,j
E Yi,j12j2 E Yi, Iii
1


(10)

where the moments of Y, i can be readily obtained from (4). It


is noteworthy that the proposed closed-form approximation is very
simple, and the determination of the distribution parameters from
(9) is straightforward. Moreover, as it will become evident from the

4112

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 8, OCTOBER 2013

numerical examples shown in the following, the proposed approximation is also accurate for a wide range of values of the parameters i,  ,
i,  , and i,  .
Using a general result presented in [28, App.], an error bound
for the proposed approximation can be derived. Specifically, let us
and fY (y) have the
assume that the pdfs fX (x)

first 2n 2 moments
identical, namely Mk = 0 xk fX (x)dx = 0 y k fY (y)dy, where
n is a nonnegative integer and k = 0, . . . 2n 2. Then, for arbitrary
real , the distance between their corresponding cdfs can be derived as
[28, Coroll. 2.5.4]

() = |FX () FY ()|

F1

1m
i , i , i , 1 + 2i ,
i A2, i

i A2, i
m
i +

m
i
 hi i  m
i
i

Hi

(i )(m
i)


2 1

|Pk ()|
is the kernel polynomial, and
where n () =
k=0
Pk (x) is defined in [28, pp. 576]. Consequently, the error bound is
dependent on and the distribution parameters. Setting n = 2 to (11)
and using (9), () can be easily evaluated.
B. OP Analysis
In a wireless communication system subject to CCI, an outage is
declared when the instantaneous SINR falls below a predetermined
outage threshold th . More specifically, in DF relaying systems, an
outage event occurs if either one of the two-hop links is in outage.
Mathematically speaking, the OP can be expressed as
Pout (th ) = Pr {min{1 , 2 } < th }

(15)

  r (m
i + r)

i 1 i k1

k=0

2


(11)

n

i A1, i

i A1, i
m
i +

If the parameter i is limited to positive integer values, by substituting


(2) and (8) to (13) and using the definition of the gamma function,
Fi () can be expressed as
Fi () 1








fY (y)dy n1 ()
= fX (x)dx


0

n=1

r!

r=0

Arn, i an, k, i
r+m
i
i
An, i + m

2) Interference-Plus-Noise-Limited Systems: In this case, to obtain


an analytical expression for Fi (), we assume that the parameter
i is limited to positive integer values only. Substituting (3) and
(8) to (13) and using the definition of the gamma function and
the binomial identity, Fi () can be expressed in closed form as
follows:

m
i
 hi i  m
i
Fi () 1

Hi

(i )(m
i)

  r

i 1 i k1

k=0

r=0

r!

r  


r

Arn,i an,k,i exp(An,i )




= F1 (th ) + F2 (th ) F1 (th )F2 (th )

(12)

n=1

j=0

where the cdf of i can be expressed as [4, Eq. (4)], [7, Eq. (7)]

.
m
i+j

(m
i +j)
i
An,i + m
i


Fi () =

. (16)

(17)
Fai [ (C + z)] fZi (z)dz.

(13)

1) Interference-Limited Systems: For the interference-limited case


and assuming arbitrary values of i , by substituting (2) and (8) to (13),
Fi () can be expressed as
Fi ()

(A1, i A2, i )i

m
i
i

m
i

(m
i )(1 + 2i )


exp

m
iz
i

Denoting Pe, 1 and Pe, 2 as the ABEP of the links S R and R


D, respectively, the ABEP of the considered system can be expressed
as [30, Eq. (6)]
P be = Pe, 1 + Pe, 2 2Pe, 1 Pe, 2 .

i +2i 1
zm

2i

(18)

The ABEP Pe, i can be expressed in terms of the cdf of i as


[31, Eq. (12)]

C. ABEP Analysis

2 (i , i ; 1 + 2i ; A1, i z, A2, i z)dz.

(14)

Pe, i =

qp
2(p)


p1 exp(q )Fi ()d
0

Using the Laplace transform pair presented in [29, Eq. (3.43.1.3)]


and [24, Eq. (9.183.1)], Fi () can be evaluated in closed form as
given in3
Fi ()

(A1, i A2, i )i (2i + m


i)
(m
i )(1 + 2i )

1+

i A1, i

m
i

i 
1+

m
i
i

2i

i A2, i

m
i

1
=
(p)

2p1

exp


Fi

2
q


d

(19)

i
2i

3 Note that the Appell F hypergeometric function is a standard built1


in function available in Mathematica. A computer program for the efficient
implementation of this function is also available in [25, App. B, pp. 295].

where the parameters p and q account for different modulation


schemes. Specifically, p = 1 and q = 1 for binary differential phaseshift keying, p = 0.5 and q = 1 for coherent binary phase-shift keying
(BPSK), and p = 0.5 and q = 0.5 for coherent binary frequency-shift
keying.
1) Interference-Limited Systems: For interference-limited systems
and assuming arbitrary values of i , we propose an efficient method
for the evaluation of Pe, i , based on the semi-infinite GaussHermite

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 8, OCTOBER 2013

4113

quadrature (SI-GHQ) [32]. In particular, by applying a Q-point SIGHQ, an accurate approximation for Pe, i can be obtained as
Pe, i

Q
1 
j x2p1
Fi
j
(p)

j=0

x2j
q


(20)

where j are the weights, and xj is the abscissa. For given values
of Q, the corresponding values of the parameters j , xj , as well as
the truncation error, are given in [32, Tab. II]. If i is restricted to
positive integers, then by substituting (16) to (19) and with the help of
[24, Eq. (9.211/4)], an accurate closed-form approximation for Pe, i
can be obtained as in

Pe, i

m
i
 hi   m
i

i 1 i k1
  (m
i
Hi

i + r)
1

2
2(i )(m
i )(p)
r!

qp

k=0



r=0

Ap
i +1,
n, i an, k, i (r+p) r+p, p m

n=1

m
iq
i
An, i


.
(21)

2) Interference-Plus-Noise-Limited Systems: For interferenceplus-noise-limited systems, to obtain an analytical expression for


Pe, i , we restrict i to integer values only. Then, by substituting
(17)(19) and following a similar to the aforementioned case process,
an accurate closed-form approximation for Pe, i can be obtained as in
(22), shown at the bottom of the page.

Fig. 1. CDF of the sum of four and six i.n.i.d.squared variates.

a change of variables = tan2 and applying a Np -point Gauss


can be numerically evaluated in an
quadrature approximation, R
i
efficient manner as

 1 F (xn )
i
1
wn
R
i
ln 2
1 + xn
Np

(25)

n=1

D. Average Capacity Analysis


The average capacity of a DF dual-hop transmission system is given
by [33, Eq. (19)] as
EDF = min

1
E log2 (1 + 1 ) , E log2 (1 + 2 )
2
2



1
, R

min R
1
2
2

where the abscissas xn and the weights wn are defined in


[34, Eqs. (22) and (23)], respectively. It is noted that, to obtain a
sufficient approximation accuracy, Np 60 must be chosen in (25).
For integer values of i , by substituting (16)(24) and with the help of

[24, Eq. (3.259/3)], an accurate closed-form approximation for R


i
can be obtained as follows:

 hi i

(23)

R
i

where
= 1
R
i
ln 2

ln(1 + )fi ()d

(24)

1) Interference-Limited Systems: For interference-limited systems


and assuming arbitrary values of i , by substituting (15)(24), an
integral involving the Appell F1 function and rational functions has to
be evaluated. Unfortunately, such an integral is very difficultif not
impossibleto be solved in closed form. Nevertheless, by performing

Pe,i

2 


k=0

r=0

r+ m
i


Arn,i an,k,i 2 F1 r+ m
i , r+1; 1+r+ m
i ; 1

i An,i

m
i


.
(26)

1 Fi ()
d.
1+

(i ) ln 2

n=1

1
=
ln 2

  (m
i )r
i /

i 1 i k1

Hi

2) Interference-Plus-Noise-Limited Systems: For interference ,


plus-noise-limited systems, to obtain an analytical expression for R
i
i is restricted to integer values only. Then, by substituting (17)(24),
can be expressed as
R
i

m
i
 hi i  m
i

R
i

i 1 i k1 2
r
hi
   p

qp H
(m
i + j)(r + p)
1
i

An,i an,k,i
2 2(i )(m
i )(p)
j!(r j)!
k=0

r=0

n=1

j=0

Hi

(i )(m
i ) ln 2

m
i
i

 

i 1 i k1

k=0

r=0

1
r!

r+pj


r + p, p + r j m
i + 1,

m
i (q + An,i )
i
An,i


(22)

4114

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 8, OCTOBER 2013

TABLE I
VALUES OF PARAMETERS R EQUIRED FOR THE A PPROXIMATE F ORMULATIONS OF FIG. 1

Fig. 2. Error bound for the approximation of cdf of the sum of four and six
i.n.i.d. squared variates.

2



Arn, i an, k, i

n=1

r  

r
j=0


I

r, An, i ,

Fig. 3. OP of dual DF systems in interference-limited and interference-plusnoise-limited environments for various numbers of interferers I1 = I2 = I,
with I 2, as a function the first-hop average SNR, 1 .

(m
i + j)

m
i
, An, i , m
i +j
i


(27)

exp ()
d. This integral can be
where I(, , , , )  0 (+)
(1+)
numerically evaluated in an efficient manner by performing a change
of variables t2 = and applying a Q-point SI-GHQ.

V. N UMERICAL AND C OMPUTER S IMULATION R ESULTS


Using the analysis presented in Section IV, various numerical
performance evaluation results compared with equivalent Monte Carlo
simulations will be presented herein. To investigate the accuracy of
the proposed approximation, in Fig. 1, the exact and approximate cdf
of the sum of four and six i.n.i.d.squared variates is depicted.
The exact cdf has been obtained by means of Monte Carlo simulations
using 106 samples. An exponentially decaying power profile is adopted
to model power imbalance, so that  = exp(( 1)), with being
i and
the decaying factor. The required values for the parameters
m
i are presented in Table I. As it can be observed, for a given set
of fading parameters and number of summands N , the accuracy of
the proposed method is affected by the power imbalance. Specifically,
for N = 6, the best scenario, hereafter referred to as Case 1, occurs
when smaller fading severities (that corresponds to high values of i

and i ) are assigned to the signals with greater mean power. In this
case, the proposed approximations are practically indistinguishable
from the exact solutions. For N = 4, the best accuracy is obtained for
the balanced test case ( = 0). For the opposite scenario (the worstcase condition), hereafter referred to as Case 2, the approximation is
still very good.
Assuming Case 2 and balanced average power, the corresponding
error bound (), which is given by (11), is portrayed in Fig. 2 for
N = 4 and N = 6 summands. As it can be observed, () has a
Note that these findings
maximum, which equals unity when = .
are in agreement with those provided in [28].
Figs. 3 and 4 show the exact and approximate OP and average
capacity, respectively, of the considered DF system, as a function of
the first-hop average SNR 1 for various numbers of interferers. In
this case, it is assumed that 1 = 2, 1 = 0.15, 2 = 2, 2 = 0.25,
th = 2 dB, and 2 = 2 1 . Moreover, the transmit power values have
been normalized to unity. Both interference-limited and interferenceplus-noise-limited environments are considered. The parameters of the
interference channels have been set as {i,  }6=1 = { }6=1 = {0.1,
0.2, 0.2, 0.4, 0.5, 0.5}; {i,  }6=1 = { }6=1 = {1.5, 2.0, 2.0, 2.5, 3.0,
3.0}; and { i,  }6=1 = {  }6=1 = {3.2, 3.2, 2.5, 2.2, 2.2, 1.8}, i {1,
2}. The resulting approximation parameters to the pdf of Zi are
{m
, i }6=2 = {4.359269434,6.797382879,9.626675989,12.91629228,
, i }6 = {6.4, 8.9, 11.1, 13.3, 15.1}. Finally,
15.95021332} and {
=2
an equal number of interferers at both hops are assumed, i.e.,
I1 = I2 = I, with I 2. For interference-limited environments, the

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 8, OCTOBER 2013

4115

as it is evident, a very close match is observed. For the same system


under consideration, Fig. 5 shows the exact and approximate ABEP,
in both interference-limited and interference-plus-noise-limited
environments, when BPSK is employed. For interference-limited
environments, the approximate ABEP was obtained based on both
(20) and (21). When (20) is used for the evaluation of ABEP, a
value of Q = 5 is adequate to obtain a good match with (21). As
far as the accuracy of the proposed approximation is concerned,
similar findings to those reported in the previous test cases can be
verified.

VI. C ONCLUDING R EMARKS

Fig. 4. Average capacity of dual DF systems in interference-limited and


interference-plus-noise-limited environments for various numbers of interferers
I1 = I2 = I, with I 2, as a function the first-hop average SNR 1 .

Fig. 5. ABEP of dual DF systems employing BPSK in interference-limited


and interference-plus-noise-limited environments for various numbers of interferers I1 = I2 = I, with I 2, as a function the first-hop average
SNR 1 .

approximate OP and average capacity were evaluated using both the


closed-form expressions given by (16) and (26) and the numerical
integration methods given by (15) and (25). As it is evident, both
methods yield identical results. For the evaluation of the Appell
function, the algorithm provided in [25, App. B, pp. 295] was used.
Using this technique, it was observed that 22 terms are enough to
obtain sufficient numerical accuracy for a broad range of parameters.
Moreover, when (25) is used for the evaluation of the average capacity,
a value of NP = 115 is adequate to obtain a good match with (26)
for all tested cases. In addition, for the evaluation of (27), a 15-point
SI-GHQ was used for sufficient accuracy. The numerical results are
compared with equivalent results, which have been obtained using
Monte Carlo simulations with more than 106 random samples, and

In this paper, capitalizing on the fading model and assuming


a dual-hop DF relaying system with faded multiple cochannel
interferers at the both the relay and the destination, simple accurate
approximate expressions for the OP, the ABEP, and the average
capacity have been derived. A comparison of the analytical results
with Monte Carlo simulations shows that, for practical purposes, the
proposed approximations are accurate for a wide range of values of
fading parameters and number of interferers.
R EFERENCES
[1] J. N. Laneman, D. N. C. Tse, and G. W. Wornell, Cooperative diversity in
wireless networks efficient protocols and outage behaviour, IEEE Trans.
Inf. Theory, vol. 50, no. 12, pp. 30623080, Dec. 2004.
[2] H. A. Suraweera, H. K. Garg, and A. Nallanathan, Performance analysis
of two hop amplify-and-forward systems with interference at the relay,
IEEE Commun. Lett., vol. 14, no. 8, pp. 692694, Aug. 2010.
[3] S. S. Ikki and S. Assa, Multi-hop wireless relaying systems in the
presence of cochannel interference: performance analysis and design
optimization, IEEE Trans. Veh. Technol., vol. 61, no. 2, pp. 566573,
Feb. 2012.
[4] T. Soithong, V. A. Aalo, G. P. Efthymoglou, and C. Chayawan, Performance of multihop relay systems with co-channel interference in Rayleigh
fading channels, IEEE Commun. Lett., vol. 15, no. 8, pp. 836838,
Aug. 2011.
[5] C. Zhong, S. Jin, and K.-K. Wong, Dual-hop systems with noisy relay
and interference-limited destination, IEEE Trans. Commun., vol. 58,
no. 3, pp. 764768, Mar. 2010.
[6] D. B. da Costa and M. D. Yacoub, Outage performance of two hop AF
relaying systems with co-channel interferers over Nakagami-m fading,
IEEE Commun. Lett., vol. 15, no. 9, pp. 980982, Sep. 2011.
[7] D. B. da Costa, H. Ding, and J. Ge, Interference-limited relaying transmissions in dual-hop cooperative networks over Nakagami-m fading,
IEEE Commun. Lett., vol. 15, no. 5, pp. 503505, May 2011.
[8] D. Lee and J. H. Lee, Outage probability for dual-hop relaying systems
with multiple interferers over Rayleigh fading channels, IEEE Trans.
Veh. Technol., vol. 60, no. 1, pp. 333338, Jan. 2011.
[9] K. P. Peppas and C. Datsikas, Outage analysis of dual-hop relaying
communications with co-channel interference over Nakagami-m fading
channels, IEICE Trans. Commun., vol. E94-B, no. 8, pp. 24142418,
Jan. 2011.
[10] D. B. da Costa, H. Ding, M. D. Yacoub, and J. Ge, Two-way relaying in interference-limited AF cooperative networks over Nakagami-m
fading, IEEE Trans. Veh. Technol., vol. 61, no. 8, pp. 37663771,
Oct. 2012.
[11] M. D. Yacoub, The - and the - distribution, IEEE Antennas
Propag. Mag., vol. 49, no. 1, pp. 6881, Feb. 2007.
[12] J. Zhang, M. Matthaiou, Z. Tan, and H. Wang, Performance analysis
of digital communication systems over composite /gamma fading
channels, IEEE Trans. Veh. Technol., vol. 61, no. 7, pp. 31143124,
Sep. 2012.
[13] N. Ermolova, Moment generating functions of the generalized and
k distributions and their applications to performance evaluations of
communication systems, IEEE Commun. Lett., vol. 12, no. 7, pp. 502
504, Jul. 2008.
[14] N. Ermolova, Useful integrals for performance evaluation of communication systems in generalized - and - fading channels, IET
Commun., vol. 3, no. 2, pp. 303308, Feb. 2009.

4116

IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 8, OCTOBER 2013

[15] N. Y. Ermolova and O. Tirkkonen, The fading distribution with


integer values of , IEEE Trans. Wireless Commun., vol. 10, no. 6,
pp. 19761982, Jun. 2011.
[16] K. P. Peppas, F. Lazarakis, T. Zervos, A. Alexandridis, and K. Dangakis,
Sum of non-identical independent squared - variates and applications
in the performance analysis of DS-CDMA systems, IEEE Trans. Wireless
Commun., vol. 9, no. 9, pp. 27182723, Sep. 2010.
[17] D. Morales-Jimenez and J. F. Paris, Outage probability analysis for -
fading channels, IEEE Commun. Lett., vol. 14, no. 6, pp. 521523,
Jun. 2010.
[18] D. Morales-Jimenez, J. F. Paris, and A. Lozano, Outage probability
analysis for MRC in fading channels with co-channel interference,
IEEE Commun. Lett., vol. 16, no. 5, pp. 674677, May 2012.
[19] J. Paris, Outage probability in / and /
interference-limited scenarios, IEEE Trans. Commun., vol. 61, no. 1,
pp. 335343, Jan. 2013.
[20] A. C. Moraes, D. B. da Costa, and M. D. Yacoub, An outage analysis
of multibranch diversity receivers with cochannel interference in ,
, and fading scenarios, Wireless Pers. Commun., vol. 64,
no. 1, pp. 319, May 2012.
[21] W.-G. Li, H.-M. Chen, and M. Chen, Outage probability of dual-hop
decode-and-forward relaying systems over generalized fading channels,
Eur. Trans. Telecommun., vol. 21, no. 1, pp. 8689, Jan. 2010.
[22] K. P. Peppas, F. Lazarakis, A. Alexandridis, and K. Dangakis, Momentsbased analysis of dual-hop amplify-and-forward relaying communications
systems over generalised fading channels, IET Commun., vol. 6, no. 13,
pp. 20402047, Sep. 2012.
[23] K. P. Peppas, G. C. Alexandropoulos, and P. T. Mathiopoulos, Performance analysis of dual-hop AF relaying systems over mixed
and fading channels, IEEE Trans. Veh. Technol., vol. 62, no. 7,
pp. 31493163, Sep. 2013.
[24] I. Gradshteyn and I. M. Ryzhik, Tables of Integrals, Series, and Products,
6th ed. New York, NY, USA: Academic, 2000.
[25] H. Exton, Multiple Hypergeometric Functions and Applications,
G. M. Bell, Ed. Sussex, U.K.: Ellis Horwood, 1976.
[26] J. C. S. S. Filho and M. D. Yacoub, Nakagami-m approximation to the
sum of m non-identical independent Nakagami-m variates, Electron.
Lett., vol. 40, no. 15, pp. 951952, Jul. 2004.
[27] N. Y. Ermolova and O. Tirkkonen, Distribution of diagonal elements of
a general central complex Wishart matrix, IEEE Commun. Lett., vol. 16,
no. 9, pp. 13731376, Sep. 2012.
[28] F. Berggren, An error bound for moment matching methods of lognormal
sum distributions, Euro. Trans. Telecomms., vol. 16, no. 6, pp. 573577,
Nov./Dec. 2005.
[29] A. P. Prudnikov, Y. A. Brychkov, and O. I. Marichev, Integrals and Series
Volume 4: Direct Laplace Transforms, 1st ed. Boca Raton, FL, USA:
CRC, 1992.
[30] D. B. da Costa and S. Assa, Dual-hop decode-and-forward relaying
systems with relay selection and maximal-ratio schemes, Electron. Lett.,
vol. 45, no. 9, pp. 460461, Apr. 2009.
[31] I. S. Ansari, S. Al-Ahmadi, F. Yilmaz, M.-S. Alouini, and
H. Yanikomeroglu, A new formula for the BER of binary modulations
with dual-branch selection over gereralized-K composite fading channels, IEEE Trans. Commun., vol. 59, no. 10, pp. 26542658,
Oct. 2011.
[32] N. M. Steen, G. D. Byrne, and E. M. Gelbard, Gaussian quadratures for
x2
b x2
the integrals
e
f (x)dx and
e
f (x)dx, Math. Comput.,
0
0
vol. 23, no. 107, pp. 661671, 1969.
[33] G. Farhadi and N. C. Beaulieu, On the ergodic capacity of multi-hop
wireless relaying systems, IEEE Trans. Wireless Commun., vol. 8, no. 5,
pp. 22862291, May 2009.
[34] F. Yilmaz and M.-S. Alouini, An MGF-based capacity analysis of equal
gain combining over fading channels, in Proc. IEEE PIMRC, Sep. 2010,
pp. 945950.

Bit-Error-Rate Performance of Companding


Transforms for OFDM
Yasir Rahmatallah, Member, IEEE,
Nidhal Bouaynaya, Member, IEEE, and
Seshadri Mohan, Member, IEEE

AbstractThis paper provides a comprehensive analytical framework


to assess the relative bit-error-rate (BER) performance of companding
transforms (CTs) employed to reduce the peak-to-average-power ratio
(PAPR) in orthogonal frequency-division multiplexing (OFDM) systems.
This paper provides a quantitative basis for several claims, which are
reported in the literature, based solely on simulation results. In particular,
we consider three main classes of CTs and provide a set of necessary and
sufficient conditions for the superiority of one CT relative to the others.
The conditions are given in terms of the companding parameters, which
are usually selected to achieve a target PAPR. Our analytical derivations
are supported by simulation results.
Index TermsCompanding transforms (CTs), orthogonal frequencydivision multiplexing (OFDM), peak-to-average power ratio (PAPR).

I. I NTRODUCTION
Despite the significant advantages offered by orthogonal frequencydivision multiplexing (OFDM), it has the major inherited drawback
of fluctuating envelope with high peaks, which leads to a high peakto-average-power ratio (PAPR) for the transmitted signal. High peaks
drive the transmitters power amplifier (PA) into the nonlinear or saturation regions of operation, hence causing distortions and out-of-band
radiation. They also demand analog-to-digital converters (ADC) with
wide dynamic ranges. Many PAPR reduction techniques have been
proposed in the literature, such as clipping and filtering, companding
transforms (CTs), selective mapping, partial transmit sequences, tone
injection, tone reservation, and linear block coding [1][3]. PAPR
reduction capability is usually measured by the empirical complementary cumulative distribution function (ccdf), which is defined as the
probability that the signals PAPR exceeds a specific threshold. In most
methods, PAPR is reduced at the expense of increasing the bit error rate
(BER), complexity, or data overhead.
CTs form an attractive and widely used PAPR reduction technique
due to their flexibility and low complexity, regardless of the number
of subcarriers in the OFDM signal. CTs attenuate the high peaks
and amplify the low amplitudes, thus decreasing the PAPR of the
signal prior to the PA. However, CTs increase the BER due to the
distortion incurred by the modulating symbols at the transmitter and
the expansion of channels noise by the decompander at the receiver.
Fig. 1 shows the block diagram of an OFDM transceiver with a
compander inserted between the parallel-to-serial converter and the

Manuscript received August 27, 2012; revised December 27, 2012 and
February 19, 2013; accepted May 3, 2013. Date of publication May 15, 2013;
date of current version October 12, 2013. This work is supported by the
National Science Foundation under Grant EPS-0701890. The review of this
paper was coordinated by Dr. A. J. Al-Dweik.
Y. Rahmatallah is with the Division of Biomedical Informatics, University
of Arkansas for Medical Sciences, Little Rock, AR 72205 USA (e-mail:
yrahmatallah@uams.edu).
N. Bouaynaya and S. Mohan are with the Department of Systems Engineering, University of Arkansas at Little Rock, Little Rock, AR 72204 USA (e-mail:
nxbouaynaya@ualr.edu; sxmohan@ualr.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TVT.2013.2263233

0018-9545 2013 IEEE

You might also like