Volume 4, Issue 4, April 2014 ISSN: 2277 128X

International Journal of Advanced Research in

Computer Science and Software Engineering
Research Paper
Available online at: www.ijarcsse.com
A Novel Multi-hop Threshold based Relay Selection Scheme
K.Shamganth* Dr.Martin Sibley Dr.Sammi Ghnmi
Research Student, Reader Head of Department
School of Computing & Engineering School of Computing & Engineering Engineering Department
University of Huddersfield,UK University of Huddersfield,UK Ibra College of Technology,Oman

Abstract— Relay selection in co-operative networks plays a vital role in increasing the efficiency of networks. If the
selected relay is not the best relay then the end-to-end co-operative network will be in failure. In this proposed relay
selection scheme, threshold-based relay selection is applied in the first-hop the source-to-relay link. From the
destination, the best relay will be selected based on Output Threshold Multiple Relay Selection (OT-MRS) scheme [8].
If the relay selected by the source and destinations are different then multi-relay transmission will take place between
the relays. In this paper closed form SER formulations of the system are derived.

Keywords— Co-operative networks; Relay selection; Outage probability; Interference; Co-operative relaying

I. INTRODUCTION
Co-operative communication is one of the fastest growing research areas, and it will be the key enabling technology in
LTE – (Long Term Evolution) -Advanced standard. The key idea in user co-operation [1] is resource sharing among
multiple network nodes. Exploration of user co-operation leads to savings of overall network resources. Relay selection
is one important research area in co-operative communication where there is numerous relay selection approaches
proposed in the literature [2]. Different relay selection schemes were compared based on Relaying candidate selection,
optimal relay assignment criterion, Co-operative transmission scheme and type of relay selection i.e. reactive or proactive
in [3]. Threshold based relay selection using Amplify and Forward (AaF) relaying scheme [4] and Decode and Forward
(DaF) relaying scheme is widely analysed in the literature [5-7]. In threshold based relay selection schemes the relays
were selected based on those relays having a received SNR higher than a threshold value. Threshold based schemes are
of two types: destination based and source based scheme. Relay selection is performed based on received SNR at the
destination during the last hop. In this paper we propose a hybrid threshold based scheme, which has the combination of
source based relay selection and destination based relay selection schemes. In the first part of the relay selection the
source selects the best relay based on the SNR value greater than the threshold SNR, and the destination selects the best
relay based on OT-MRS scheme [8]. If the relays selected by the source and the destination are different, then multi-relay
communication will take place between the relays. In the Multi-Relay Selection (MRS) schemes in the literature the
number of active relays is many and so the power consumption is very high. In the MRS schemes like OT-MRS [8]
threshold is checked for the output SNR at the destination. But in the proposed method the relays will forward the
amplified version of the source information only when the SNR at the relays is above the threshold.
The rest of this paper is organized as follows. In Section II, we describe the system model and the proposed algorithms
in Section III, performance analysis is presented in Section IV and Section V concludes the paper with the summary of
results.
II. SYSTEM MODEL
We consider a half-duplex, multi-hop system where there are source S1, destination D1 and N relay nodes with Ri i∈
{1,2,….N}. Relay selection is classified into two types, reactive and proactive [6]. In reactive mode, the source
broadcasts its information and selects the best relay. On the other hand, in proactive mode the source transmission takes
place only after selection of the best relay. In this paper the proactive approach is used. In the system model shown in
Fig.1, the source S1 transmits information to the destination D1 though the best relay. Each relay in the system model
uses the amplify and forward relaying scheme. It is assumed that all the relays use half-duplex communication and each
relay has one antenna. So each relay amplifies the received signal and retransmits it to the destination.

Fig. 1 System Model

© 2014, IJARCSSE All Rights Reserved Page | 526

 Shamganth et al., International Journal of Advanced Research in Computer Science and Software Engineering 4(4),
April - 2014, pp. 526-530
There are two-phases followed in this work: In the first phase the source broadcasts its information. In this phase,
destination D1 receives the signal transmitted by source S1 direct link and from the relays. If the SNR for the link
between S1  R1 is above the threshold i.e  S R1   T then the relay R1 will forward the amplified version of the
1

source information. If not, the first relay will be in silent mode. All other relays will compare  S1Ri   T for
 
i  1,2....M the SNR of the first hop is above the threshold then the relays will be in active mode. Relays in active
mode will forward the amplified version in the first time slot. The combiner at destination D1 combines the signal
received from the source and from the relays. If the combiner output exceeds the threshold value then that relay will be
selected as the best relay from the destination side. The destination will send the best relay information to the source S1.
The relays with  S1Ri   T calculate the maximum SNR between the Source and the relays i.e max  S1 ,Ri    T
 
for i  1,2....M . The best relay from the source S1 will be selected based on the above condition. If the relays
selected by the source S1 and destination D1 are different then multi-relay transmission will take place between the
source S1 and destination D1 as shown in Fig.1.

The received signals ys1,D1 and ys1,Ri at the ith relay can be written as

ys1,D1  P1 hs1,D1x  ns1,D1

(1)
y s1, Ri  P1 hs1, Ri x  ns1, Ri for i  1,2...N  (2)
Notations followed in the equations of this paper are as follows:
 P1 – Transmitted power at the source
 hs1,Ri – Channel coefficient from the source to ith relay
  s1,Ri –SNR between source S1 to the relay node Ri
 ns1,Ri – Additive noise added in the channel between source S1and Relay Ri
 Pi – ith relay node power.

The received signal at the destination node in phase2 due to the ith relay transmission is
Pi
y Ri,D1  h Ri,d1 ys1,Ri  n Ri,D1
2
P1 h s1,Ri  N0
(3)
The channel coefficients hRi ,D1 and h s1,Ri are modelled as zero-mean, complex Gaussian random variables with

variances 2Ri,D1 , s1,Ri

2
, at the receiving node but not at the transmitting nodes. The noise terms are modeled as zero-
mean, complex Gaussian random variable with variance N0. The signal received from the source and from relays is
combined at the destination by the use of Maximum Ratio Combining (MRC).

III. REVIEW OF OT-MRS SCHEME

OT-MRS scheme is proposed in [4] selects the Lc relays out of total L relays when the combined SNR of the Lc
relayed paths and the direct path exceeds the pre-set threshold (γth). In this relay selection scheme, the destination (D)
receives the signal transmitted by source (s1) during the first phase. Next, the first relay R1 forwards the amplified
version of the source information to the destination in the second phase. The combiner at the destination combines this
signal from the relay and the direct path signal. If the SNR at the output of the combiner at the destination exceeds the
threshold, then no more relays will be selected. Otherwise, the relays R2,R3……RN will be selected in the subsequent
timeslots until the sum of the output SNR exceeds the threshold.

The instantaneous output SNR i L

i 1 at D with i active AF given in [8] as
N  s 1,Ri  Ri ,D 1
 i   s 1,D 1  
i 1 s 1,Ri   Ri ,D 1  1 (4)

To analyse the system performance, the distribution of

l is required, so
 l is replaced by a tight upper bound  ub
l in [4]
and it is written as
N
 l <  lub =  s1,D   i
i 1 (5)

© 2014, IJARCSSE All Rights Reserved Page | 527

Shamganth et al., International Journal of Advanced Research in Computer Science and Software Engineering 4(4),
April - 2014, pp. 526-530
 i  min(  s1,Ri ,  Ri ,D )
(6)
The tight lower bound for ζi is given by
N
1
 l ≥  llb   s1D     i
i 1  
2 (7)

IV. PROPOSED RELAY SELECTION SCHEME

In this section, we describe the proposed relay selection algorithm. The main idea of this algorithm is to reduce the
channel estimation and power at the relay nodes. Since the power control is applied in this algorithm the lifetime of the
relay node is maximized. In the algorithm proposed in [8] the bandwidth consumption is more due to increased
signalling message exchange between the source and the relay node.
To overcome the problems of power consumption and complexity issues, we propose the relay selection algorithm
with the predetermined threshold at the relay nodes and at the destination node. Mode of operation in the proposed
algorithm is described as follows:

Step1: Initialize i0

Step2: Set i  i  1 for i  1,2,...N , if i  N  1 , go to Step 6
Step3: If  s1,Ri   T go to step 2 else go to step 4
Step4: If Relays with  S1,Ri   T then it will be in listening mode, all other relays will be in sleep mode. So power
consumption is reduced.
Step5: Relays in listening mode will amplify and forward the Source (S1) information to the destination (D1).
Step6: Compute 1  max  s1,Ri  
Step7: Destination D1 will combine the signal received directly from the source s1 and the relayed transmission by
using optimal combiner as in [4], if the value of the particular relayed transmission is above the predetermined threshold
 
i.e,  2  max  Ri , D1   s1, D1   T then the relay will be chosen as the best relay from destination.
Step8: If 1   2 then multi-relay transmission will takes place between the best relay selected by the source S1 and
destination D1.
.
A. Difference of Proposed algorithm from OT-MRS scheme
In the proposed algorithm, in the first timeslot the source broadcasts its information so the relays in the coverage area
and the destination D1 will receive source s1 information. During this phase the SNR at the relays will be calculated and
compared with the predefined threshold γT. This SNR threshold can be chosen to be the minimum required SNR for
successful symbol decoding for a given modulation scheme.
Relays with SNR γRi>γT for i= {1, 2…N} will be in active mode and all other relays will be in sleep mode. By this
proposed method the power consumption due to the relays is less. Also spectral efficiency is good in the proposed
method since only relays with SNR γRi>γT will be in listening mode and all other relays will not forward the source
information as in OT-MRS. If more than one relay has a SNR more than the threshold SNR then the relays will forward
to the destination. In the destination node the SNR of the direct path and the best relayed path will be combined by the
maximal ratio combining (MRC) technique and this output compared with preset threshold SNR set at the destination. If
the cumulative SNR of the best relay and the direct path is greater than the threshold SNR then the relay will be chosen
as the best relay by the destination.

B. Numerical Example
Consider there are four relays Ri for {i=1,2,3,4} during the first phase source broadcasts its information, so all the four
relays and destination D will receive the source information. In the relays, a predefined threshold is set and if two relays
R1 and R2 with SNR greater than the threshold value, then both relays R1 and R2 will amplify and forward source
information to the destination.
 
If γRi>γT and  s1,D1   R1   T
In this case relay R1 will not be selected as the best relay by the destination. So the destination will combine the SNR
of the second best relay R2 with the destination SNR s1, D and if this output is more than the threshold value at the
destination then relay R2 will be chosen as best relay by the destination.

V. SER FORMULATIONS OF THE PROPOSED RELAY SELECTION ALGORITHM

A closed form derivation of SER expression for AF cooperation protocol [9] of BPSK signal is presented. With the
knowledge of the channel coefficients hs1,D1 and hRi,D1, the destination detects the transmitted symbols by jointly
combining the received signal ys1,D1 from the source and yRi,D1 from the relay.
The combined signal at the output of MRC detector [11]is given by

© 2014, IJARCSSE All Rights Reserved Page | 528

Shamganth et al., International Journal of Advanced Research in Computer Science and Software Engineering 4(4),
April - 2014, pp. 526-530
N
y D 1  a1 y s 1,D 1   ai y Ri ,D 1
i 1 (8)
where a1 and a2 are the combining factors,

P0 h* s1, D1
a1 
N0
P0 Pi
h* s1,Ri h* Ri , D1
2
P0 hs1, Ri  N 0
ai 
 P h 2

 i Ri , D1

P h 2
 N 
 0 s1, Ri 0 

If we assume the transmitted symbol „x‟ has an average energy of 1, then the SNR at the MRC detector output is
N
 D1   s1    i
i 1 (9)
2 2 2
P0 hs1,D 1 P0 Pi hs1, Ri h Ri , D1
 s1  i  2 2
Where N0 and Pi h Ri , D1  P0 hs1, Ri  N 0

The instantaneous SNR γi can be tightly upper bounded [11]as

2 2
P0 Pi hs1, Ri h Ri , D1
i  2 2
Pi h Ri , D1  P0 hs1, Ri  N 0
(10)
It is the harmonic mean of
2 2
P0 hs1, Ri Pi h Ri , D1
and
N0 N0
For BPSK modulation with instantaneous SNR
 i , the SER with the CSI is given by

 
  sin 2   
  PSK   
1 2  2  d
 exp  
PSK
PCSI
 0  sin    
2

 
  (11)
Let us denote the MGF of a random variable

M z  s    exp  sz Pz  z 
 (12)
Averaging the conditional SER over the Rayleigh fading channels the SER of BPSK signals is given by

   

 2
sin 2     sin 2   
1  2   M 
N
 2   d
PSER  M s  
 0  sin     i 1  sin    
2 i 2
(13)
   
   

Derivation of MGF of γS is shown in Appendix A, and it can be written as

1
M s 
sP  2 (14)
1  0 s1, D1
N0
© 2014, IJARCSSE All Rights Reserved Page | 529
 Shamganth et al., International Journal of Advanced Research in Computer Science and Software Engineering 4(4),
April - 2014, pp. 526-530
VI. CONCLUSION
In this paper, we introduced the idea of combining two different relay selection schemes; threshold based relay
selection and OT-MRS schemes. It has been shown that the proposed scheme reduces the power consumption of the
relay nodes and also improves the spectral efficiency. Closed form expressions of SER for the proposed scheme were
derived.

REFERENCES
[1] Andrew Sendonaris, Elza Erkip and Behnaam Azhang, “User Cooperation Diversity-PartI-System
Description,”IEEE Transactions on Communications,vol.51,pp1927-1958,Nov.2003.
[2] S.Abdulhadi,M.Jaseemuddin and A.Anpalagan, “A Survey of Distributed Relay Selection Scheme in Co-operative
Networks,” Springer Wireless Personal Communication,vol.63,pp917-935.2012.
[3] Aggelos Beltas,Yundon Shin and Moe Z, “Co-operative Communications with Outage-Optimal Opportunistic
Relaying,” IEEE Transactions on Wireless Communications,vol.6,Sep.2007.
[4] J.Nicholas Laneman, “Co-operative Diversity in Wireless Networks: Efficient protocols and Outage behavior,”
IEEE Trans.Information Theory, vol.50, ,pp.3062-3080,Dec.2004.
[5] Kyu-sung Hwang, Young-Chai Ko, “An Efficient Relay Selection Algorithm for Co-operative Networks,” IEEE
66th Vehicular Technology Conference, pp.81-85,2007.
[6] Furzan Atlay Onat, “Threshold based relay selection in Co-operative Wireless Networks,” IEEE-GLOBECOM,
2008.
[7] Furzan Atlay Onat, “Threshold Selection for SNR-based Selective Digital Relaying in Co-operative Wireless
Networks,” IEEE Transactions on Wireless Communications, vol.7, pp.4226-4237,Nov.2008.
[8] Amarasuriya, G., Ardakani, M., and Tellambura, C, “Output-threshold multiple-relay-selection scheme for co-
operative wireless networks,” IEEE Transactions on Vehicular Technology, vol.59, pp. 3091–3097,2010.
[9] P. A. Anghel and M. Kaveh, “Exact symbol error probability of a co-operative network in a Rayleigh fading
environment,” IEEE Transactions on Wireless Communications, vol. 3, pp. 1416–1421, Sept.2004.
[10] Mischa Dohler and Yonghui Li, “Co-operative Communications: Hardware,Channel & Phy,”1st edition,John Wiley
& Sons,2010.
[11] K.J.RayLiu, Ahmed K.Sadek,Weifung Su and Andres Kwansenski,”Co-operative Communication and
Networking,”Cambridge University Press,2010.

APPENDIX A
X1X2
If X1 and X 2 are two independent exponential random variable with parameters 1 ,  2 and z  is the
X1  X2
harmonic mean of z. MGF of z in [10] is given as
     1   2  s  21 2s ln 1   2  s   
2 2

Mz s   1 2
2 3 41 2 (15)

1  2   2 1   2  s  s2
2
Where 
N0 N0
With 1  and  2 
P0  s1, Ri
2
Pi  Ri
2
, D1

At high SNR, for any relay if both 1 and  2 is zero then   s

Thus MGF of (15) can be approximated as
1   2 21 2  s 2 
Mz s    2 ln  
s s  1 2  (16)
At high SNR, the MGF can be further simplified as
1   2
Mz s  
s

© 2014, IJARCSSE All Rights Reserved Page | 530