You are on page 1of 7

Frequenz 2014; 68(9–10): 489 – 495

Jelena A. Anastasov*, Goran T. Djordjevic, Stefan R. Panic and Mihajlo C. Stefanovic

Evaluations of SSC Diversity Receiver over EGK


Fading Channels
Abstract: In this paper, a detailed performance analysis propagation. In the literature, there are a lot of exploited
of  switch-and-stay combining receivers over non identi- models for describing transmission over mentioned con-
cal  distributed extended generalized K fading channels ditions. For example Rayleigh, Ricean, Nakagami-m and
is  presented. The analysis has been performed in both Weibull distributions fit well in describing multipath
noise and interference-limited environments (the level of fading channel conditions, and Gamma distribution is a
interference or noise is ignored, respectively). The expres- good approximation for the lognormal-shadowed chan-
sions for the output signal-to-noise (SNR) and signal-to-­ nels [1]. To depict the composite fading channels, K [2]
interference (SIR) probability density function and cumu- and  generalized-K (GK) [3] distributions as acceptable
lative distribution function, in a form of Fox’s H functions, solutions were proposed. More recently, the extended
are presented. Based on this, analytical expressions for ­generalized-K (EGK) distribution [4] was approved as a
evaluating the moments, outage probability, average bit distribution which includes all important distributions as
error rate and average channel capacity are derived. The its special or limiting cases and can be observed as very
influence of fading and shadowing phenomena, as well general and convenient for describing great variety of
as  the influence of unbalanced input SNR/SIR on the fading/shadowing channel conditions. Summary of distri-
most  important performance metrics is obtained. Pre- butions that can be obtained from the EGK distribution
sented numerical results are confirmed by Monte Carlo are given in [4, Table I].
simulations. In most of wireless communication systems Gaussian
noise is a front-end source of the performance impair-
Keywords: co-channel interference, diversity, error proba- ment. The systems that are under dominant noise influ-
bility, fading, noise, shadowing ence are called noise-limited (NL) ones. Apart from this
fact, in cellular communication systems, the level of
PACS® (2010). 84.40.Ua noise  can be ignored because of dominant influence of
co-channel interference (CCI) imposed by the neighboring
DOI 10.1515/freq-2014-0019 cells using the same frequency. The severe CCI in the cell-
Received February 11, 2014. edge area can considerably degrade the system perfor-
mance [1]. Those are so-called interference-limited (IL)
systems. In diminishing fading influence in both NL and
1 Introduction IL systems, space diversity techniques are shown to be
very good solution. Among them the switch-and-stay com-
Signal transmission in wireless communication systems is bining (SSC) technique is the most exploited one as a less
often simultaneously corrupted by multipath fading and expensive and with the lowest complexity nature. The SSC
shadowing phenomena. The main cause of the fading ex- receiver selects a particular branch until the value of the
istence is delaying, reflecting and scattering of the signal signal-to-noise ratio (SNR) or signal-to-interference ratio
components, opposed to the shadowing phenomena (SIR) of that branch (in NL or IL environment, respec-
caused by the presence of many ambient objects during tively) drops below a predetermined threshold. Then the
combiner switches to another branch and stays there re-
gardless of the SNR/SIR value of that branch [1].
*Corresponding author: Jelena A. Anastasov: Department of The performance analysis of SSC receiver over gener-
Telecommunications, Faculty of Electronic Engineering, University alized Gamma channels in NL environment was proposed
of Nis, Nis, Serbia. E-mail: jelena.anastasov@elfak.ni.ac.rs in [5]. Also, brief performance evaluations for the output
Goran T. Djordjevic, Mihajlo C. Stefanovic: Department of
SNR of many diversity receivers including SSC receiver,
Telecommunications, Faculty of Electronic Engineering, University
of Nis, Nis, Serbia
operating over GK fading channels are presented in [6]. A
Stefan R. Panic: Faculty of Natural Science and Mathematics, lot of papers were published relaying on the performance
University of Pristina, Kosovska Mitrovica, Serbia analysis in environment corrupted by CCI [7].

Brought to you by | Umea University Library


Authenticated | 10.248.254.158
Download Date | 9/1/14 2:57 AM
490   J. A. Anastasov et al., Evaluations of SSC Diversity Receiver over EGK Fading Channels

As we mentioned before EGK distribution is the most In the presence of the additive white Gaussian noise
general distribution. The analytical expressions of the (AWGN), the instantaneous SNR is denoted by G = X 2 / N 0,
­important statistic measures for the system operating over where N0 represents the power spectral density of
the EGK fading and Gaussian noise were given in [4]. Some AWGN,  while the average SNR is expressed as= γ E[=G]
elementary performance evaluations over EGK single E[ X 2 ]/ N 0 = Ω / N 0, where E[.] denotes the expectation
channel IL environment were presented in [8]. Based on operator. According to this and starting from (1), the
­
the expressions derived in [4] and [8], some new evalua- PDF  of the instantaneous SNR can be obtained in the
tions are performed in this paper, and the SSC receiver form
in  both NL and IL EGK environment is analyzed for the
first time. The expressions for evaluating the moments of  − ,− 
1  β sβ 
the SSC output SNR/SIR, outage probability, average bit ()
pG γ = H 2,0
 γ 1   1   . (2)
error rate (ABER) for some binary modulation formats and ( ) ( )
Γ ms Γ m γ
0,2
 γ  ms ,  ,  m,  
 ξs   ξ  

average channel capacity are given. Some of the expres-
sions, derived in this paper, can be simplified into already
Starting from the basic definition, mn = E γ n  , the nth
published results. The analytical results are confirmed by
order moment of γ, can be derived by using [11, eq. (2.8)] in
Monte Carlo simulations.
the following form
The rest of the paper is organized as follows. In Sec­
tion 2, the basic statistic measures of single NL and IL
 n  n
EGK channels are given. The detailed performance analy- Γ  ms +  Γ  m +  n
 ξs   ξ  γ 
sis of SSC receiver over mentioned channel conditions is mn =
NL
  . (3)
reported in Section 3. In Section 4, numerical and simula- ( ) ( )
Γ ms Γ m  β sβ 
tion results are discussed and conclusions are offered in
Furthermore, we obtained the CDF of the instanta-
Section 5.
()
γ

()
neous SNR, FG γ = ∫ pG t dt , by using [10, eq. (2.8.17)]
0

2 Channel models and simplifying the Fox’s H function using [10, eq. (2.1.9)],
in the following form

2.1 Noise-limited environment
1

 (1,1) 

()
FG γ = 2,1 β s β
H 1,3   1   1 .
( ) ( ) ( )
γ
Let us observe an environment where the desired signal Γ ms Γ m  γ  ms ,  ,  m,  , 0,1 
  ξs   ξ  
envelope, X, experiences EGK fading. The PDF of the de­ (4)
sired signal envelope considering [4, eq. (2), with correc-
tion of one typo] and using transformation given in [9,
eq. (6.22)], can be described as
2.2 Interference-limited environment
 − ,− 
2  
()
pX x = 2,0 β s β 2
H 0,2  x  1   1  , (1)
 In the IL environment, both the desired signal envelope,
( ) ( )
Γ ms Γ m x  Ω m ,
 s 

, m,
ξ s   ξ  
 R, and interfering signal envelope, r, have EGK distri­

butions. The PDF of the instantaneous SIR, defined as
where parameters m and ξ represent the fading order z = R 2 / r 2, is given by [8, eq. (6)]
and  fading shaping factor, respectively, while ms and ξs
represent the shadowing severity and shadowing shaping 1
()
pz z =
factor, respectively. The average signal power is denoted ( ) ( ) ( ) ( )
Γ msd Γ md Γ msc Γ mc z
by Ω, β = ( ) ( )
Γ m + 1 /ξ / Γ m , βs = ( ) ( )
Γ ms + 1 / ξ s / Γ ms , and   1  1 
()
Γ . is the Gamma function [9, (1.1)]. The Fox’s H function, 
2,2  β sd β d
 1 − msc ,  ,  1 − mc ,  
 ξ sc   ξc  
, (5)
(
 ai , Ai )  × H 2,2 β β Sz  1  1  
defined as [10, eq. (1.1.2)], is denoted by H mp ,,qn  z 1, p 
,  sc c m
 sd , ,
 dm ,  

j(
 b , B
j ) 
1,q 

  ξ sd   ξd  
where p, q, m, n are integers such that 0 ≤ m ≤ q , 0 ≤ n ≤ p;
ai , b j ∈ C , where C is the set of complex numbers, and where the previously defined parameters in Eq. (1) are
( )
Ai , B j ∈ R+ = 0, ∞ , (i = 1, p; j = 1, q ). indexed by d for the desired signal and c for interfer-

Brought to you by | Umea University Library


Authenticated | 10.248.254.158
Download Date | 9/1/14 2:57 AM
 J. A. Anastasov et al., Evaluations of SSC Diversity Receiver over EGK Fading Channels   491

ing  one, and S =Ω d / Ω c represents value of the average  P1 P2 2


SIR.
 ∑ pΔ i Δ , ( ) 0 < Δ ssc ≤ ΔT
 P1 + P2 i=1
We derived the nth order moment of the instanta- pΔ ssc ( )
Δ = 2  p Δ ( )  , ,
 P1 P2 ( )
neous SIR by using [11, eq. (2.8)], in the following form ∑  p Δ + Δi
 P1 + P2 i=1  Δi Pi 
Δ ssc > ΔT
  
(9)

Γ  msd +
n   n  n   n
 Γ  md +  Γ  msc −  Γ  mc −  ( )
where pΔi Δ is defined as Eq. (2) or (5) in NL or IL envi-
mnIL = 
ξ sd   ξd   ξ sc   ξc  ronment, respectively.
( ) ( ) ( ) ( )
Γ msd Γ md Γ msc Γ mc The nth order moment of the output SNR is derived
β β S
n in  the closed-form with the aid of Eq. (3) and by using
×  sc c  (6) [10, eqs. (2.8.17) and (2.1.9)], in the following way:
 β sd β d 

P1 P2 2  NL mnNL 1 n 2,1
The CDF of the instantaneous SIR, is obtained by mnNL ,SSC
= ∑ mn + P − P ΔT H 1,3
P1 + P2 i=1  i i
using [10, eq. (2.8.17)] and simplifying the Fox’s H func-
tion  using the relation [10, eq. (2.1.9)], in the following

β β (1 − n,1) 

×  si i ΔT  1  1  (10)
form [8, eq. (7)]  γ i  msi ,  ,  mi ,  , ( )
−n,1 

  ξ si   ξ i  
1
()
Fz z = The previous equation can be reduced into [6, eq. (9)]
( ) ( ) ( ) ( )
Γ msd Γ md Γ msc Γ mc
for the case of SSC receiver over GK fading channels
  1  1 


( 1,1 ) ,  1 − m sc ,
 ,  1 − mc ,  
ξc  
as follows. By setting ξ = ξs = 1 into Eq. (3), mnNL is simpli-
β
× H 3,3
β  ξ sc   fied into [3, eq. (5)]. Also, by setting ξ = ξs = 1 in Eq. (4) and
β β Sz  .
2,3 sd d

1  1 
 sc c

m
 sd , ,
 dm ,  , ( )
0,1 

using [9, eq. (6.108)] the Fox’s H function is simplified
  ξ sd   ξd   into  Meijer’s G function and forward, by using [13, eq.
(7) (07.34.03.0871.01)], the Meijer’s G function is reduced into
the regularized hypergeometric function, which simplifies
CDF, Pi, into [3, eq. (3)]. Finally, the Fox’s H function in

3 Performance analysis of SSC Eq. (10) can be simplified into the regularized hypergeo-
metric function by using [13, eq. (07.34.03.0727.01)], which
receiver in NL/IL environment simplifies the derived Eq. (10) into [6, eq. (9)].
We also derived the nth order moment of output SIR
Let us denote instantaneous SNR or SIR at the output of by using Eq. (6) and following the same procedure as for
a  SSC receiver over the channel corrupted only by noise deriving Eq. (10), in the form
or  interference, respectively, as Δssc. The predetermined
switching threshold is denoted by ΔT. Then, the CDF of Δssc
P1 P2 2  IL mnIL 1 n 2,3
for the proposed analysis in NL/IL environment, for un-= mnIL ,SSC ∑ mn + P − P ΔT H 3,3
P1 + P2 i=1 
correlated branches can be employed as follows [1, eq. i i

  1  1  
(9.291)]:  (
1 − n,1 ,  1 − msci ,)  ,  1 − mci ,  
ξci  
×  si i ΔT 
β β ξ sci  
 1  1   .
( )
γi
 P1 P2 2 
  msdi ,  ,  mdi ,  , −n,1 

 ∑ FΔi Δ ,( ) 0 < Δ ssc ≤ ΔT   ξ sdi   ξdi  
 P1 + P2 i=1 (11)
( )
FΔ ssc Δ = 2  F Δ ( ) ,
 P1 P2 ( )
 P1 + P2 ∑
 FΔ Δ + Δi − 1  Δ ssc > ΔT
i =1 
i
Pi  Derived Eqs. (10) and (11) are novel and can be
  
(8) straightforwardly used for evaluating the average output
SNR/SIR and amount of fading.
( )
where FΔi Δ is defined following Eqs. (4) and (7) for both As one of the most important performance metrics,
considered environments; and = ( )
Pi FΔi ΔT . Differentiat- the outage probability, Pout, is the main focus in this paper.
ing Eq. (8) with respect to Δ, the PDF of SSC forwardly can The Pout is defined as a probability that the value of Δssc
be obtained as falls below given outage threshold q. Using Eq. (8) as

Brought to you by | Umea University Library


Authenticated | 10.248.254.158
Download Date | 9/1/14 2:57 AM
492   J. A. Anastasov et al., Evaluations of SSC Diversity Receiver over EGK Fading Channels

() ()
Pout q = FΔ ssc q , this probability can be obtained in NL/IL The finite integral in Eq. (14) can be evaluated via nu-
environment. merical integration similarly as suggested in Eqs. [3] and
Furthermore, the ABER at the output of SSC receiver, [6]. The optimum threshold, ΔT*, for min{ Pe } can be found
Pe , can be obtained by averaging the conditional error by solving the equation ∂Pe / ∂ΔT Δ =Δ* = 0. Also, it can be
T T
probability over the PDF of the output Δssc as numerically evaluated by using root-finding techniques

available in Mathematica or Matlab software package.
Pe= ∫ P ( Δ ) p ( Δ ) dΔ .
e Δ ssc (12) Besides these performance measures, the average
0 normalized channel capacity, C e / W , with W as the trans-
mission bandwidth, is also considered in the paper. This
For binary modulation formats, the conditional error
capacity is obtained by averaging Shannon capacity over
probability has the following form [1]:
the PDF (Eq. (9)), in the following way:

Pe ( Δ ) =
(
Γ b, aΔ
,
) (13) ∞

2 Γ(b) C e=
/W ∫ log (1 + λ ) p ( Δ ) dΔ .
2 Δ ssc (17)
0

( )
where Γ α ,x denotes the complementary incomplete
Substituting Eq. (9) in Eq. (17), the C e / W is derived in
Gamma function, [9, eq. (2.2)], and a and b are param­
the following way:
eters: (a, b) = (1, 0.5) for the binary phase shift keying
(BPSK), (a, b) = (0.5, 0.5) for the coherent binary frequency
P1 P2 2  CΔ 
Δ
1 T
shift keying (BFSK) and (a, b) = (1, 1) for differential BPSK
= Ce / W ∑ C Δi + i − ∫ log 2 1 + Δ pΔi Δ dΔ  ,
P1 + P2 i=1  Pi Pi 0 
( ) ( )
(DBPSK). Substituting Eq. (9) in Eq. (12), we get 
(18)
2  P ΔT 
P1 P2 1
= Pe
Δ
( ) ( )
∑  PΔ + i − ∫ Pe Δ pΔi Δ dΔ  , (14) where C∞Δi is the average capacity of the ith channel, given
(
2 P1 + P2 i=1  i Pi Pi 0 ) 
by C Δi ∫ log 2 1 + Δ pΔi Δ dΔ . If we consider the NL case,
= ( ) ( )
where PΔi is the ABER at the ith input branch, PΔi = 0
∞ substituting Eq. (2) in the previous definition of C Δi and
( ) ( )
∫ Pe Δ pΔi Δ dΔ. Using the representation of the using the transform of logarithmic function into the Fox’s
( )  and also using
0
Gamma  function into the Meijer’s G, e.g. into H  1,1
H function ln b,= (
aΔ H 1,22,0 
)

 1,1  ( ) 

0,1 , b,1 ( ) ( ) 
function Γ b,=aΔ H 1,2 (
2,0 
aΔ ) , [13, eqs.


0,1 , b,1 
 ( )( ) [10, eq. (2.8.12)], we get [4, eq. (15)]

(06.06.26.0005.01), (07.34.26.0008.01)], and employing


1
[10, eq. (2.8.12)], we derived the expressions for PΔi in the C ΔNL =
case of NL environment in the form
i
() ( ) ( )
ln 2 Γ msi Γ mi

 β si β i ( )( )
0,1 , 1,1 

1 ×H
PΔNL =
4,1
  1  1 . (19)
i
( ) ( ) ()
2 Γ msi Γ mi Γ b
2,4
m ,
 i  si ξ si   i ξi 
γ ( )( )
, m , , 0,1 , 0,1 

 
  1  1 
×H 2,2
 γi
a
 1 − msi ,  ,  1 − mi ,  , 1,1 ( )
 , (15)
3,2  ξ si   ξi 
On the other hand, in the case of IL environment, substi-
 β si β i
 ( )( )
0,1 , b,1 
 tuting Eq. (5) in the definition of C Δi , and following the
similar procedure as for NL case, we derive
and in the case of IL environment in the form

1 1
PΔIL = C ΔIL =
i
( ) ( ) (
2 Γ msdi Γ mdi Γ msci Γ mci Γ b ) ( ) () i
() ( ) ( ) (
ln 2 Γ msdi Γ mdi Γ msci Γ mci ) ( )
  1  1     1  1  
  1 − msdi , ( )
 ,  1 − mdi ,  , 1,1    1 − msci , ( )( )
 ,  1 − mci ,  , 0,1 , 1,1 
4,2  β sci β ci Si 4,3  β sdi β di  .
 ξ sdi   ξdi  . ξ sci   ξci 
× H 3,4 × H 4,4
a β β  1  1  β β S  1   1  


sdi di
( )( )
0,1 , 0, b ,  msci ,  ,  mci ,    sci ci i  msdi ,
 ( )( )
 ,  mdi ,  , 0,1 , 0,1 
  ξ sci   ξci     ξ sdi   ξdi  
(16) (20)

Brought to you by | Umea University Library


Authenticated | 10.248.254.158
Download Date | 9/1/14 2:57 AM
 J. A. Anastasov et al., Evaluations of SSC Diversity Receiver over EGK Fading Channels   493

New derived Eq. (20) coincides with [14, eq. (34)] for
one dominant interfering signal (N = 1), by setting ξdi =
ξci = ξsdi = ξsci = 1, when the Fox’s H function is simplified
into the Meijer’s G function by using [9, eq. (6.108)]. This
also confirms generality of EGK analysis. By combining
Eqs. (18), (19) and (20), one can determine the capacity of
the SSC system over EGK fading channels.

4 Numerical results
The analytical expressions derived in the paper are given
in the form of the Fox’s H function. Presented numerical
results are performed in Mathematica 7, using transform
of the Fox’s H function into the Meijer’s G function [15,
p. 531, eq. (22)]. The numerical values of Fox’s H function Fig. 2: Pout in NL environment as a function of the normalized
can be also computed by applying the algorithm presented switching threshold
in [16]. The Monte Carlo simulations are performed in C++.
The EGK fading envelope is simulated by using [8, eq. (9)].
The Pout dependence on the normalized average input
SNR, γ / q (γ=1 γ=
2 γ ), of SSC receiver in NL environment
for different values of switching threshold ΔT is presented
in Fig. 1. The curve relating ΔT = q is lower compared to the
curves relating ΔT ≠ q . For Pout = 10−4, when ΔT = q/2, the
diversity gain compared with single channel transmission
is only about 1.3 dB. Opposite to this, when ΔT = q this di-
versity gain compared with single channel transmission is
even about 13.3 dB.
The considerable effect of proper selection of switch-
ing threshold is additionally illustrated in Fig. 2. This
Figure depicts Pout versus the normalized switching thresh-
old ΔT/q. It is evident that there exists an optimal switch-

Fig. 3: Pout of unbalanced receiver in IL environment versus the


normalized average SIR at the first branch

ing threshold that minimizes Pout. This optimal switching


threshold is equal to the protection ratio q. For that case,
SSC receiver can be considered as SC receiver. In order to
verify analytical results, Monte Carlo simulations were
performed and for comparison purposes were included in
this figure. The Pout is estimated using 2 × 109 generated
samples. It is evident that there is a very good agreement
between analytical and simulation results.
Fig. 3 presents the effect of branch unbalancing on
the outage performance in IL environment for different
values of the fading/shadowing severity parameters. It is
Fig. 1: Pout in NL environment as a function of the normalized average evident that outage performance could be considerably
SNRs degraded due to the branch unbalancing. The results

Brought to you by | Umea University Library


Authenticated | 10.248.254.158
Download Date | 9/1/14 2:57 AM
494   J. A. Anastasov et al., Evaluations of SSC Diversity Receiver over EGK Fading Channels

Fig. 4: ABER in NL environment versus the average input SNR at the Fig. 5: ABER in NL environment versus the switching threshold
first branch

show that this degradation does not depend on fading/


shadowing severity parameters. For example, to maintain
Pout of 10−5 the required value of the normalized average
input SIR, S1/q, in the case of unbalanced SIRs at the input
(S2 = S1/4) is about 3 dB greater then in the case of balanced
SIRs (S2 = S1) for all three values of fading/shadowing pa-
rameters. In addition, it is confirmed that the Pout values de-
crease with increase in fading/shadowing shaping factors.
Fig. 4 presents the ABER dependence on the average
input SNR at the first branch γ 1 for BPSK and DBPSK sig-
naling and randomly picked switching threshold ΔT = 8 dB.
In this figure we analyzed the influence of balanced and
unbalanced SNRs on ABER performance. The unbalanced
case is considered through the exponentially decaying
power delay profile γ 2 = γ 1 e −0.5 as in [6]. It is obvious that
the balanced input gives the lowest ABER values. At
ABER of 10−5, for both BPSK and DBPSK, the branch un- Fig. 6: Channel capacity in IL environment versus the average SIRs
balancing produce the degradation in γ 1 of about 1 dB
and  1.6 dB in the case of ξ1 = ξ2 = ξs1 = ξs2 = 1 and ξ1 = ξ2 =
ξs1 = ξs2 = 2, respectively. The more the values of fading/ minimum ABER sense and its value depends on the
shadowing p ­arameters, the greater are the effects of average input SNR, as well as the fading/shadowing pa-
branch unbalancing on ABER performance. Also, Fig. 3 rameters. Presented numerical results illustrate that the
shows better performance of BPSK modulation scheme optimal switching threshold increases with γ= 1 γ=
2 γ in-
compared to DBPSK scheme under the same conditions. creasing and fading/shadowing parameters increasing.
As expected, the diversity gain is larger for less value of To confirm the accuracy of the analytical approach we
fading/shadowing parameters. The simulation results are have included simulation results in the same figure.
in agreement with analytical ones. An ABER value is esti- Fig. 6 depicts the channel capacity versus the average
mated on the basis of 3 × 103 bit errors, while the maximum SIR (S1 = S2 = S) for different values of the shaping param­
number of bits used in simulation is 2 × 109. eters of fading and shadowing (bottom horizontal and left
Fig. 5 presents the ABER dependence on the switch- vertical axis). As expected, the system capacity increases
ing threshold for DBPSK modulation scheme in NL envi- with S increasing. In addition, the more the value of
ronment. There is an optimal switching threshold in the shaping parameters the more the value of capacity is.

Brought to you by | Umea University Library


Authenticated | 10.248.254.158
Download Date | 9/1/14 2:57 AM
 J. A. Anastasov et al., Evaluations of SSC Diversity Receiver over EGK Fading Channels   495

These dependences are obtained for the optimal switch- References


ing threshold ( ΔT*) that is computed numerically by solving
∂C e / ∂ΔT Δ =Δ* = 0. The second layer of this figure (top hor- [1] M. K. Simon and M. S. Alouni, Digital Communication over
T T
izontal and right vertical axis) presents the capacity de- Fading Channels, 2nd ed. New York: Wiley, 2004.
pendence on switching threshold. The results show that [2] A. Abdi and M. Kaveh, “K distribution: an appropriate
substitute for Rayleigh-lognormal distribution in fading-
the capacity is considerably influenced by value of switch-
shadowing wireless channels,” Electronic Letters, vol. 34,
ing threshold. The optimal switching threshold increases no. 9, pp. 851–852, April 1998.
with increasing S and also with increasing the values of [3] P. S. Bithas, N. C. Sagias, P. T. Mathiopoulos, G. K.
the shaping parameters. Numerical results obtained based Karagiannidis and A. A. Rontogiannis, “On the performance
on Eq. (18) are presented with the results obtained by analysis of digital communications over Generalized-K fading
channels,” IEEE Communications Letters, vol. 5, no. 10,
Monte Carlo integration [17]. The Monte Carlo integration
pp. 353–355, May 2006.
is performed using 109 samples. The good match between [4] F. Yilmaz and M. S. Alouini, “A new simple model for composite
these results is evident. fading channels: second order statistics and channel capacity,”
in 7th International Symposium on Wireless Communication
Systems (ISWCS 2010), pp. 676–680, Sep. 2010.
4 Conclusions [5] N. C. Sagias and P. T. Mathioulos, “Switched diversity receivers
over generalized Gamma fading channels,” IEEE
Communications Letters, vol. 9, no. 10, pp. 871–873, Oct. 2005.
In this paper, we have derived analytical expressions for [6] P. S. Bithas, P. T. Mathiopoulos and S. A. Kotsopoulos,
the moments, outage probability, error probability and “Diversity reception over generalized-K (KG) fading channels,”
spectral efficiency for unbalanced SSC receiver over the IEEE Transaction on Wireless Communications, vol. 6, no. 12,
EGK fading separately in the case of IL and NL environ- pp. 4238–4243, Dec. 2007.
[7] I. M. Kostic, “Analytical approach to performance analysis for
ment. Novel analytical expressions can be efficiently used
channel subject to shadowing and fading,” IEE Proceedings,
for performance evaluation for various types of fading vol. 152, no. 6, pp. 821–827, Dec. 2005.
­environments that are special or limiting cases of the EGK [8] J. A. Anastasov, G. T. Đorđević and M. Č. Stefanović, “Analytical
distribution. In the case of special values of fading param- model for outage probability of interference-limited systems
eters, some newly derived analytical expressions can be over extended generalized-K fading channels,” IEEE
simplified to the previously reported ones in the literature. Communications Letters, vol. 16, no. 4, pp. 473–475, Apr. 2012.
[9] M. A. Chaudhry and S. M. Zubair, On a Class of Incomplete
Numerical results, obtained on the basis of analytical ap-
Gamma Functions with Applications. Boca Raton: Chapman &
proach, are verified by Monte Carlo simulations. Hall/CRC, 2002.
The results have shown that the outage probability, [10] A. Kilbas and M. Saigo, H-Transforms: Theory and Applications.
over the observed EGK fading in NL environment, is con- Boca Raton, FL: CRC Press LLC, 2004.
siderably influenced by switching threshold value. Optimal [11] A. M. Mathai, R. K. Saxena and H. J. Haubold, The H-Function:
Theory and Applications, 1st ed. New York: Springer Science,
switching threshold which minimizes the outage proba-
2009.
bility is equal to the protection ratio. Numerical results [12] C. W. Helstrom, Probability and Stochastic Processes for
have also indicated that the outage probability is de- Engineers, 2nd ed. New York: MakMillian, 1991.
graded  by the receiver branches unbalancing and this [13] The Wolfram Functions site, 2008. [Online] Available:
degradation does not depend on fading parameters in IL http://functions.wolfram.com
environment. Presented results have illustrated the ABER [14] I. Trigui, A. Laourine, S. Affes and A. Stephenne,
“Performance analysis of mobile radio systems over composite
degradation due to the receiver branches unbalancing in
fading/shadowing channels with co-located interference,”
NL environment. The results have shown that effects of IEEE Transactions on Wireless Communications, vol. 8, no. 7,
branch unbalancing on ABER performance are greater pp. 3448–3453, Jul. 2009.
when fading/shadowing parameters are greater. The ABER [15] A. P. Prudnikov, Y. A. Brychkov and O. I. Marichev, Integral
and capacity are strongly dependent on the switching and Series: Part 3, More Special Functions. NAUKA Publisher,
1962 (in Russian).
threshold. Also, it can be concluded that optimum switch-
[16] F. Yilmaz and M. S. Alouini, “Product of the powers of
ing threshold in the minimum ABER or maximum capac- generalized Nakagami-m variates and performance of
ity sense increases with SNR and fading/shadowing pa- cascaded fading channels,” IEEE Global Telecommun.
rameters increasing. Conference (GLOBECOM ’09); p. 1, 2009.
[17] M. C. Jeruchim, P. Balaban and K. S. Shanmugan, Simulation
Acknowledgments: This work was supported by the Min- of Communication Systems. New York: Kluwer, 2002.

istry of Science of the Republic of Serbia under the grants


TR-32028 and III-044006.

Brought to you by | Umea University Library


Authenticated | 10.248.254.158
Download Date | 9/1/14 2:57 AM

You might also like