You are on page 1of 5

Outage Performance of MIMO Cognitive Relay

Networks with Antenna Selection


Hao Zhang 1,2 Xinjie Wang 1,3 T. Aaron Gulliver2 Wei Shi1 Hongjiao Zhang1
1
Department of Electrical Engineering, Ocean University of China, Qingdao, China
2
Department of Electrical and Computer Engineering, University of Victoria, Victoria, BC Canada
3
College of Communication and Electronics, Qingdao Technological University, China
Email: zhanghao@ouc.edu.cn wangxinjie@qtech.edu.cn agullive@ece.uvic.ca
shiwei6670595@126.com zhjouc@126.com

Abstract—In this paper, a novel antenna selection strategy is among the Random Variables (RVs) associated with the
proposed and analyzed for dual hop MIMO cognitive amplify- end-to-end Signal-to-Noise Ratio (SNR) was considered.
and-forward (AF) relay networks over independent non- This is the key difference between cognitive relay networks
identically distributed (i.n.i.d.) Nakagami-m fading channels. and conventional relay networks.
Spectrum sharing is employed with secondary users
Multiple-Input Multiple-Output (MIMO) technology is an
constrained by a limit on the interference to the primary
receivers. All secondary users are equipped with multiple effective means of providing multiplexing and diversity
antennas and have a maximum power constraint. The best gains in radio systems [10]. It has been considered in
antenna pairs are selected for secondary user communications. cognitive radio networks for interference control at the
The outage probability is derived and used to evaluate the primary receiver [11]. MIMO techniques have typically
impact of antenna selection on the outage performance. The been employed in cognitive networks with only multiple
analytic results are validated by Monte Carlo simulation. antennas at the transmitter [12],[13]. Consequently, along
with the performance gains comes hardware complexity at
Keywords-MIMO, Congnitve relay network, Amplify-and- the radio front end. Antenna selection can be used to reduce
forward, Antenna selection
this complexity while retaining much of the benefit of
multiple antennas [14],[15]. To the best of our knowledge,
I. INTRODUCTION the performance of antenna selection in MIMO cognitive
Cognitive radio is a promising technique to improve relay networks has not yet been investigated.
spectrum utilization [1]. Several cognitive radio techniques In this paper, a MIMO cognitive AF cooperative relay
have been proposed, in particular spectrum sharing. In this network is considered with antenna selection. The statistical
case, secondary users are allowed to use the licensed dependence among the RVs associated with the SNR is
spectrum of a primary user under interference power examined, and an accurate closed form expression for the
constraints. outage probability (OP) is derived for the best antenna
Cooperative relays are an efficient means of mitigating selection over independent and non-identically distributed
the effects of fading in wireless networks [2]. They are well (i.n.i.d.) Nakagami-m fading channels. Rayleigh fading is a
suited to a cognitive spectrum sharing radio network special case of Nakagami-m fading, so the corresponding
because the power of the secondary users is limited. There OP can easily be obtained using the results presented here.
are several possible cooperative protocols such as Amplify- The analytic expressions obtained are validated through
and-Forward (AF) and Decode-and-Forward (DF) [2]. Monte Carlo simulation, and the impact of several key
These protocols can be used in a Cognitive Relay Network system parameters on the performance, such as the number
(CRN) to enhance the performance of secondary users. CRN of Secondary User (SU) antennas and the fading parameters,
performance with various protocols has been investigated in is evaluated.
the literature. The outage performance of a cognitive AF The rest of this paper is organized as follows. In Section
relay network has been derived for Rayleigh fading II, the system model is presented. The outage performance of
channels with an interference constraint from the primary the MIMO cognitive AF relay network is derived in Section
user [3],[4]. For a cognitive DF relay network, the outage III. Some performance results are presented in Section IV to
performance with multiple relays over Rayleigh fading verify the analysis and determine the impact of the system
channels has been derived [5]-[7]. The outage performance parameters. Finally, Section V gives some concluding
for a cognitive AF relay network with Nakagami-m fading remarks.
channels was analyzed in [8],[9]. The statistical dependence

10 978-1-4799-1501-9/13/$31.00 ©2013 IEEE


II. SYSTEM MODEL 2
hSt Rn ~ (m1 , ω1 ) , t = 1, 2,! , N S , n = n = 1, 2,! , N R ,
The MIMO cognitive AF relay network model is shown in
Fig. 1. The primary transmitter (PUT) u sends messages to 2

the primary receiver (PUD) v. The primary receiver has one


hRk Dl ~ (m2 , ω 2 ) , k = 1, 2,! , N R , l = 1, 2,! , N D ,
antenna. The secondary source S sends messages to the 2
secondary destination D via a single relay R. There are NS hSt v ~ (m3 , ω 3 ) , t = 1, 2,! , N S , and
antennas at the source S, NR antennas at the relay R, and ND 2
antennas at the destination D. It is assumed that a direct hRk v ~ (m4 , ω 4 ) , k = 1, 2,! , N R .
communication link between the secondary source and
secondary destination is not available, as is reasonable with As is typical, time division multiplexing is assumed. In
fading channels and limited transmit power. the first time slot, based on (1), the received signal at
antenna n of the relay R from antenna t of the source S can
be written as
ySt Rn = PSt hSt Rn x + nRn (4)

where x denotes the signal sent by the source, hSt Rn is the


channel coefficient between antenna n of the relay and
antenna t of the source, and nRn ~ CN (0, N 0 ) . The “best”
transmitter-relay antenna pair is selected between the source
and relay. This selection should consider the channel
between S and R and the interference power constraint at the
PUD. Thus the best antenna pair (a,b) between S and R is
given by
­ ­ ½ ½
Fig. 1. The MIMO cognitive relay network model.
hSa Rb
2
= arg max ® min ® P,
° ° I ° 2 ° (5)
h
2 ¾ St Rn ¾
hSt Rn t =1,2.... N S
n =1,2.... N R °
¯ °¯ hSt v °¿ °¿
The interference generated by the secondary users at the
In the second time slot, the received signal at antenna l of
PUD should remain below the interference threshold I. The
the destination D from antenna k of the relay R can be
transmit power at antenna t of source S, denoted St
written as
t=1,2,..,NS, and antenna k of relay R, denoted Rk k=1,2,..,NR,
must satisfy yRk Dl = PRk hRk Dl xR + nDl , (6)
­ ½
)
­ ½
°
PSt = min ® P,
I °,
2 ¾
°¯ hSt v °¿
°
PRk = min ® P,
I °,
° hRk v °

(1) where x R = Gy S R , G = 1/ PS hS R
a b a a b ( 2
+ N 0 , and nDl ~
¯ ¿ CN (0, N 0 ) . Note that antenna c of the best relay-destination
where P is the maximum secondary user transmit power, pair may be different from antenna b of the best transmitter-
hSt v is the channel coefficient between St and v, and hR v is
k
relay pair. The “best” transmitter-receiver antenna pair (c,d)
the channel coefficient between Rk and v. The instantaneous is selected between the relay R and destination D. Similar to
channel gains between antenna pairs are assumed to be above, this pair is given by
Nakagami-m random variables. If X is a Nakagami-m ­ ­ ½ ½
random variable with parameters (m,Ȧ), the probability hRc Dd
2
= arg max ® min ® P,
° ° I ° 2 ° (7)
h
2 ¾ Rk Dl ¾
k =1,2.... N
density function (PDF) and cumulative distribution function hRk Dl R
° °¯ hRkν °¿ °
(CDF) can be written as
l =1,2.... N D ¯ ¿
m
§m· The received SNR at the destination can be expressed as
¨ ¸
ω § m ·
f X ( x ) = © ¹ x m −1 exp ¨ − x ¸ (2) γ 1γ 2
γD = ≈ γ = min {γ 1 , γ 2 } (8)
Γ ( m) © ω ¹ γ 1 + γ 2 + 1 up
§ m · PS PR
Γ¨ m, x ¸ 2 2
ω ¹ where γ 1 = a hSa Rb and γ 2 = c hRc Dd .
FX ( x ) = 1 − © (3)
N0 N0
Γ(m )
To simplify the notation, let

11
III. OUTAGE PERFORMANCE ANALYSIS K! M

¦ ∏( x )
nj
( x0 + x1 " + xM ) =
K
M j
The Outage Probability (OP) is defined as the probability
that the instantaneous SNR at D falls below a predefined
n0 , n1 ,!, nM
n0 + n1 +"+ nM = K ∏ ( n )!
j =0
j
j =0

threshold γ th (12)
P2 can be expressed as
Pout = Pr (γ D ≤ γ th )
C Nk R ( ( m1 − 1) !) § m3 ·
k m3
NR

( ) P2 = ¦ ( −1)
k
= Pr min {γ 1 , γ 2 } ≤ γ th ¨ ¸
( Γ ( m1 ) ) Γ ( m3 ) © ω3 ¹
k
k =0

( 1
)(
= 1 − 1 − Fγ ( γ th ) 1 − Fγ ( γ th )
2
) (9) m −1
1
¦ jl j
Thus Fγ ( γ th ) and Fγ ( γ th ) are required. Since the k! § N 0 m1 · j =0
1 2 × ¦ ¨ z¸
© I ω1
m1 −1
derivations are similar, we focus on Fγ ( z ) , which ¹
1
is l0 ,l1 ,!,lm1 −1
l0 + l1 +"+ lm1−1 = k ∏ ª
¬( )
l j ! ( j !)
lj
º
¼
j=0
given by
§ ­ ­ ½ 2
½ · § m1 −1 §N m m ·I ·
¨ ° ° I ° hst Rn ° ¸ × Γ ¨¨ ¦ jl j + m3 , ¨ 0 1 zk + 3 ¸ ¸¸
Fγ 1 ( z ) = Pr max ®min ® P, 2¾ ¾< z¸ © j =0 © I ω1 ω3 ¹ P ¹
¨ t =1,2.... NS
¨ ° °¯ hst v °¿ N 0 ° ¸ 1
© n =1,2.... NR ¯ ¿ ¹ × m −1
(13)
1
ª
{ }
I/P ¦ jl j + m3
§ 2 N · § N 0 m1 m · j =0
= « ³ F ¨ max hS R < z 0 ¸ f 2 ( y ) dy ¨ zk + 3 ¸
¬ 0 © n =1,2.... NR
t n
P ¹ hStν © I ω1 ω3 ¹
Substituting (11) and (13) into (10), Fγ ( z ) is obtained as
NS

{ } º

§ 2 N ·
+ ³ F ¨ max hSt Rn < zy 0 ¸ f 2 ( y ) dy »
1

I /P © n =1,2.... N R I ¹ hStν ¼ Fγ 1 ( z , m1 , ω1, m3 , ω3, N R , N S , I , P ) (14)


ª and Fγ ( z ) can be obtained similarly as
« § § m N ··
NR 2

« I / P ¨ Γ ¨ m1 , 1 0 z ¸ ¸ Fγ 2 ( z , m2 , ω2, m4 , ω4, N D , N R , I , P ) (15)


= « ¨1 − ©
ω1 P ¹ ¸
«³¨ f 2 ( y ) dy
Γ ( m1 ) ¸ hStν Then substituting (14) and (15) into (9) gives Pout.
« ¨ 0
¸¸
« © ¨
¹
« 

IV. NUMERICAL RESULTS


¬ P1

NS In this section, numerical results are presented to


º
NR
» validate the above analysis and evaluate the system
§ § mN ·· performance. Figure 2 shows the Outage Probability (OP) of
∞ ¨
Γ ¨ m1 , 1 0 yz ¸ ¸ »
¨ © ω1 I ¹ ¸ » (10) the proposed antenna selection strategy for a MIMO
+ ³ 1− f 2 ( y ) dy » cognitive AF relay network with different numbers of
¨ Γ ( 1)
m ¸ h
»
Stν
I /P
¨¨ ¸¸ antennas. The analytic results are shown as solid lines, and
© ¹ » the Monte Carlo simulation results are shown as points. We

»
P2 ¼ assume P is infinite, m1=m2=2, m3=m4=1, γ th = 3 dB,
so P1 can be expressed as Ȧ1=Ȧ2=Ȧ3=Ȧ4=1, and N0=1. The OP corresponding to
NR
§ single antenna transmitters and receivers is also given for
§ § m1 N 0 · · § m3 I · ·
¨ Γ ¨ m1 , z¸¸ ¨ Γ ¨ m3 , ¸¸ comparison in order to show the performance gains with
¨ © ω1P ¹ ¸ ¨1 − © ω3 P ¹ ¸ (11) antenna selection. From this figure, we can conclude that
P1 = 1 −
¨ Γ ( m1 ) ¸ ¨ Γ ( m3 ) ¸ increasing the number of antennas at the relay provides a
¨¨ ¸¸ ¨¨ ¸¸ greater performance improvement than increasing the
© ¹ © ¹ number of antennas at the source or destination. The reason
is that relay antennas can improve the performance for both
After some mathematical manipulation, and using (8.352.4) hops. Further, the agreement between the Monte Carlo
and (8.350.2) in [16] and the following well known simulation and analytic curves verifies the accuracy of the
expression above derivation.

12
figure shows that a change in the fading parameter for one
of the two hops affects the outage probability equally, i.e.,
m1=1 and m3=3 or m1=3 and m3=1. Also, an increase in m
or Ȧ for one hop improves the performance.

Fig. 2. Outage performance with different numbers of


antennas.

Fig. 4. Outage performance with different fading parameters.

V. CONCLUSION
In this paper, a MIMO cognitive radio network with
amplify-and-forward cooperative relaying was considered.
To reduce the complexity, antenna selection was used at the
transmitters and receivers. A closed form expression for the
outage probability was derived for independent and non-
identically distributed Nakagami-m fading channels and
validated by Monte Carlo simulation. The impact of key
parameters on the system performance was examined. The
results presented show the advantage of antenna selection in
terms of reducing the outage probability.

Fig. 3 Outage performance with different maximum power


ACKNOWLEDGEMENT
values P (in dB).
This work was supported in part by the International
Figure 3 shows the outage probability of the proposed S&T Cooperation Program of Qingdao, China, under Grant
antenna selection strategy for a MIMO cognitive AF relay No. 12-1-4-137-hz.
network with different values of P and numbers of antennas.
REFERENCES
The analytic results are shown as solid lines, and the Monte
Carlo simulation results are shown as points. The channel [1] A. Goldsmith, S. A. Jafar, I. Maric, and S. Srinivasa, “Breaking
spectrum gridlock with cognitive radios: An information theoretic
fading parameters are the same as for Fig. 2. This figure perspective,” Proc. IEEE, vol. 97, no. 5, pp. 894-914, May 2009.
shows that the maximum power limit P can cause outage [2] J. N. Laneman, D. Tse, and G. W. Wornell, “Cooperative diversity in
saturation. However, increasing the number of antennas can wireless networks: Effcient protocols and outage behavior,” IEEE
overcome this problem. The agreement between the Monte Trans. Inform. Theory, vol. 50, no. 12, pp. 3062-3080, Dec. 2004.
Carlo simulation and analytic curves also verifies the [3] T.Q. Duong, V. N. Q. Bao, H. Tran, and H.-J. Zepernick, “Exact
accuracy of the preceding analysis. outage probability of cognitive AF relaying with underlay spectrum
sharing,” Electron. Lett., vol. 47, no. 17, pp. 1001-1002, Aug. 2011.
Figure 4 shows the effect of the fading parameters on
[4] J. Chen, Z. Li, J. Si, H. Huang, and R. Gao, “Outage probability
the outage probability with NS=NR=ND=2. In this figure, analysis of partial AF relay selection in spectrum-sharing scenario
only analytic results are given, and P is again assumed to be with multiple primary users,” Electron. Lett., vol. 48, no. 19, pp.
infinite. Note that m=1 corresponds to Rayleigh fading. This 1211-1212, Sep. 2012.

13
[5] Y. Guo, G. Kang, N. Zhang, W. Zhou, and P. Zhang, “Outage [11] R. Zhang and Y.-C. Liang, “Exploiting multi-antennas for
performance of relay-assisted cognitive-radio system under spectrum- opportunistic spectrum sharing in cognitive radio networks,” IEEE J.
sharing constraints,” Electron. Lett., vol. 46, no. 2, pp. 182-183, Jan. Sel. Topics Signal Process., vol. 2, no. 1, pp. 88-102, Feb. 2008.
2010. [12] J. Zhou and J. Thompson, “Linear precoding for the downlink of
[6] L. Luo, P. Zhang, G. Zhang, and J. Qin, “Outage performance for multiple input single output coexisting wireless systems,” IET
cognitive relay networks with underlay spectrum sharing,” IEEE Commun., vol. 2, no. 6, pp. 742-752, July 2008.
Commun. Lett., vol. 15, no. 7, pp. 710-712, July 2011. [13] E. G. Larsson and E. A. Jorswieck, “Competition versus collaboration
[7] Z. Yan, X. Zhang, and W. Wang, “Exact outage performance of on the MISO interference channel,” IEEE J. Sel. Areas Commun., vol.
cognitive relay networks with maximum transmit power limits,” 26, no. 7, pp. 1059-1069, Sept. 2008.
IEEE Commun. Lett., vol. 15, no. 12, pp. 1317-1319, Dec. 2011. [14] K. Hamdi, W. Zhang, and K. B. Letaief, “Low-complexity antenna
[8] V. Asghari and S. Aissa, “Performance of cooperative spectrum- selection and user scheduling in cognitive MIMO broadcast systems,”
sharing systems with amplify-and-forward relaying,” IEEE Trans. in Proc. IEEE Int. Conf. on Commun., Beijing, China, pp. 4038-4042,
Wireless Commun., vol. 11, no. 4, pp. 1295-1300, Apr. 2012. May 2008.
[9] T. Q. Duong, D. B. da Costa, M. Elkashlan, and V. N. Q. Bao, [15] J. Zhou, Y. Li, and B. Evans, “Antenna selection for multiple-input
“Cognitive amplify-and-forward relay networks over Nakagami-m and single-output cognitive radio systems,” IET Commun., vol. 6, no.
fading,” IEEE Trans. Veh. Tech., vol. 61, no. 5, pp. 2368-2374, June 8, pp. 917–930, May 2012.
2012. [16] I. S. Gradshteyn and I. M. Ryzhik, Tables of Integrals, Series, and
[10] I. E. Telatar, “Capacity of multi-antenna Gaussian channels,” Eur. Products, 7th ed., Academic Press, Burlington, MA, 2007.
Trans. Telecommun., vol. 10, no. 6, pp. 585–595, Nov.-Dec. 1999.

14

You might also like