Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Manuscript received February 15, 2016; revised July 18, 2016. Corresponding author email: [email protected].
doi:10.12720/jcm.11.7.667-676
667
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
Performance Analysis of Coded Cooperation Based on
Distributed Turbo Codes with Multiple Relays
Jing An1 and Chang Li
2
1 Department of Information Engineering, Shijiazhuang Vocational Technology Institute, Shijiazhuang050081, China
2 Network Optimization Center, Network Department,
China Mobile Group Hebei Co. Ltd.,
Shijiazhuang 050035, China
Email: {anj.sjz, lichang.sjz}@hotmail.com
Abstract—Diversity is an effective technique to enhance link
quality and enable multiple users to pool their resources to form
a virtual antenna array that obtains spatial diversity. In this
paper, we propose a generalized Distributed Turbo Codes
(DTC)-based coded cooperation protocol for two-hop relay
networks with an arbitrary number of relays. This scheme aims
at achieving improved diversity over the classical coded
cooperation method in Rayleigh fading channels. We derive a
finite range single integral solution for the outage probability,
which characterizes the coded performance with multiple relays
at various rates. We also develop a closed-form expression for
the Pairwise Error Probability (PEP) and a tight upper bound for
the Bit Error Rate (BER) using DTC. The results demonstrate
the merits of DTC-based coded cooperation with multiple relays,
under various relay and uplink channel conditions, and show
that the proposed scheme is able to achieve the full diversity
order. Moreover, the analytical upper bounds are validated with
simulation results. Index Terms—Bit error rate, channel coding, coded
cooperation, distributed turbo code, fading channels, multiple
relays, outage probability, pairwise error probability
I. INTRODUCTION
In order to provide reliable transmission at high data
rates and offer a variety of multimedia services in future
wireless communication systems, diversity has been
considered as an effective technique in combating
detrimental effects of channel fading caused by multi-
path propagation and Doppler spread. However, in
cellular network, due to the limited size, cost and
hardware limitations, it may not be possible equip a
mobile terminal with multiple transmit antennas.
Recently, the cooperative communications technique is
emerging to enable a new form of spatial diversity in
wireless systems. Single antenna mobiles in a multi-user
environment share their antennas and form a virtual
multiple antenna transmitter that is able to achieve
transmit diversity. The recent surge of research interests
in cooperative communications is subsequent to the
pioneering work of Sendonaris et al. [1], [2]. Shortly
afterwards, Laneman et al. [3]–[5] proposed several
cooperative protocols and analyzed their performance in
both ergodic as well as quasistatic channels. Among the
cooperative protocols they proposed and analyzed are
Decode and Forward (DF), Amplify and forward (AF),
and adaptive methods that switch between the two. In
parallel to the work of Laneman, an alternative
framework, dubbed coded cooperation was proposed, in
which cooperative signaling is integrated with channel
coding [6]–[9]. The basic idea is that each user, instead of
repeating the received bits (either via amplification or
decoding), tries to transmit incremental redundancy for
its partner. In the original coded cooperation strategy [7],
users transmit only their partners’ data in the cooperation
frame whenever possible, which can achieve full
diversity and lead to significant performance
improvements in slow fading channels. However, this
strategy is unable to achieve diversity gains, and suffers
from cooperation imbalances between users operating in
fading environments. Then, space time coded cooperation
is proposed to achieve improved diversity over the
original coded cooperation protocol and reduce the
impact of cooperation imbalances [10]. However,
synchronization for space-time cooperation especially in
Time Division Multiple Access (TDMA) systems
becomes a serious problem because the symbol-level
synchronization between users in the uplink cannot be
guaranteed.
In [11], Elfituri et al. proposed a convolutional coding
based distributed coded cooperation scheme for multiple
relay channels. The authors derived simplified closed-
form expressions for the outage probability in the case of
error free and erroneous relaying. However, their
analytical results (i.e., (28) and (34) in [11]) are flawed
and do not match with the presented simulation curves
(e.g., Fig. 7 and Fig. 9 in [11]). This is because that the L-
fold multiple integral in (27) of [11] is overly simplified
into L single integrals.
To further explore the cooperative spatial diversity and
cooperative coding gains, various distributed coding
schemes have been proposed, such as convolutional
codes [8], [12], distributed space time coding [13], [14],
distributed low density parity check codes (D-LDPC)
[15], [16] as well as the distributed turbo coding (DTC)
scheme [17]-[19].
As the idea of relay networks attracted attention,
researchers began to investigate their information
theoretic aspects. In particular, determining achievable
rate regions was investigated under a number of different
668
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
assumptions [19]-[24]. [25] proposed an extended
distributed turbo coding scheme for half-duplex relay by
transferring information via the timing of the relay-
receive and relay transmit phases to the destination.
However, most existing DTC schemes are only designed
for a relay networks with a single relay node. In [26] a
generalized hybrid relaying scheme was proposed for
multiple relays network with any number of relays.
However, the hybrid relaying protocol including the AF
and DF strategy is too complex [27]-[38]. Therefore, it is
very important to develop a DTC scheme for the general
relay network with any number of relays.
In literature [39], they develop a closed-form
expression for the Pairwise Error Probability (PEP) and a
tight upper bound for the Bit Error Rate (BER) using
DTC.
In this work, we consider the multiple-relay Rayleigh
fading channels, and study the outage probability and
BER of the coded cooperation protocol based on DTC
with multiple relays. We aim to derive a finite range
single integral solution for the outage probability, and a
closed-form expression for the Pairwise Error Probability
(PEP) of the proposed DTC-based coded cooperative
protocol. In comparison with the original coded
cooperation protocol, the proposed strategy based on
DTC with multiple relays is capable of providing the
added benefits of improved diversity gains, and reducing
the impact of cooperation imbalances.
This paper is organized as follows. Section 2 describes
the system model for the proposed coded cooperation
scheme. Section 3 presents a mathematical probability
model and derives a finite range single integral solution
for the outage probability. Section 4 evaluates the bit
error probability of DTC-based coded cooperation with
an arbitrary number of relays when Binary Phase-Shift
Keying (BPSK) modulation is used. Section 5 presents
simulation results and discussions. Conclusion is drawn
in Section 6.
S D
R
.
.
.
R
R
M
S D
R
.
.
.
R
R
M
Frame 1 Frame 2 Fig. 1. System model of coded cooperation with multiple relays.
II. SYSTEM MODEL
In our model, we consider the scenario of one source
with the help of M relays. We design a distributed
coding scheme at both the source and relays, where the
encoding process is divided into two frame transmissions.
In what follows, we denote by s , ir and d the source,
thi relay, and destination nodes, respectively. Consider
the multiple-relay channel shown in Fig. 1, where the
data is sent from s to d with the assistance of M relay
nodes.
As illustrated in Fig. 1, both the source and relay nodes
employ convolutional coding to protect information bits.
There are K bits in each source block, and the length of
the codeword is N . In the distributed coding scheme
under consideration here, the -bitN codeword is
partitioned into two successive frames (i.e., frame 1 and
frame 2) of 1N and 2N bits ( 1 2N N N ) with rate 1R
and 2R [6], [7], using a Rate-Compatible Punctured
Convolutional (RCPC) code puncturing matrix [8]. We
can quantify the level of cooperation with the parameter
1 1N N R R (1)
which indicates the portion of the total channel codeword
allocated for the first frame.
In frame 1, the source broadcasts the first 1N bits to
both the relays and destination. If a relay correctly
decodes the message received from the source, it re-
encodes the message into the original -bitN codeword.
Otherwise, the relay becomes silent. In the second frame,
the source and the relays whose Cyclic Redundancy
Checks (CRCs) are validated transmit the second 2N bits
to the destination, note the transmit energy is split equally
among all the active relays. The received copies of the
second part are combined using Maximal Ratio
Combining (MRC), and the information bits in the first
and second frames are decoded by a Viterbi decoder at
the receiver.
A. Transmission Protocols
In the following analysis, we analyze the performance
of the proposed cooperative coding scheme based upon
the DTC over slow fading channels, where fading
coefficients remain constant during each frame
transmission interval.
For the first frame, the signals that are received at the
relay and destination nodes are given by
, , , ,s d s d s d s dy h E x n (2)
, , , ,i i i is r s r s r s ry h E x n (3)
where x denotes the modulated source signal, and
1,2,...,i M . ,s dh and , is rh represent the fading channel
coefficients from s to d and s to ir , respectively. ,s dE
and , is rE are the transmitted signal power for the
corresponding links, while ,s dn and ,s dn represent complex
Additive White Gaussian Noise (AWGN) over the
corresponding channels.
Let L be the number of relays that are used for
cooperation in the second phase (i.e., these relays
correctly decode the received message). Accordingly, the
received signals at the destination node are given by
669
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
,
, , ,ˆi
i i i
r d
r d r d r d
Ey h x n
L (4)
, , , ,s d s d s d s dy h E x n (5)
where x̂ denotes the relay transmitted signal, and 1 L is
the ratio for maintaining the same average power
compared to the single relay scenario. That is, the
transmit energy is split equally among all the active
relays in the second phase, each of the relays transmits
with the same power. Note every two frames have a
different L .
SourceCRC
RSC2πViterbi
Decoder
Systematic bits
Parity 1
R1
Sys+Parity 1
RSC1
RSC2π Parity 2
RSC2πViterbi
Decoder
Parity 2RM
Sys+Parity 1
M
R
C
Parity 2
Turbo
Iterative
DecoderParity 2
Sys+Parity 1
...
Sys bits
Sys bits
Frame 2
Frame 1
C
T
N
Fig. 2. Distributed turbo encoding in a coded cooperation scheme.
B. Distributed Turbo Codes with Multiple Relays
The implementation of coded cooperation using DTC
is illustrated in Fig. 2. The turbo codes employ two
constituent Recursive Systematic Convolutional (RSC)
codes with interleaving. The source and relay share the
same random interleaver, shown as . To reduce
implementation complexity, the relay employs
conventional Viterbi decoding. Note that the systematic
and the first parity bits are 1N , the second parity bits are
2N . It is possible to have a more flexible level of
cooperation, as well as better performance, by using
punctured turbo codes or Rate Compatible Punctured
Turbo codes (RCPT) [40].
Fig. 2 shows the sketches a block diagram for coded
cooperation scheme. In the first frame, the source
broadcasts the 1N bits (systematic and the first parity bits)
to both the relays and destination. If a relay correctly
decodes the message received from the source using the
Viterbi decoder, it interleaves the source bits and re-
encodes them into 2N bits (the second parity bits).
Otherwise, that relay keeps silent. In the second frame,
the source and the relays whose Cyclic Redundancy
Checks (CRCs) are validated transmit the 2N bits to the
destination. At the destination, the received copies of the
second parity bits are combined using Maximal Ratio
Combining (MRC), and the information bits in the first
and second frame are concatenated (CTN) and then
decoded by a turbo decoder. The low-complexity iterative
decoder offers near-optimum decoding performance for
turbo codes.
III. OUTAGE PROBABILITY ANALYSIS
First, we consider non-cooperative direct transmission
between the source and destination. With quasi-static
fading, the capacity conditioned on channel realization,
characterized by the instantaneous Signal-to-Noise Ratio
(SNR), can be expressed by the classic Shannon formula
2log 1C in b/s/Hz. The channel is in outage if
the conditional capacity falls below a selected threshold
rate R, and the corresponding outage event is
C R , or equivalently, 2 1R . The outage
probability is defined as
2 1
0Pr 2 1
R
R
out rP p d
(6)
where rp denotes the probability density function
(pdf) of random variable . For Rayleigh fading, has an
exponentially distributed with parameter 1/ , where
denotes the mean of SNR of the fading channel and
accounts for the combined effects of large-scale path loss
and shadowing. The outage probability for Rayleigh
fading can thus be evaluated as
2 1
0
1 2 1exp 1 exp
R R
outP d
(7)
A. Outage Probability
The PDF document should be sent as an open file, i.e.
without any data protection.
In this section, we consider the situation where some
of the relays may fail to correctly decode the message
that they received from the source. That is, their CRCs do
not pass. Let L be the number of cooperative relay
stations. Apparently, L ranges from 0 to M. Denote by
the set of indices of the cooperating relays
1 2, , , 1,2, ,Li i i M (8)
Note that the cardinality of is L ( 0 L M ).
As discussed in the previous section, in coded
cooperation user data are transmitted over two successive
frames. In the first frame, each user transmits a rate
1R R codeword. There are two possible cases for the
second frame transmission depending on the number of
relay stations that are able to successfully decode the first
frame. Case I (denoted by = 1): None of the relay
670
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
stations decodes the source frame successfully. In the
information theoretical sense, this scenario corresponds to
the following event (Note that every two frames may
have a different L ).
, , 2 ,log 1i i is r s r s rC R (9)
where subscripts s and ir denote the transmission from
the source to the thi relay, and 1,2, , .i M In the
second frame, the source transmits additional parity bits.
For a given source, the destination will receive the first
frame and second transmission frames both from the
source. The first frame occupies fraction, whereas the
second frame uses the remaining 1 fraction of the
total bits. These two transmission can thus be viewed as
time sharing between two independent channels, where
the first channel uses a fraction of the time. Thus, we can
write the outage events as
, , 2 ,
2 ,
1 log 1
1 log 1
s d s d s d
s d
C
R
(10)
In case 2 (denoted by 2 ), some of the relays
decode the source frame successfully
, , 2 ,log 1i i is r s r s rC R (11)
where i . On the other hand, for the relays that failed
to decode the source message, we have
, , 2 ,log 1i i is r s r s rC R (12)
where i . In the second frame, both the L (1 L M )
error-free relays and the source transmit additional parity
bits to the destination. The corresponding outage events
are given as
, , 2 ,
2 , ,
2 log 1
1 log 1i
s d s d s d
s d r d
i
C
R
(13)
Since the above two cases are disjoint, we can write
the overall outage probability as in (14). Note that the
first, and second terms correspond to cases 1 and 2,
respectively. It should be emphasized that, in all the two
cases, the destination is in outage. For the case of
Rayleigh fading, (14) can be shown to evolve into (15),
where LS is an integration region given by (16).
In contrast, the L-dimensional region of integration in
(27) of [11] is decomposed into L independent one-
dimensional regions of integration. Therefore, the L-
dimensional integral is overly simplified into L single
integrals. The derived outage probability expressions
given in (28) and (34) in [11] are thus flawed. This is
further evidenced by the fact that the analytical results do
not match the simulation results presented in Fig. 7 and
Fig. 9 in [11].
As can be seen from (15), the outage probability for
DTC-based coded cooperation with multiple relays is a
function of the mean channel signal-to-noise ratio (SNR)
values ( ,s d , , is r and ,ir d ) the allocated rate R , and the
cooperation level . While the mean channel SNR
values and R may often be constrained by underlying
communications channels, is a free parameter that can
be varied to optimize performance.
According to the derivations in Appendix, we can
simplify (15) into (17). Note that we can simplify the
multiple integrals function to the single integral function,
thanks to the independence of different ,ir d .
/
, ,
11
/ /
, , , , ,
1
Pr 2 1 Pr 2 1
Pr 2 1 Pr 2 1 Pr 1 1 2
i
i i i
MR R
out s r s d
i
MR R R
s r s r s d s d r d
K ii i
P
M
L
(14)
/
1 , ,
/ /,,
,
, , , , ,
1 2 1 21 exp 1 exp
1 2 1 2 1 11 exp exp exp exp
i
i
i i iL
R RM
out
i s r s d
R Rr ds d
s d
i is r s r s d s d r d rS
P
Md
L
,
1 ,i
i
M
r d
L i d
d
(15)
1
, , , , , , ,= , 0, 0, , 1 1 2i i i
R
L s d r d s d r d s d s d r d
i
S i
(16)
/
, ,
/ /,
0, , , , ,
1 2 1 21 exp 1 exp
! 1 2 1 2 11 exp exp exp 1 exp
! !
i
i i i
MR R
out
s r s d
M L LR R
s d
s r s r s d s d r d
P
M a
L M L
2 1
,
1
R
LM
s d
L
d
(17)
671
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
B. Asymptotic Analysis and Diversity Order
The PDF document should be sent as an open file, i.e.
without any data protection.
In order to study the asymptotic behavior of the outage
probability at high SNRs, we assume the following re-
parameterization for the mean SNR values among the
source, relay and destination [7]
, ,s d T s d (18)
where T is the ratio of the user transmit power to the
received noise, and ,s d is a finite constant accounting for
large-scale path loss and shadowing effects. Relative
differences in quality between the various channels are
still captured by the ,s d values. Thus, by expressing
outage probability as a function of1 T , and then letting
T (i.e., the high-SNR regime), the diversity order
is given by the smallest exponent of 1 T . Under the
Rayleigh fading assumption, the first term of (15) in the
asymptote of large behaves as (19). The same as the
above, we can obtain (20). For the integral of (15), using
the fact that exp 1x for all 0x x _ 0, we have (21).
This means that each term contributing to the sum in (15)
behaves as 11 M
T
, full diversity order is achieved.
/ /
11 1, , , ,
1 2 1 2 1 2 1 2 11 exp 1 exp 1 exp 1 exp
i i
R R R RM M
Mi is r s d T s r T s d T
(19)
/ /
, ,
1 2 1 2 11 exp exp
i i
R R
M Li is r s r T
(20)
, ,, ,
, , , ,1
, , , , , , , ,
, ,1
, ,
1 1 1 1 1exp exp exp
1 1 1
i i
i i
i i i iL L
i
iL
r d r ds d s d
s d r d s d r dLii i is d s d r d r d s d r d T s d T r dT
S S
s d r dLi is d r dT
S
d d d d
d d
1 1
, ,
1 1 1 1
i
L Lis d r dT T
(21)
IV. BIT ERROR RATE ANALYSIS
The pairwise error probability (PEP) for a coded
system is defined as deciding in favor of code word
1 2( , , , )e Ne e e when code word 1 2( , , , )c Nc c c was
transmitted. Therefore, for a binary code with BPSK
modulation, the PEP of noncooperative transmission can
be written as [10]
( | ) 2 ( )c en
P Q n
(22)
where ( )Q x denotes the Gaussian Q-function [12]. The
instantaneous received SNR values are denoted by vector
γ . The transmitted codeword is c , the erroneously
decoded codeword is e , and the set is the set of all n
for which ( ) ( )c en n , thus d is the Hamming
distance between c and e . For linear codes, the PEP
depends only on d and not the particular code words c
and e , so that the conditional PEP is typically denoted by
( | )P d .
To obtain the unconditional PEP we must average (22)
over the fading distributions
0( ) ( | ) ( )P d P d p d
(23)
where ( )p is the probability density function (PDF) of
.We can obtain an exact solution to (23) using the
techniques of Simon and Alouini. The first step is to use
the following alternative representation for the Gaussian
Qfunction, originally derived by Craig [41]-[42], and then
applied to performance analysis in fading channels in
[43]-[45]
22
20
1( ) exp
2sin
xQ x d
(24)
Substituting (24) in (22) and (23), Using the well
known MGF function method [45], for the case of
Rayleigh fading, we can obtain
12
20
1( ) 1
sin
dP d d
(25)
where is the average SNR. Equation (25) is an exact
expression for the unconditional PEP and can be easily
evaluated with numerical integration techniques.
We can obtain the following upper bound for (25) by
noting that the integrand is maximized when 2sin =1 , so
that
11
( ) 12
P d d
(26)
As can be seen from (26), for large SNRs, the PEP is
inversely proportional to the product of the average SNR
and d .
A. Pairwise Error Probability
In this section, we analyze the PEP of the proposed
coded cooperation protocol. Analogous to the outage
probability analysis, there are two possible cases
depending on how many relay stations are able to
672
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
successfully decode the first frame. Assuming,s d , , is r
and ,ir d are the mean channel SNR values. In the
following, we derive a closed-form expression for the
pairwise error probability. Although the expression is
developed for the specific case of BPSK modulation, the
analysis can be readily extended to accommodate other
types of modulation.
For Case I, none of the relay stations decodes the
source frame successfully, all the relays keep silent. So
that the conditional PEP for the destination is
, , 1 , 2 ,
,
( | , ) 2( )
2
is d r d s d s d
s d
P d Q d d
Q d
(27)
where 1d and 2d are the portions of the error event bits
transmitted respectively, such that 1 2d d d .
The average sequence error probability can be derived
by averaging (2) over the fading coefficients. Let ( )P d
be the average probability of decoding an erroneous code
sequence with weight d, we can obtain
1
2 1,
,20
1 1( ) 1 1
2sin
s d
s d
dP d d d
(28)
For Case II, some of the relays decode the source node
successfully
, , 1 , 2 , 2 ,
, 2 ,
( | , ) 2
2
i i
i
s d r d s d s d r d
i
s d r d
i
P d Q d d d
Q d d
(29)
which can be evaluated as
112 2 ,,
2 20
11
, 2 ,
1( ) 1 1
sin sin
11 1
2
i
i
r ds d
i
s d r d
i
ddP d d
d d
(30)
The BER for a turbo code is bounded by (32).
Substituting ( )P d into (32), we can obtain the BER for
every case. into (32), we can obtain the BER for every
case.
B. Bit Error Rate of Turbo Codes
The bounds for the BER and BLER of turbo codes can
be obtained using the weight enumerating function (WEF)
of the equivalent block code, as shown in [46] - [48]. The
WEF of the overall concatenated code is given based on
the WEF of the constituent codes. We follow the same
method of [47] with a minor modification for a turbo
code with 1C and 2C as the constituent recursive
convolutional codes and an interleaver with size K .
The conditional WEF of a block code, ( )C
WA Z ,gives
all possible code words generated by the set of input
sequences with weight w (note that Z is a dummy
variable). Denote by 1 ( )C
WA Z and 2 ( )C
WA Z the conditional
WEFs of 1C and
2C , respectively.
Then using the probabilistic uniform interleaver the
conditional WEF of the turbo code is [47]
1 1( ) ( )( , )
C C
C W W
W
A Z A ZA Z Y
K
w
(31)
The BER and BLER of the turbo code are obtained
using the union bound argument [47]
, ,
0 0 1
( )K K K
b w z y
z y w
wP a P d
K
(32)
, ,
0 0 1
( )K K K
block w z y
z y w
P a P d
(33)
where , ,w z ya denotes the multiplicity of code words
corresponding to input weight w and parity weights z
and y , and ( )P d is the corresponding PEP expression.
Note that 1d is equal to the summation of the exponents
of w and z , and 2d is equal to the exponent of y .
C. End-to-end Bit Error Rate
We now use the above PEP results to determine the
end to end bit error probabilities for the coded
cooperation protocol. The first step is to calculate the
probabilities of the cooperative cases. The cooperative
case probabilities are determined by the BLER of the first
frame transmission. The BLER for a turbo code is
bounded by (33).
As mentioned before, we parameterize the two cases
by 1,2 and we can express the probability for Case
I ( 1 ) as follows
,
1
( 1)i
M
block r
i
P P
(34)
where , iblock rP denote the BLER for the thi relay.
For case II, the probability is given as
1
, ,
1
( 2) 1i i
M
block r block r
L i i
MP P P
L
(35)
The overall end-to-end BER equals the average of the
BER over the two transmission scenarios as [49]
2
1
b b
i
P P P i
(36)
where bP denote the end-to-end BER.
V. NUMERICAL RESULT
In this section, we present simulation results to
compare various relaying schemes with multiple relays.
All simulations are performed assuming BPSK
modulation and a frame size of 130 symbols over quasi-
673
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
static fading channels. We use a 4-state recursive
systematic convolutional code (RSC) with the code rate
of 1/2 and the generator matrix of (1, 5/7). It is possible
to have flexible code rates through the use of RCPT [40]
codes. The baseline scheme for comparisons is a
noncooperative turbo coded system. The different
channels between the source, relays and destination are
assumed to be independent flat quasi-static Rayleigh
fading channels, where the channel coefficients are fixed
for the duration of the frame and change independently
from one frame to another.
Fig. 3. Outage probability vs. mean SNR for various cooperative
protocols for M=1,2 and 3 relay channels.
Fig. 3 plots the outage probability given by (17) versus
the SNR for four different system configurations at rate
1R b/s/Hz. All channels have equal mean SNRs. As
can be observed from Fig. 3, coded cooperation with
1M is able to achieve around 1 dB gain over repetition
based DF. Moreover, the coding gains increase with the
increase of the number of relays. For example, the gain is
nearly 10 dB for coded cooperation with 1M at the
outage probability of 510 as can be observed from Fig. 3.
By contrast, we see that the repetition based DF scheme
is not very effective in the low SNR regime. This is due
to the fact that the DF codeword is equivalent to
repetition coding, which is relatively inefficient as
opposed to distributed coding. Coded cooperation,
however, performs much better than all the other systems
in Fig. 3 at all SNRs. This is attributed to the coding
gains induced by the coded cooperation protocol.
Fig. 4. Outage probability vs. rate. All the channels have a mean SNR of 10 dB.
Fig. 4 compares the outage probabilities at different
rates for the various cooperative schemes. All the
channels have a mean SNR of 10 dB. In the low rate
regime, both DF and coded cooperation provide
significant improvements, with the latter offering
generally better overall performance. As the rate
increases, the outage probability of the DF protocol will
exceed that of its coded cooperation counterpart. This is
again a manifestation of the inefficiency of the repetition
coding nature of DF. We note that coded cooperation in
the worst scenario always performs at least as well as the
no cooperation and DF schemes.
Fig. 5 plots the outage probabilities versus the number
of relays at the fixed rate of R =1b/s/Hz. Two typical
channel SNR values are considered in Fig. 5, i.e., 10 dB
and 15 dB. As can be observed from the figure, as the
number of relays increases, the outage probability
decreases due to the increased diversity gains. However,
the additional gains obtained from increasing the number
of the relays come at the expense of approximately
linearly increased complexity, resulting in diminishing
returns.
Fig. 5. Outage probability vs. the number of relays. All the channels have the equal average SNRs, i.e., 10 dB or 15 dB.
Fig. 6. Comparing bit error rates of coded cooperation for 1, 2, and 3
relays, and all the channels have equal SNR (varies 0-20 dB).
Fig. 6 shows the comparison of the simulated BER
results and the analytical BER using (36) for the proposed
transmission scheme with L = 1, 2 and 3 relay channels,
assuming all the nodes have the equal mean SNR. As
-5 0 5 10 15 20 25 3010
-7
10-6
10-5
10-4
10-3
10-2
10-1
100
Mean SNR (dB)
Outa
ge P
robabili
ty
No cooperation (M=0)
DF (M=1)
Coded cooperation (M=1)
Coded cooperation (M=2)
Coded cooperation (M=3)
0 0.5 1 1.5 2 2.5 30
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Rate (b/s/Hz)
Outa
ge P
robabilit
y
No cooperation (M=0)
DF (M=1)
Coded cooperation (M=1)
Coded cooperation (M=2)
Coded cooperation (M=3)
1 2 3 4 5 6 7 8 9 1010
-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
The Number of Relays
Outa
ge P
robabili
ty
SNR=10dB
SNR=15dB
0 2 4 6 8 10 12 14 16 18 2010
-6
10-5
10-4
10-3
10-2
10-1
100
Mean SNR (dB)
Bit E
rror
Rate
Bound
Simulation
No cooperation (M=0)
Coded cooperation (M=1)
Coded cooperation (M=2)
Coded cooperation (M=3)
674
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
shown in Fig. 3, the analytical results are in close
agreement with our simulated BER results. Additionally,
the diversity gain that is achieved using a different
number of relays is evident in these results. Compared
with the noncooperative scheme, the gain at BER of 510
is from 12 dB (for the case of L =1) to 18 dB (for the
case of L = 3).
Fig. 7 shows the simulation results for the BER of
turbo coded cooperation using two relays under various
channel conditions. For most application scenarios, the
backhaul and cooperative links are more reliable than the
direction transmission link. In this simulation, the mean
SNR of the source to relay and relay to destination
channels are 5 dB, 10 dB or 15 dB higher. That is,
, , ,i is r r d s d +5 dB (or +10 dB, +15 dB). The union
bounds well match the simulation results curve.
Fig. 7. Comparison of the bit error rates of turbo coded cooperation for two relays with di_erent mean SNRs. The x-axis is the SNR from
source to destination. Note that four cases are considered. All the
channel have the same SNR. The mean SNR of the source to relay and relay to destination channels are 5 dB, 10 dB or 15 dB higher than that
direct transmission link.
VI. CONCLUSION
In this paper, we investigated the outage probability
and the BER of coded cooperation based on DTC with
multiple relays in Rayleigh fading channels. A finite
range single integral solution for the outage probability
expressions of the proposed turbo coded cooperation
protocol was derived. Furthermore, we also developed a
closed-form expression for the PEP and a tight upper
bound for the BER using DTC. The analytical upper
bounds were verified with simulation results. Our results
demonstrated the efficacy of the proposed protocol in
comparison with other cooperative strategies, and showed
that our scheme is able to achieve a full diversity order.
APPENDIX A APPENDIX TITLE
The integration region LS in (15) can be rewritten
1
, ,1
,
21
1i
R
r d s d
is d
(A1)
1
, ,1
,
1 21
1i
R
r d s d
s d
aL
(A2)
note that (A2) is true because each of the L cooperative
relays has the same transmission power during the second
frame transmission. Since , 0s d we have
1
,1
,
21
1
R
s d
s d
(A3)
, 2 1R
s d (A4)
We are now ready to simplify the multiple integrals in
(15) as shown in (A5).
1
1
,,
, ,
, , , ,
2 ,,
, ,0 0
, , , ,
2 ,
0, ,
1 1exp exp
1 1exp exp
1exp 1 exp
i
i
i iL
R
i
i
i i
R
r ds d
s d r d
is d s d r d r dS
a r ds d
r d s d
is d s d r d r d
s d
s d s d
d d
d d
,
,i
L
s d
r d
ad
(A5)
REFERENCES
[1] A. Sendonaris, E. Erkip, and B. Aazhang, “User
cooperation diversity. Part I: System description,” IEEE
Trans. Commun., vol. 51, no. 11, pp. 1927-1938, Nov.
2003.
[2] A. Sendonaris, E. Erkip, and B. Aazhang, “User
cooperation diversity. Part II: Implementation aspects and
performance analysis,” IEEE Trans. Commun., vol. 51, no.
11, pp. 1939-1948, Nov. 2003.
[3] J. N. Laneman, G. W. Wornell, and D. N. C. Tse, “An
efficient protocol for realizing cooperative diversity in
wireless networks,” in Proc. IEEE Int. Symp. Information
Theory, Washington, DC, Jun. 2001, pp. 294.
[4] J. N. Laneman, D. N. C. Tse, and G.W.Wornell,
“Cooperative diversity in wireless networks: E_cient
protocols and outage behavior,” IEEE Trans. Inf. Theory,
vol. 50, no. 12, pp. 3062-3080, Dec. 2004.
[5] J. N. Laneman, “Cooperative diversity in wireless
networks: Algorithms and architectures,” Ph.D.
dissertation, MIT, Cambridge, MA, 2002.
[6] T. Hunter and A. Nosratinia, “Cooperation diversity
through coding,” in Proc. IEEE Int. Symp. Information
Theory, Lausanne, Switzerland, Jun. 2002, p. 220.
[7] T. E. Hunter, S. Sanayei, and A. Nosratinia, “Outage
analysis of coded cooperation,” IEEE Trans. Inf. Theory,
vol. 52, no. 2, pp. 375-391, Feb.2006.
[8] T. E. Hunter and A. Nosratinia, “Diversity through coded
cooperation,” IEEE Trans. Wireless Commun., vol. 5, no. 2,
pp. 283-289, Feb. 2006.
[9] T. E. Hunter, “Coded cooperation: a new framework for
user cooperation in wireless networks,” Ph. D. dissertation,
Univ. Texas at Dallas, Richardson, 2004.
0 2 4 6 8 10 1210
-7
10-6
10-5
10-4
10-3
10-2
10-1
Mean SNR (dB), M=2
Bit E
rror
Pro
babili
ty
Bound
Simulation
All equal
5dB
10dB
15dB
675
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
[10] M. Janani, A. Hedayat, T. E. Hunter, and A. Nosratinia,
“Coded cooperation in wireless communications: Space-
Time transmission and iterative decoding,” IEEE Trans. on
Signal Processing, vol. 52, no. 2, pp. 362-371, Feb. 2004.
[11] M. Elfituri, W. Hamoud., and A. Ghrayeb, “A
convolutional-based distributed coded cooperation scheme
for relay channels,” IEEE Trans. Vehicular Technology,
vol. 58, no. 2, pp. 655-669, Feb. 2009.
[12] A. Stefanov and E. Erkip, “Cooperative coding for wireless
networks,” IEEE Trans. on Commun., vol. 52, no. 9, pp.
1470-1476, Feb. 2009.
[13] S. Yiu, R. Schober, and L. Lampe, “Distributed space-time
block coding,” IEEE Trans Commun., vol. 54, no. 7, pp.
1195-1206, July 2006.
[14] Y. Li and X. G. Xia, “Full diversity distributed space-time
trellis codes for asynchronous cooperative
communications,” in Proc. IEEE ISIT 2005, Sept. 2005, pp.
911-915.
[15] A. Chakrabarti, et al., “Low density parity check codes for
the relay channel,” IEEE J. Select. Area Commun., vol. 25,
no. 2, pp. 280-291, Feb. 2007.
[16] J. Hu and T. Duman, “Low density parity check codes over
wireless relay channels,” IEEE Trans Wireless Commun.,
vol. 6, no. 9, pp. 3384-3394, Sept. 2007.
[17] M. C. Valenti and B. Zhao, “Distributed turbo codes:
towards the capacity of the relay channel,” in Proc. IEEE
VTC 2003-Fall, 6-9 Oct. 2003, vol. 1, pp. 322-326.
[18] B. Zhao and M. C. Valenti, “Distributed turbo coded
diversity for relay channel,” IEE Electronics Letters, vol.
39, pp. 786-787, May 2003.
[19] Y. Li, B. Vucetic, Tan F. Wong, and M. Dohler,
“Distributed turbo coding with soft information relaying on
multi-hop relaying networks,” IEEE J. Select. Area
Commun., vol. 24, pp. 2040-2050, Nov. 2006.
[20] P. K. Sharma and P. Garg, “Outage analysis of coded
cooperation with multiple relays and Nakagami-m fading,”
in Proc. IEEE International Conference on
Communications Workshops, Budapest, 2013, pp. 322-326.
[21] P. K. Sharma and P. Garg, “Coded cooperation:
Generalised outage analysis,” IET Communications, vol. 8,
pp. 972 - 979, Jun. 2014.
[22] T. X. Vu, P. Duhamel, and M. D. Renzo, “Performance
analysis of network coded cooperation with channel coding
and adaptive DF-Based relaying in rayleigh fading
channels,” IEEE Signal Processing Letters, vol. 22, pp.
1354-1358, Sept. 2015.
[23] A. R. Heidarpour, G. K. Kurt, and M. Uysal, “Diversity-
multiplexing tradeoff for network coded cooperative
OFDMA systems,” in Proc. IEEE International
Conference on Communications, London, 2015, pp. 4368 -
4373.
[24] K. Ishii, K. Ishibashi, and H. Ochiai, “Multilevel coded
cooperation for multiple sources,” IEEE Transactions on
Wireless Communications, vol. 10, pp. 4258-4269, Dec.
2011.
[25] C. Hausl, T. Lutz, and R. Koetter, “Distributed turbo
coding with information transfer via timing of the half-
duplex relay-phases,” in Proc. IEEE ICC 2009, May 2009,
pp. 1-5.
[26] Y. Li, B. Vucetic, and J. Yuan, “Distributed turbo coding
with hybrid relaying protocols,” in Proc. IEEE PIMRC
2008, Sept. 2008, pp.1-5.
[27] Y. M. Al-Moliki, M. A. Aldhaeebi, G. A. Almwald, and M.
A. Shaobi, “The performance of RS and RSCC Coded
cooperation systems using higher order modulation
schemes,” in Proc. 6th Intelligent Systems, Modelling and
Simulation Annu., Kuala lumpur, 2015, pp. 211-214. [28] R. Bordon, S. Montejo, S. Mafra, and R. Souza, “Energy
efficient power allocation schemes for a two-user network-
coded cooperative cognitive radio network,” IEEE
Transactions on Signal Processing, no. 99, pp. 1-12, Jan.
2016.
[29] Y. Zhao, Y. Li, X. Chen, and N. Ge, “Joint optimization of
resource allocation and relay selection for network coding
aided device-to-device communications,” IEEE
Communications Letters, vol. 19, pp. 807-810, May 2015.
[30] W. Xu, Y. Wang, J. Lin, and L. Fu, “Compressed
cooperation in an amplify-and-forward relay channel,” in
Proc. 26th Annu. Conf. Personal, Indoor, and Mobile
Radio Communications, Hong Kong, 2015, pp. 932-936.
[31] R. Tutgun and E. Aktas, “Cooperative network coded ARQ
strategy for broadcast networks,” in Proc. IEEE
International Conference on Communications, London,
2015, pp. 1819-1824.
[32] M. M. Salah and A. A. Elrahman, “Power allocation
strategies for distributed turbo codes in relay networks,” in
Proc. Signal Processing, Communications and Computing,
Ning Bo, 1994, pp. 1-5.
[33] P. K. Sharma and P. Garg, “Outage analysis of full duplex
coded cooperation,” in Proc. Twentieth National
Conference on Communications, Kanpur, Uttar Pradesh,
2014, pp. 1-5.
[34] P. K. Sharma and P. Garg, “Coded cooperation:
Generalised outage analysis,” IET Communications, vol. 8,
pp. 972-979, June 2014.
[35] C. Sunu and E. Aktas, “Adaptive modulation and coding
(AMC) in cooperative communication channels,” in Proc.
22nd Signal Processing and Communications Applications
Conference, Trabzon, 2014, pp. 1571-1574.
[36] H. Feng, H. Wang, L. Dai, and L. J. Cimini, “To cooperate
or not to cooperate: An outage analysis of interference-
limited wireless networks,” IEEE Trans. on Wireless
Communications, vol. 13, pp. 822-833, Feb. 2014. [37] X. Chen, H. H. Chen, and W. Meng, “Cooperative
communications for cognitive radio networks - from theory
to applications,” IEEE Communications Surveys &
Tutorials, vol. 16, pp. 1180-1192, July 2014.
[38] A. Gupta, P. K. Sharma, P. Garg, and G. Sharma, “Outage
analysis of free space optical coded cooperation in weak
atmospheric turbulence,” in Proc. Signal Processing and
Integrated Networks, Noida- Delhi NCR, 2015, pp. 207-
210.
[39] C. Li, Y. Wang, W. Xiang, and D. Yang, “Performance
analysis for coded cooperative multiple-relay in distributed
turbo channels,” in Proc. IEEE 22nd International
676
Journal of Communications Vol. 11, No. 7, July 2016
©2016 Journal of Communications
Symposium on Personal, Indoor and Mobile Radio
Communications, Toronto, 2011, pp. 1511-1515.
[40] D. N. Rowitch and L. B. Milstein, “On the performance of
hybrid FEC/ARQ systems using rate compatible punctured
turbo (RCPT) codes,” IEEE Trans. Commun., vol. 48, pp.
948-959, June 2000.
[41] J. G. Proakis, Digital Communications, 3rd ed., New York:
McGraw-Hill, 1995.
[42] J. W. Craig, “A new, simple, and exact result for
calculating the probability of error for two-dimensional
signal constellations,” in Proc. IEEE MILCOM, McLean,
VA, Oct. 1991, pp. 571-575.
[43] J. Hagenauer, E. O_er, and L. Papke, “Iterative decoding
of binary block and convolutional codes,” IEEE Trans.
Inform. Theory, vol. 42, pp. 429-445, Mar. 1996.
[44] S. Benedetto, D. Divsalar, G. Montorsi, and F. Pollara, “A
softinput soft-ouput APP module for iterative decoding of
concatenated codes,” IEEE Trans. Commun., vol. 1, pp.
22-24, Jan. 1997.
[45] M. K. Simon and M. S. Alouini, Digital Communication
over Fading Channels: A Unified Approach to
Performance Analysis, NewYork: John Wiley and Sons,
2000.
[46] C. Berrou and A. Glavieux, “Near optimum error
correcting coding and decoding: Turbo codes,” IEEE Trans.
Commun., vol. 44, pp. 1261-1271, Oct. 1996.
[47] S. Benedetto and G. Montorsi, “Unveiling turbo codes:
Some results on parallel concatenated coding schemes,”
IEEE Transactions on Information Theory, vol. 42, no. 2,
pp. 409-428, March 1996.
[48] S. Benedetto, D. Divsalar, G. Montorsi, and F. Pollara,
“Serial concatenation of interleaved codes: Performance
analysis, design, and iterative decoding,” IEEE Trans.
Inform. Theory, vol. 44, pp. 909- 926, May 1998.
[49] A. Nosratinia, T. E. Hunter, and A. Hedayat, “Cooperative
communication in wireless networks,” IEEE Commun.
Mag., vol. 42, pp. 74-80, Oct. 2004.
Jing An was born in Hebei Province,
China, in 1980. She received the B.S.
and M.S. degree from the Shijiazhuang
Railway Institute, Shijiazhuang, in 2004
and 2008, respectively, both in computer
science. She is currently working in
Department of Information Engineering,
Shijiazhuang Vocational Technology
Institute. Her research interests are in the broad area of
communications and information theory, particularly coding
and signal processing for multimedia communications systems.
Chang Li was born in Hebei Province,
China, in 1980. He received the B.S.
degree from the Hebei University of
Science and Technology of China,
Shijiazhuang, in 2003, the M.S. degree
from the Shijiazhuang Railway Institute,
Shijiazhuang, in 2006, and the Ph.D.
degree from Beijing University of Posts
and Telecommunications (BUPT), Beijing, in 2011, all in
electrical engineering and communications. He is currently
working in China Mobile Group Hebei Co., Ltd. as senior
engineer. His research interests focus on wireless
communications theories and Technologies, especially the
cooperative technology and channel coding.