LOG-LIKELIHOOD-RATIO BASED SELECTIVE DECODE AND

FORWARD COOPERATIVE COMMUNICATION

Dr. R. Chembil Palat
1
, Dr. A. Annamalai
2
, Dr. Jeffrey H. Reed
1
Wireless @ Virginia Tech – Virginia Tech, Blacksburg, VA 24060, USA
2
Electrical & Computer Engineering, Prairie View A&M University, Prairie View, TX 77446 USA

Abstract- This paper presents exact ABER
performance analysis for selective decode and
forward (SDF) cooperative diversity system
with BPSK modulation under Rayleigh fading
where the relay has a MAP based receiver
and the retransmission is based on log-
likelihood-ratio (LLR) threshold. We also
derive the optimum LLR threshold that
minimizes ABER performance. It is shown
that the LLR relay based SDF cooperative
diversity system performs better than a SNR
threshold based SDF system with lower
implementation complexity than λ -MRC and
C-MRC schemes.

I. INTRODUCTION
Cooperative communication that exploits spatial
diversity from relay node transmissions has been
shown to perform better than systems with
collocated antennas [1], [2]. Several adaptive
versions of decode and forward cooperative
diversity system such as selective decode and
forward (SDF) [2], λ -MRC [3] and C-MRC [4]
have been proposed and studied in the past.
Average bit error rate (ABER) performance of
SDF schemes were studied in [5] and [6].
However the issue of wrongful combining at the
destination due to transmission of incorrect
symbols from the relay has not been addressed
correctly. It is either assumed that the relay uses
CRC to check for errors or the bit error
probability (BEP) due to wrongful combing is
approximated to 0.5. Hence it is only feasible to
determine Block or Frame error rates through
simulation in the first case or achieve upper
bounds for ABER performance in the second
case.
Recently, [7] suggested using log-likelihood-
ratio threshold based adaptive forwarding for
spatial-multiplexing based cooperative
communication. In this paper we provide an
efficient analysis for ABER performance of a
LLR based SDF cooperative diversity system
under Rayleigh faded links and compare it with
other adaptive decode and forward schemes. We
also address the issue of correctly determining
the effect of wrongful combining on ABER
performance.
II. SYSTEM MODEL
We consider a classical relay based cooperative
communication system consisting of a source, a
relay and a destination node as shown in Fig. 1.
The protocol used for the DF scheme is similar
to the adaptive decode and forward scheme
described in [6]. Communication takes place in
two phases due to the half-duplex constraint at
the relay. In phase-1 the source broadcasts
information to both the relay and destination.
The relay has a maximum a posteriori (MAP)
receiver where the relay decides to retransmit
only when the LLR is above a certain preset
threshold or else it remains silent in phase-2. At
the destination in phase-2, the receiver combines
the signal stored in phase-1 and the received
signal from the relay using MRC diversity
combining.
BPSK modulation is considered and the links
between source-to-destination, source-to-relay
and relay-to-destination are assumed to
experience frequency non-selective Rayleigh
fading. The normalized mean path SNR is given
by
{ }
2 2
ij ij
E σ α = where ij represents the link
between
th
i and
th
j nodes and
ij
α is a random
variable representing the fading amplitude
process. We let
2
SD
ij
ij
d
d
µ
σ
| |
=
|
\ .
where
SD
d the
distance from source to the destination and
ij
d is
the distance between
th
i and
th
j nodes while the
pathloss exponent is denoted by µ .
The signal received at the relay and the
destination in phase one and two can be written
as

phase 1
ˆ
phase 2
SR S SR S SR
SD S SD S SD
RD S RD R RD
r b E n
r b E n
r b E n
α
α
α
¹
= +
¦
`
= +
¦
)
= +
(1)

978-1-4244-1645-5/08/$25.00 ©2008 IEEE 615
S
D
R
θ
d
SD
=1
d
RD
=0.5
d
RS
=0.5
S
D
R
θ
d
SD
=1
d
RD
=0.5
d
RS
=0.5

Fig. 1. Classical relay based cooperative system
with different node geometries with relay
subtending angles
4
π
θ = with respect to the
source-to-destination link at the destination

where
ij
r denotes signal received at node j from
node i ,
S
b is the transmitted symbol,
ˆ
S
b is the
symbol transmitted by the relay after decoding
S
b ,
ij
n is the zero-mean unit-variance complex
additive noise sample at node j and
i
E is the
energy with which a symbol was transmitted
form node i . To provide a fair comparison while
evaluating the performance of a relay based
system and a direct link, the total energy
consumed by the cooperative diversity system is
assumed to be equal to that of a direct
transmission system. This is done as follows. If
T
E is the total energy per symbol available for
direct transmission, then a fraction δ of the total
energy is allocated to the source and the rest
allocated to the relay transmit power. This can be
written as

( ) and 1
T S R
S T R T
E E E
E E E E δ δ
= +
= = −
(2)
The average received SNR in each link is given
by
2
2
2
T SD
SD
SD
E δ σ
γ
η
= for source-to-destination
link,
2
2
2
T SR
SR
SR
E δ σ
γ
η
= for source-to-relay link,
( )
2
2
1
2
T RD
RD
RD
E δ σ
γ
η

= where ( ) ( )
2
0
T
T SD
E s t dt δ =

,
( ) ( ) ( )
2
0
1
T
T RD
E s t dt δ − =

and
2 0
2
ij
N
η = is the
two-sided power spectral density of AWGN.
Next we present the ABER performance analysis
for the above scheme.
III. ABER PERFORMANCE ANALYSIS
The error probability of the decode and forward
scheme is dictated by three mutually exclusive
events which is given by
( )
( ) ( )
1
DIV X SD SR SR
e e e e
e e
P P F P P P P
∗ ∗

Λ
Λ>Λ Λ>Λ
| |
= ⋅ Λ + − ⋅ + ⋅
|
\ .
(3)
The first term on the RHS of equation (3)
represents the event where there is no
retransmission from the relay as the LLR is
below the preset threshold denoted
by

Λ represented by the CDF
( ) ( )
Pr F
∗ ∗
Λ
Λ = Λ < Λ .Thus the detection at the
destination is solely based on probability of
symbol error in source-to-destination link
represented by
SD
e
P . The second and third terms
represent the event where the relay decodes and
decides to transmit in phase-2. The second term
represents the case where the relay correctly
decoded the symbol transmitted from the source
in phase one with probability 1
SR
e
P

Λ>Λ
| |

|
\ .
and in
phase-2 after coherent diversity combining of the
two received signals at the destination, the
transmitted symbol is decoded in error with
probability
( ) DIV
e
P . The third term represents the
case where the relay in phase one decodes the
symbol transmitted from the source in
error
SR
e
P

Λ>Λ
. This can happen due to improper
choice of threshold level. This leads to wrongful
combining at the destination in phase two where
( ) X
e
P denotes the probability of error due to
wrongful combining.
The probability of error
SD
e
P in (3) can now be
computed as
( ) ( )
( )
2
2
0
1 1
,
sin
SD SD
SD
e e
P P f d d
π
γ γ
γ
γ γ γ φ γ θ
π θ ∗


| |
= = |
|
\ .
∫ ∫

(4)
where ( )
SD
γ
φ ⋅ is the MGF of the instantaneous
received SNR
SD
γ which is readily available in
closed form for Rayleigh fading [9]. Following
the treatment in [8],
( ) DIV
e
P can be computed as
( )
( ) ( )
1 2
2 2
0
1 1 1
sin sin
SD RD
DIV
e
P d
γ γ
φ φ θ
π θ θ
| | | |
=
| |
| |
\ . \ .


616
(5)
The decision statistic as a result of wrongful
combining can be written as
( )
( )
{ }
2 2 2 2
ˆ
1
1
D T SD SD T RD RD
T SD SD SD T RD RD RD
Z E E
E n E n
δ σ α δ σ α
δ σ α δ σ α
∗ ∗
= − −
+ ℜ + −
(6)
The bit error probability for the decision statistic
in (6) can no be evaluated as
( )
1 2
X
e
P P P = + where
( ) ( )
( )
( ) ( )
2 2 2 2
1 2 2 2 2
0 0 1
2 1
1
1

RD
SD RD
T
SD SD RD RD
SD SD RD RD
SD RD SD RD
E
P Q
N
f f d d
δ
α
δ
α α
δσ α δ σ α
δσ α δ σ α
α α α α
∞ ∞

| |
= − − |
|
+ −
\ .
×
∫ ∫
(7)
and
( ) ( )
( )
( ) ( )
1
2 2 2 2
2 2 2 2 2
0 0 0
2 1
1 1
1

RD
SD RD
T
RD RD SD SD
SD SD RD RD
SD RD SD RD
E
P Q
N
f f d d
δ
α
δ
α α
δ σ α δσ α
δσ α δ σ α
α α α α


( | |
= − − − ( |
|
+ − (
\ . ¸ ¸
×
∫ ∫

(8)
Now we turn our attention to the MAP receiver
at the relay. For BPSK modulation the decision
statistic is given by

{ } { }
2
ˆ
SR SR SR SR S S SR SR
Z r E b n α α α
∗ ∗
= ℜ = + ℜ (9)
The LLR Λ for this receiver is given by
ˆ
4
SR S
Z E Λ = [9]. Now if we let
ˆ
S SR
h E Z = and thus 4h Λ = then for
Rayleigh fading channel the PDF of h is given
by [9]
( )
2
1 1
exp 2 1 1 exp 2 1 1
h
h h
f h
β β
β β
( ( | | | |
− + + + − + − ( ( | |
| |
( ( \ . \ . ¸ ¸ ¸ ¸
=
+

(10)
where
{ }
2
SR s
E E β α = . Now
SR
e
P

Λ>Λ
can be
computed as

( )
( )
4
1
1 exp 4
SR
h
e
P f h dh
h



Λ>Λ
Λ
=
+

(11)
The CDF
( ) ( )
Pr F
∗ ∗
Λ
Λ = Λ < Λ can also be
evaluated as [9]
( ) ( )
( ) ( )
4
2 2
0
2 1 2 1
1 exp 1 1 1 exp 1 1
4 4
2 1 2 1
h
F f h dh
β β
β β β β β β

∗ ∗
Λ

Λ
( ( | | | | − Λ − Λ
− + + − + − ( ( | |
( ( \ . \ . ¸ ¸ ¸ ¸
Λ = = +
+ + + + − +


(12)
Using (4)-(12) the bit error probability in (3) can
now be evaluated. Next the derivation of
optimum LLR threshold is discussed.
IV. OPTIMAL LLR THRESHOLD
From (3) it can be observed that an optimum

Λ can be obtained by minimizing the ABER in
(3). Noting that,
( ) ( )
b
a
d
f x f a
da
= −

and ( ) ( )
b
a
d
f x f b
db
=

; we
can derive optimum

Λ by taking the derivative
of (3) and equating it to zero. The derivative of
(3) is given by

( )
( )
( )
( )
( ) ( )
( )
( )
1 1 1
4 4 4 4
1 exp
1 1

4 4 1 exp
1

4 4
1 exp
DIV SD e
e h e h
X
e h
DIV X
e e
SD
h e
dP
P f P f
d
P f
P P
f P
∗ ∗






(
| | | | Λ Λ
(
= − −
| |
( Λ
+ Λ \ . \ .
¸ ¸
(
| | Λ
(
+ −
|
(
+ Λ \ .
¸ ¸
(

| | Λ
(
= +
|
(
+ Λ \ .
¸ ¸
(13)
Equating (13) to zero and observing
that 0
4
h
f

| | Λ

|
\ .
, the optimum LLR threshold is
obtained when
( )
( ) ( )
1
1 exp
SD
e
X DIV
e e opt
P
P P

=
− + Λ
and
the optimum LLR is given by
( ) ( )
ln 1
X DIV
e e
opt SD
e
P P
P

(

Λ = −
(
(
¸ ¸
. In the next section
we present the results.
V. NUMERICAL RESULTS
Fig. 2 shows the plots for the performance
comparison of the DF scheme that results in
optimum ABER performance using ML
detection at the destination [3], C-MRC [4], SDF
scheme based on SNR threshold at the relay and
the SDF scheme with MAP-receiver based relay
that uses LLR threshold. The system model is
based on the discussion in section II. The
parameters for the system model with reference
to Fig. 1 are 3 µ = , 1
SD
d = , 0.5
RD
d = and
1.25 cos
4
SR
d
π | | | |
= −
| |
\ . \ .
. The figure includes
plots ABER for different values of LLR
thresholds and SNR threshold levels. It can be
observed that MAP based scheme outperforms
the simple SNR threshold based SDF scheme as
617
it is based on a better decision metric at the relay.
The optimum LLR computed using the
derivation in section IV shows an ABER
performance that is approximately 1 dB worse
than optimum ML based DF scheme [3] and C-
MRC [4] which have also been simulated for the
same set up. The SDF-MAP scheme only
requires the knowledge of average SNRs to
compute the BEP of the source-to-destination
link and the relay-to-destination link to set the
LLR threshold whereas λ -MRC [3] and C-MRC
schemes require constant update based on
instantaneous values bit error rate and SNRs
over each link. Hence the implementation
complexity is much lower in the case of MAP
based SDF. It is also observed that higher

Λ performs better at higher
0 T
E N compared to
lower values. This is due to lesser wrongful
combining at higher LLR thresholds. More on
the effect of location of the relay and threshold
can be found in [10].
VI. CONCLUSIONS
In this correspondence we derived an efficient
method to compute the ABER of a SDF based
cooperative diversity system under Rayleigh
fading that uses a MAP receiver at the relay that
forwards symbols based on LLR threshold.
ABER performance comparison showed that
SDF-MAP system performs about 1 dB worse
than the optimum DF system at considerably low
implementation complexity.

0 5 10 15 20 25 30
10
-5
10
-4
10
-3
10
-2
10
-1
E
T
/N
0
dB
A
B
E
R
Direct link
ML detection (optimum)
C-MRC
SDF-SNR, γ
*
= -10dB
SDF-SNR, γ
*
= 10dB
SDF-SNR, γ
*
opt
SDF-MAP, Λ
*
= -10dB
SDF-MAP, Λ
*
= 10dB
SDF-MAP, Λ* = Λ
*
opt
δ = 0.5
Rx Cluster θ = π/4

Fig. 2 ABER performance comparison of SDF
cooperative communication system with BPSK
modulation under Rayleigh fading with relay in
receive-cluster (
4
π
θ = ).

REFERENCES

[1] A. Sendonaris, E. Erkip, and B. Aazhang,
“User cooperative diversity -part I: system
description,” IEEE Trans. Commun., vol. 51, pp.
1927–1938, Nov. 2003.
[2] J. N. Laneman, D. N. C. Tse, and G. W.
Wornell, “Cooperative diversity in wireless
networks: efficient protocols and outage
behavior,” IEEE Trans. Inform. Theory, vol. 50,
pp. 3062–3080, Dec. 2004.
[3] A. Sendonaris, E. Erkip, and B. Aazhang,
“User cooperative diversity -part II:
Implementation aspects and performance
analysis,” IEEE Trans. Commun., vol. 51, pp.
1939–1948, Nov. 2003.
[4] Tairan Wang, A. Cano, G.B. Giannakis,
“Efficient Demodulation in Cooperative
Schemes Using Decode-and-Forward Relays,”
IEEE Thirty-Ninth Asilomar Conf. on Signals,
Systems and Computers, pp. 1051 – 1055, Oct.
2005.
[5] P. Herhold, E. Zimmermann, G. Fettweis, “A
simple cooperative extension to wireless
relaying,” International Zurich Seminar on
Communications, pp. 36 – 39, 2004.
[6] Weifeng Su, A.K. Sadek, K.J.R. Liu, “SER
performance analysis and optimum power
allocation for decode-and-forward cooperation
protocol in wireless networks,” IEEE WCNC, vol.
2, pp. 984 – 989, 13-17 March 2005.
[7] Sang Wu Kim, “Cooperative spatial
multiplexing in mobile ad hoc networks,” IEEE
International Conference on Mobile Adhoc and
Sensor Systems, 7-10 Nov. 2005.
[8] Vanganuru K., Annamalai A., “Combined
transmit and receive antenna diversity for
WCDMA in multipath fading channels,” IEEE
Commun. Letters, vol. 7, pp. 352 – 354, Aug.
2003.
[9] Sang Wu Kim, Eun Yong Kim, “Optimum
receive antenna selection minimizing error
probability,” IEEE WCNC, vol. 1, pp. 441 – 447,
13-17 March 2003.
[10] R. Chembil Palat, A. Annamalai, J.H. Reed,
“Probability of error under path loss arbitrary
fading and power allocation for selective decode
and forward cooperative communication,”
Unpublished





618

The probability of error PeSD in (3) can now be computed as PeSD = ∞ where rij denotes signal received at node j from ˆ node i . The second and third terms represent the event where the relay decodes and decides to transmit in phase-2. where δ ET = ∫ ( sSD ( t ) ) dt . Pe( DIV ) can be computed as Pe( DIV ) γ RD = 2 (1 − δ ) ET σ RD N0 is the 2 0 two-sided power spectral density of AWGN. b is the S S symbol transmitted by the relay after decoding bS . the transmitted symbol is decoded in error with probability Pe( DIV ) . 1. the total energy consumed by the cooperative diversity system is assumed to be equal to that of a direct transmission system. then a fraction δ of the total energy is allocated to the source and the rest allocated to the relay transmit power.R Next we present the ABER performance analysis for the above scheme. The second term represents the case where the relay correctly decoded the symbol transmitted from the source in phase one with probability  1 − PeSR ∗  and in  Λ>Λ    phase-2 after coherent diversity combining of the two received signals at the destination. (1 − δ ) ET = ∫ ( sRD ( t ) ) 2 dt and ηij = = 1 12 π ∫ φγ 0 SD  1   1  dθ  2 φγ RD  2  sin (θ )   sin (θ )       616 . γ ∗ dθ  2  sin (θ )    δ ET σ 2 2ηSR 2 2η RD T 2 SR for source-to-relay link. This is done as follows. III.5 S θ dRD=0.5 D The error probability of the decode and forward scheme is dictated by three mutually exclusive events which is given by DIV X Pe = PeSD ⋅ FΛ ( Λ ∗ ) + 1 − PeSR>Λ∗  ⋅ Pe( ) + PeSR> Λ∗ ⋅ Pe( )   Λ Λ   (3) dSD=1 Fig. Classical relay based cooperative system with different node geometries with relay subtending angles θ = with respect to the 4 source-to-destination link at the destination π The first term on the RHS of equation (3) represents the event where there is no retransmission from the relay as the LLR is below the preset threshold denoted by Λ∗ represented by the CDF ∗ ∗ FΛ ( Λ ) = Pr ( Λ < Λ ) . Following the treatment in [8]. This leads to wrongful combining at the destination in phase two where X Pe( ) denotes the probability of error due to wrongful combining. This can be written as ET = ES + ER ES = δ ET and ER = (1 − δ ) ET (2) The average received SNR in each link is given 2 δ ET σ SD for source-to-destination by γ SD = 2 2η SD link. This can happen due to improper Λ> choice of threshold level. b is the transmitted symbol. To provide a fair comparison while evaluating the performance of a relay based system and a direct link. γ SR = γ∗ 1 ∫ P (γ ) fγ (γ )d γ = π ∫ φγ e SD π 2 SD 0  1  . ABER PERFORMANCE ANALYSIS dRS=0.Thus the detection at the destination is solely based on probability of symbol error in source-to-destination link represented by PeSD . If ET is the total energy per symbol available for direct transmission. nij is the zero-mean unit-variance complex additive noise sample at node j and Ei is the energy with which a symbol was transmitted form node i . 2 0 2 T (4) where φγ SD ( ⋅ ) is the MGF of the instantaneous received SNR γ SD which is readily available in closed form for Rayleigh fading [9]. The third term represents the case where the relay in phase one decodes the symbol transmitted from the source in error PeSR Λ∗ .

1 are µ = 3 . V. OPTIMAL LLR THRESHOLD +ℜ { ∗ δ ET σ SDα SD nSD + ∗ (1 − δ ) ET σ RDα RD nRD } (6) The bit error probability for the decision statistic in (6) can no be evaluated as X Pe( ) = P + P2 where 1  2 ET 1 2 2 2 2 P = ∫ ∫ Q  ( δσ SDα SD − (1 − δ ) σ RDα RD ) 1 2 2 2 2  N 0 δσ SDα SD + (1 − δ ) σ RDα RD 0 1−δ  α RD ∞ ∞ From (3) it can be observed that an optimum Λ∗ can be obtained by minimizing the ABER in (3). Now if we let ˆ h = ES Z SR and thus Λ = 4h then for Rayleigh fading channel the PDF of h is given by [9]       1 1 exp  −2h  1 + + 1  + exp  −2h  1 + − 1      β β           fh (h) = β2 + β optimum LLR is given by  Pe( X ) − Pe( DIV )  Λ ∗ = ln  − 1 . b b d d f ( x ) = − f ( a ) and ∫ f ( x ) = f ( b ) . d SD = 1 . d RD = 0. Now PeSR ∗ can be Λ>Λ computed as PeSR ∗ = Λ>Λ ∞ Λ ∫ ∗ 1 f h ( h ) dh 1 + exp ( 4h ) 4 (11) The CDF FΛ ( Λ∗ ) = Pr ( Λ < Λ ∗ ) can also be evaluated as [9] FΛ ( Λ ∗ Fig. the optimum LLR threshold is  4  obtained when the 1 1 + exp Λ ∗ opt The LLR Λ for this receiver is given by ˆ Λ = 4Z SR ES [9]. The figure includes  4   plots ABER for different values of LLR thresholds and SNR threshold levels.5 and   π  d SR = 1.25 − cos    . NUMERICAL RESULTS ( ) = Pe( X) PeSD − Pe( DIV ) and (10) 2 where β = E {α SR } Es . Noting that. Next the derivation of optimum LLR threshold is discussed. The derivative of (3) is given by dPe 1  Λ∗ = PeSD f h  ∗ 4  4 dΛ   1 ( DIV )  1 −  − Pe  4 1 + exp Λ∗    Λ∗  fh    4   × fαSD (α SD ) fα RD (α RD ) d α SD d α RD (7) and P2 = ∫ 0 ∞ ( ) α RD 1−δ ∫ 0 δ   2ET 1 2 2 2 2 1 − Q  ( (1 − δ ) σ RDα RD − δσ SDα SD ) 2 2 2 2  N0 δσ SDα SD + (1 − δ )σ RDα RD    × fα SD (α SD ) fα RD (α RD ) dα SD dα RD     + Pe (X )  1 1 −  4 1 + exp Λ ∗  ∗ (8) Now we turn our attention to the MAP receiver at the relay. In the next section opt PeSD     we present the results. we da ∫ db a a     δ can derive optimum Λ∗ by taking the derivative of (3) and equating it to zero. SDF scheme based on SNR threshold at the relay and the SDF scheme with MAP-receiver based relay that uses LLR threshold.(5) The decision statistic as a result of wrongful combining can be written as 2 2 2 2 ˆ Z D = δ ET σ SDα SD − (1 − δ ) ET σ RDα RD Using (4)-(12) the bit error probability in (3) can now be evaluated. It can be observed that MAP based scheme outperforms the simple SNR threshold based SDF scheme as )= ∫ 0 Λ∗ 4  −2 Λ∗   −2 Λ∗   1  1 1 − exp  1 + 1 +  1 − exp   1 + − 1 β  β  4   4    +   f h ( h ) dh = 2 β + 1+ β 2 + β 2 β +1 − β 2 + β ( ) ( ) (12) 617 . C-MRC [4]. IV. The parameters for the system model with reference to Fig. 2 shows the plots for the performance comparison of the DF scheme that results in optimum ABER performance using ML detection at the destination [3]. The system model is based on the discussion in section II. For BPSK modulation the decision statistic is given by ∗ 2 ∗ ˆ Z SR = ℜ {α SR rSR } = α SR ES bS + ℜ {α SR nSR } (9) DIV X  Pe( ) − Pe( ) 1  Λ   SD = fh    Pe + ∗ 4  4  1 + exp Λ  ( ) (  Λ∗  fh    4    ( )  ) (13) Equating (13) to zero and observing  Λ∗  that f h   ≥ 0 .

H. 51. on Signals. 13-17 March 2005. J. C. 2003. Λ* = -10dB 10 -5 SDF-MAP.” IEEE International Conference on Mobile Adhoc and Sensor Systems. The SDF-MAP scheme only requires the knowledge of average SNRs to compute the BEP of the source-to-destination link and the relay-to-destination link to set the LLR threshold whereas λ -MRC [3] and C-MRC schemes require constant update based on instantaneous values bit error rate and SNRs over each link. pp. N. vol. Λ* = 10dB SDF-MAP. vol.” IEEE Trans. 1. Laneman. Tse. “Optimum receive antenna selection minimizing error probability. “Cooperative diversity in wireless networks: efficient protocols and outage behavior. Nov. 2 ABER performance comparison of SDF cooperative communication system with BPSK modulation under Rayleigh fading with relay in π receive-cluster ( θ = ). vol. “Probability of error under path loss arbitrary fading and power allocation for selective decode and forward cooperative communication. γ* = -10dB SDF-SNR. Aug. A. 2004. Eun Yong Kim. [3] A. N. Theory. G.K. pp. 3062–3080. and B. [9] Sang Wu Kim. 50. Herhold. 51..” IEEE WCNC. Wornell. and B. Liu. [8] Vanganuru K. 441 – 447. “A simple cooperative extension to wireless relaying.J. pp. Nov. K. ABER performance comparison showed that SDF-MAP system performs about 1 dB worse than the optimum DF system at considerably low implementation complexity. Zimmermann. [7] Sang Wu Kim. “Efficient Demodulation in Cooperative Schemes Using Decode-and-Forward Relays. A. 2003. Sendonaris. vol. Commun. 36 – 39. “SER performance analysis and optimum power allocation for decode-and-forward cooperation protocol in wireless networks.5 Rx Cluster θ = π/4 10 -2 10 -3 Direct link ML detection (optimum) C-MRC SDF-SNR. This is due to lesser wrongful combining at higher LLR thresholds. “User cooperative diversity -part II: Implementation aspects and performance analysis. A. “Combined transmit and receive antenna diversity for WCDMA in multipath fading channels. VI. E. 7-10 Nov. Sendonaris..” IEEE WCNC. More on the effect of location of the relay and threshold can be found in [10]. E.” IEEE Trans. Λ* = Λ* opt 0 5 10 15 E /N dB T 0 20 25 30 [1] A. 4 618 . and G. Cano.R. [2] J. 1927–1938. 7. pp. pp. 2005.B. 2.” IEEE Commun. [5] P. 13-17 March 2003. vol. CONCLUSIONS REFERENCES In this correspondence we derived an efficient method to compute the ABER of a SDF based cooperative diversity system under Rayleigh fading that uses a MAP receiver at the relay that forwards symbols based on LLR threshold.” International Zurich Seminar on Communications. [4] Tairan Wang.” IEEE Trans.” Unpublished ABER Fig. “User cooperative diversity -part I: system description. 2004. 352 – 354. W. Reed. γ* opt 10 -4 SDF-MAP.” IEEE Thirty-Ninth Asilomar Conf. 984 – 989. Annamalai A. pp. Hence the implementation complexity is much lower in the case of MAP based SDF.it is based on a better decision metric at the relay. Chembil Palat. γ* = 10dB SDF-SNR. Oct. G. Giannakis. [6] Weifeng Su. It is also observed that higher Λ∗ performs better at higher ET N 0 compared to lower values. 10 -1 δ = 0. pp. 1051 – 1055.. Inform. E. “Cooperative spatial multiplexing in mobile ad hoc networks. 1939–1948. D. 2003. [10] R. 2005. Erkip. Dec. Aazhang. Aazhang. Fettweis. The optimum LLR computed using the derivation in section IV shows an ABER performance that is approximately 1 dB worse than optimum ML based DF scheme [3] and CMRC [4] which have also been simulated for the same set up. Annamalai. vol.. Systems and Computers. Erkip. Letters. Commun. pp. Sadek.

Sign up to vote on this title
UsefulNot useful