Sei sulla pagina 1di 6

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323966801

Secrecy Outage Probability in Cognitive Radio Networks Subject to Rayleigh


Fading Channels

Conference Paper · April 2018


DOI: 10.1109/COMMNET.2018.8360281

CITATIONS READS
11 209

3 authors:

Mounia Bouabdellah Faissal El Bouanani


Ecole Nationale Supérieure d'Informatique et d'Analyse des Systèmes Mohammed V University of Rabat
9 PUBLICATIONS   44 CITATIONS    64 PUBLICATIONS   210 CITATIONS   

SEE PROFILE SEE PROFILE

Hussain Ben-azza
ensam-Meknès. UMI Meknès
55 PUBLICATIONS   130 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Genetic algorithms for Linear Block codes View project

Performance analysis View project

All content following this page was uploaded by Mounia Bouabdellah on 23 March 2018.

The user has requested enhancement of the downloaded file.


Secrecy Outage Probability in Cognitive Radio
Networks Subject to Rayleigh Fading Channels
Mounia Bouabdellah 1 , Faissal El Bouanani 1 , Hussain Ben-azza 2
1
ENSIAS, Mohammed V University in Rabat, Morocco
2
ENSAM, Moulay Ismail University in Meknes, Morocco
Emails: mounia_bouabdellah@um5.ac.ma, f.elbouanani@um5s.net.ma, hbenazza@yahoo.com

Abstract—Wireless communication systems are vulnerable to cooperative transmission in which the data is transmitted from
eavesdropping attack in which the attacker overhear the trans- a source node to a destination through relays.
mitted data in the network. Hence, the physical-layer security In the direct transmission, the source node sends directly the
is of utmost importance to protect the wireless communica-
tions against the eavesdropping attack. In this paper, we study confidential messages to the destination without relying on
the physical-layer security of a two-hops cognitive radio-based any intermediate nodes. Hence, this type of communication
communication system where the secondary user sends some system assumes that the transmitter, the receiver, and the
confidential information to a destination through a relay. This eavesdropper are in the same transmission range. The secrecy
relay is assumed to be equipped with multiple antennas for outage probability (SOP) for CR communication system over
reception, applies the maximum-ratio combining technique for
the received signals, and uses only one antenna to forward Nakagami-m fading channels were derived in [4]-[6]. In [5],
the combined signal to the destination. The transmission is the authors assumed that the source uses one antenna for
performed under the eavesdropper’s attempt to listen to the transmission and both the destination and the eavesdropper
communication channel. The secrecy outage probability and the use one antenna for reception. In addition to the SOP, the
intercept probability are both derived based on the statistical authors derived also the probability of non- zero secrecy
characteristics of the channels and under the constraint of
avoiding the communication outage of the primary user. All capacity (PNSC). A communication system consisting of an
results have been validated using Monte Carlo method. SU source node, a legitimate SU receiver, and an eavesdropper
Index Terms—Cognitive radio networks, physical layer secu- that are equipped all with multiple antennas is studied in [4].
rity, secrecy outage probability, intercept probability, maximum The generalized selection combining scheme has been used
ratio combining. to combine the multiple copies of signals. In [6], both the
receiver and the eavesdropper are supposed to be equipped
I. I NTRODUCTION with multiple antennas, while the transmitter uses a single
antenna. The SOP of a system consisting of a transmitter, re-
With the proliferation of mobile devices, the demand for ceiver, and eavesdropper equipped with a single antenna under
wireless radio has increased which leads to a spectrum scarcity Rayleigh and log-normal fading channels has been investigated
problem. As the static spectrum allocation technique is unable in [8]. In [7], the studied CR communication system consists
to solve this issue, cognitive radio (CR) has been proposed to of primary and secondary transmitters that are equipped with
allow a dynamic spectrum access and increase the efficiency single antenna whereas the primary and secondary receivers
of spectrum utilization. In cognitive radio networks (CRN), use multiple antennas. The communication is performed in
unlicensed users, also called secondary users (SUs), share the the presence of two eavesdroppers, one is overhearing the
licensed spectrum bands with primary users (PUs) [1], [2]. communication of the primary network whereas the second
Indeed, the SUs sense the spectrum in order to determine the is listening to the communication of the secondary network.
availability of unused bands and opportunistically use them The SOP, in this case, has been derived under the Rayleigh
without causing any harmful interference to the PUs signals. fading channels.
Like any traditional wireless network, CRN can be vulner- However, direct transmission is not always practical in wire-
able to several attacks that can disrupt their operation [3]. less networks. Hence, the cooperative communication seems
Eavesdropping attack is one of the security threats that can to be an efficient solution when no direct communication link
occur at the physical layer. Therein, unauthorized users try to exists between the source and the destination [10]. The SOP of
listen to the communication between legitimate users. Thus, two-hops CR cooperative communication system was derived
under the constraint of eavesdropping risks as well as the under the Rayleigh fading channels in [9] by taking into
potential interference to PU communication, the SUs have to consideration the self-interference at the relays. Such system
adjust their transmission powers in order to avoid these risks. consists of a single SU source that communicates with a
The physical layer security in the context of CRN has been unique destination through multiple SU relays in the presence
studied in [4]-[9]. Most of these research papers focused on of PUs and one eavesdropper that is assumed to be listening
the direct transmission between the SU source and the SU to both communication hops.
destination. However, few papers considered the case of a In this paper, we investigate the SOP as well as the intercept
probability over Rayleigh fading channels for two-hops CR where PR is the transmission power of R, xr is the transmitted
communication system in which the data transmitted by an signal of R after performing the relaying technique, and nD
SU source node is forwarded to the destination through an SU is the AWGN of mean zero and variance N .
relay under the malicious attempt of a single eavesdropper. In The received signals at the eavesdropper from the source and
this scheme, we assume that the relay performs the maximum- from the relay are, respectively, written as
ratio combining (MRC) technique at the reception and uses p
one antenna to forward the message to the final destination. To y1E = PS hSE xs + nE , (3)
the best of our knowledge, none of the existing work studied
the case of a CR-based communication system with a relay p
node equipped with multiple antennas. y2E = PR hRE xr + nE , (4)
The remainder of this paper is organized as follows. In where nE is the additive noise assumed to be AWGN of
Section II we derive the secrecy outage probability and the variance NE and mean zero.
intercept probability of the studied system. In Section III, In what follows, we are interested in deriving the secrecy
we discuss the numerical and simulation results. Finally, in capacity, the secrecy outage probability, and the intercept
Section IV, we give a brief conclusion. probability of the system under the constraint of avoiding, as
II. M ETHODOLOGY possible, the interference at the PU receiver.
In this paper, we are considering the CR-based system
represented in Fig.1. It consists of a two-hops communication
system in which one SU source node S is transmitting data Data link
to an SU destination node D through an SU relay R under Interference link E
the eavesdropping attempt of E. In the first transmission Eavesdropping link
hop, the relay receives the signal on its n branch MRC
combiner. In the second transmission hop, the relay R uses
only one antenna to forward the data to the destination. During 1
the data transmission, the PU PRx can be subjected to the
interference signals coming from both S and R. Hence, the S 2
. R ℎ𝑅𝐷
data transmitted from S to D through R should not cause . D
any harmful interference at the PU receiver. Thus, the node
S and the relay R have to adjust their transmission powers in n
order to satisfy the PUs’ quality of service. The expressions
of transmission powers PS and PR of both nodes S and R
are calculated in [9] Eq. (10) and (11), respectively.
Without loss of generality, we assume that all links’ fading 𝑷𝑹𝒙
amplitudes are Rayleigh distributed, i.e the channel coeffi- 𝑷𝑻𝒙
cients of links S→Rk , R→D, S→E, R→E, PT x →PRx ,
R→PRx , S→PRx are hSRk , hRD , hSE , hRE , hP ,hRP ,
hSP , respectively. For simplicity, we write the channel power
gains as gSRk =|hSRk |2 , gRD =|hRD |2 , gSE =|hSE |2 ,
gRE =|hRE |2 , gP =|hP |2 , gRP =|hRP |2 , gSP =|hSP |2 . We Fig. 1. The studied CRN system model.
assume that all fading amplitudes are normalized. Thus, all
coefficients λSRk , λRD , λSE , λRE , λP , λRP , λSP of the
A. Secrecy capacity
exponential distribution are equal to one.
The signal at the MRC output of the relay R is expressed The secrecy capacity of the communication system in which
as p the eavesdropper E is trying to listen to both transmission hops
yR = PS ||hSR ||xs + wR nR , (1) can be expressed as:
where PS is the transmission power of S, hSR denotes the CS = min(CS1 , CS2 ), (5)
n × 1 channel vector from the SU source node to the relay
R, xs is the transmitted signal of S, and nR stands for the where CS1 and CS2 are the secrecy capacities at the first
additive white Gaussian noise (AWGN) n × 1 channel vector and second hop, respectively. These secrecy capacities can be
whose entries have variance N and mean zero. The MRC written as
h†
weight vector is given by wR = ||hSRSR ||
, the symbol † denotes
the transpose conjugate. CS1 = log2 (γ1 ) , (6)
The received signal at the destination D is given by
p
yD = PR hRD xr + nD , (2) CS2 = log2 (γ2 ) , (7)
where secrecy capacity CS falls below a predefined security rate Rs .
1 + γR
γ1 = . (8)
1 + γ1E SOP = P r (CS < Rs ) , (17)
and
1 + γD By using equation (10), the SOP can be written as:
γ2 = . (9)
1 + γ2E SOP (γ) = P r (min[γ1 , γ2 ] < γ)
Hence, the secrecy capacity of the system can be written as = 1 − P r γ1 > 2R .P r (γ2 > γ)


Cs = log2 (min[γ1 , γ2 ]) . (10) = 1 − [1 − F γ1 (γ)] [1 − F γ2 (γ)] , (18)


Rs
On the other hand, the combined signal-to-noise-ratio (SNR) where γ = 2 .
at the relay R can be written as From Eq. (18), we see that to obtain the SOP of the
n considered system we need to derive F γ1 (γ) and F γ2 (γ).
By using Eq. (8) the CDF of γ1 can be written as
X
γR = γk,R , (11)
 
k=1 1 + γR
F γ1 (γ) = P r <γ
while the SNR at the destination D is 1 + γ1E
= P r (γR < γ(1 + γ1E ) − 1) , (19)
PR gRD
γD = , (12)
N According to [12], the Eq. (19) can be written as
By substituting the transmission power PP of the PU in the Z ∝
equations (11) and (12), the SNR at the relay R and the F γ1 (γ) = FγR (γ(1 + y) − 1)fy (y)dy, (20)
0
destination D can be, respectively, expressed as:
n
From Eq. (20), we notice that to calculate F γ1 (γ), we have
X PS to calculate first the CDF of γR which can be computed
γR = αgSRk , (13)
PP according to [12] as follows
k=1
1
PR  n
X −γ x
γD = αgRD , (14) 1 1−e j,R

PP FγR (x) = Qn  .
i=1 γ i,R 1 Qn 1 1
PP j=1
k=1,k6=j −
where α = . We refer the reader to [9] for the detailed γ j,R γ j,R γ k,R
N (21)
expressions of PS and PR .
Hence, using Eq. (21) the CDF of γ1 is derived in Eq. (24),
Finally, the SNRs at the eavesdropper from the source S and
as it is shown in the bottom of the page.
the relay R are γ1E and γ2E , respectively and are expressed
The CDF of γ2 is expressed as follows
as follows  
1 + γD
PS gSE F γ2 (γ) = P r <γ
γ1E = . (15) 1 + γ2E
NE Z ∝
PR gRE = FγD (γ(1 + z) − 1) fz (z)dz, (22)
γ1E = . (16) 0
NE
where z = γ2E
B. Secrecy outage probability 1
γD − (γ−1)
The SOP is an important metric that is used to evaluate the F γ2 (γ) = 1 − e γD . (23)
γ D + γγ 2E
security performance and it consists of the probability that the

   
1 γ−1 1 1
n ∝ − y − + y−
γ
 X Z
1 e γ 1E − e γ j,R γ 1E γ j,R 
F γ1 (γ) = Qn  j,R   dy
γ 1E i=1 γ i,R j=1 Qn 1 1 0
 
k=1,k6=j −
γ k,R γ j,R
 γ − 1

γ j,R
γ j,R 1 − e γ j,R 
 
 Xn γ j,R + γ 1E γ
1
= Qn   . (24)
i=1 γ i,R
Qn 1 1
j=1
k=1,k6=j −
γ k,R γ j,R
C. Intercept probability
1
According to [11], when the channel capacity of the main
link is less than that of the wiretap link, then the eavesdropper, 0.9
Simulation
in this case, most likely succeeds in decoding and intercepting Analytic
0.8 n=4
the source message. Thus, an intercept event is considered
to be occurring. The intercept probability consists of the 0.7 n=2
probability when the secrecy capacity becomes non-positive

SOP
and can be expressed as 0.6

Pint = P r(CS < 0), (25) 0.5

By substituting the value of CS given in Eq. (8) the intercept 0.4


probability can be written as
0.3
Pint = P r(log2 (min[γ1 , γ2 ]) < 0)
0.2
= 1 − P r (γ1 > 1) .P r (γ2 > 1) 0 0.5 1 1.5 2 2.5 3
Rate(bit/s)
= 1 − [1 − F γ1 (1)] [1 − F γ2 (1)] , (26)
Fig. 2. Secrecy outage probability versus secrecy rate for γP = 15 dB and
III. R ESULTS AND DISCUSSION  = 10−4
In this section, we present the analytical and simulation
results for the considered communication system. We study
the impact of the transmission power of the PU transmitter in 1
PP Simulation
terms of α = , the secrecy rate Rs , and the number of the 0.9 Analytic
N
relay’s antennas on the secrecy outage probability for the SU
0.8
transmission. The simulation has been performed by using the
Monte Carlo method. n=2
0.7 n=4
Fig.2 shows the secrecy outage probability as a function of
SOP

0.6
the secrecy rate Rs for different values of the number of the
relay’s antennas. As one can see, the SOP increases with the 0.5
increasing values of Rs . In other words, when a higher secrecy
rate Rs is used by SUs for better throughput performance, it 0.4
is less likely to achieve the perfect secure transmission against
0.3
eavesdropping attacks. In addition and as expected, the SOP
is less when using multiple antennas at the relay instead of 0.2
0 10 20 30 40 50
a single antenna. For instance, for Rs = 1 bits/s, the SOP is
α(dB)
almost equal to 0.78 when using only two antenna (n = 2)
and 0.82 when using 4 antennas. Fig. 3. Secrecy outage probability versus transmission power of the PU for
Fig. 3 illustrates the SOP as a function of α for different Rs = 1 bit/s and  = 10−4
numbers of the relay antennas. As it is observed, this proba-
bility decreases with the increasing values of α. Moreover, it
is clear that using multiple antennas at the relay improves the secrecy of the system as the intercept probability tends to be
secrecy of the system. For instance for α = 40 dB, the SOP smaller than the case of using a relay with a single antenna.
is almost equal to 0.57 for n = 2 whereas for n = 4 the SOP
is almost equal to 0.62. IV. C ONCLUSION
Fig. 4 depicts the SOP as a function of  for different
numbers of the relay’s antennas. As it is observed, the SOP In this paper, we studied the physical layer security in CRN
has high values for the small values of . According to [9],  where the data transmission from an SU source to an SU desti-
is the desired outage probability for the link PT X → PRX . In nation through an SU relay is under the risk of eavesdropping
other words when  is very small the transmission powers of attack and under the constraint of the interference to the PU’s
the source S and relay R tend to be small which affect the transmission. In the studied CR-based communication system,
secrecy performance of the system. we considered the case of an SU relay with multiple-input
Fig. 5 shows the intercept probability as a function of α antennas for reception and single antenna for transmission.
for different numbers of the relay’s antennas. As one can see, We analyzed the secrecy outage probability of the considered
the intercept probability decreases with increasing values of α system under different circumferences. The results showed that
and it becomes null once the PU’s transmission power is high. the system has better secrecy performance when using a relay
Furthermore, using multiple antennas at the relay enhance the with multiple antennas instead of using only one antenna.
[7] H. Tran, G. Kaddoum, F. Gagnon, "Cognitive radio network with secrecy
and interference constraints," Physical Communication, 2017, vol. 22,
Simulation p. 32-41.
0.8 Analytic [8] H. Liu, H. Zhao, C. Tang, G. Pan, T. Li, "Physical-layer secrecy outage
of spectrum sharing CR systems over fading channels," Science China
Information Sciences, 2016, vol. 59, no 10, p. 102308.
[9] M. bin Azaman, N. Nguyen, D. B. Ha, T. V Truong, "Secrecy outage
0.7 probability of full-duplex networks with cognitive radio environment and
partial relay selection," Recent Advances in Signal Processing, Telecom-
SOP

munications & Computing (SigTelCom), International Conference on.


0.6 n=2
IEEE, 2017. p. 119-123.
[10] M. Bouabdellah, F. El Bouanani, H. Ben-Azza, "A secure cooperative
n=4 transmission model in VANET using attribute based encryption," Ad-
vanced Communication Systems and Information Security (ACOSIS),
0.5 International Conference on. IEEE, 2016. p. 1-6.
[11] Y. Zou, G. Wang, "Intercept behavior analysis of industrial wireless
sensor networks in the presence of eavesdropping attack," IEEE Trans-
actions on Industrial Informatics, 2016, vol. 12, no 2, p. 780-787.
0.4 [12] ROSS, Sheldon M. Introduction to probability models. Academic press,
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 2014.
ǫ

Fig. 4. Secrecy outage probability versus  for Rs = 1 bit/s and α = 15 dB

1
Simulation
Analytic
0.8 n=2
Intercept probability

0.6

n=4

0.4

0.2

0
0 10 20 30 40 50
α(dB)

Fig. 5. Intercept probability versus transmission power of the PU for  =


10−4

R EFERENCES
[1] I. Akyildiz, W. Lee, M. Vuran, S. Mohanty, "NeXt generation/dynamic
spectrum access/cognitive radio wireless networks: A survey," Computer
networks, 2006, vol. 50, no 13, p. 2127-2159.
[2] Y. Saleem and M. Rehmani, "Primary radio user activity models for
cognitive radio networks: A survey," Journal of Network and Computer
Applications, 2014, vol. 43, p. 1-16.
[3] M. Bouabdellah, N. Kaabouch, F. El Bouanani, H. Ben-Azza, "Network
layer attacks and countermeasures in cognitive radio networks: A
survey," Journal of Information Security and Applications, 2018, vol.
38, p. 40-49.
[4] H. Lei, C. Gao, I. Ansari, Y. Guo, Y. Zou, G. Pan, K. A. Qaraqe,
"Secrecy Outage Performance of Transmit Antenna Selection for MIMO
Underlay Cognitive Radio Systems Over Nakagami- Channels," IEEE
Transactions on Vehicular Technology, 2017, vol. 66, no 3, p. 2237-
2250.
[5] C. Tang, G. Pan, T. Li, "Secrecy Outage Analysis of Underlay Cog-
nitive Radio Unit Over Nakagami- Fading Channels," IEEE Wireless
Communications Letters, 2014, vol. 3, no 6, p. 609-612.
[6] N. Nguyen, T. Thanh, T. Duong, "Secure communications in cognitive
underlay networks over Nakagami-m channel," Physical Communica-
tion, 2016.

View publication stats

Potrebbero piacerti anche