IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO.
5, MAY 2020
Energy-Aware Relay Selection Improves
Security-Reliability Tradeoff in Energy
Harvesting Cooperative Cognitive Radio Systems
Peishun Yan , Yulong Zou , Senior Member, IEEE, Xiaojin Ding , and Jia Zhu
Abstract—This article investigates the physical layer security for
a cooperative cognitive radio (CCR) system with energy-harvesting
(EH) technique, which consists of a cognitive source (CS), multiple
cognitive relays (CRs) and a cognitive destination (CD) as well as
multiple eavesdroppers (Es) who are considered to tap the confi-
dential transmission from CS via CRs to CD. Both CS and CRs
are equipped with energy-harvesters to harvest energy from the
surrounding radio frequency environment. Additionally, in order
to guarantee the quality-of-service of primary communications,
the transmit powers of CS and CRs are limited by the maximum
tolerant power at primary destination. Depending on the avail-
ability of channel state information (CSI) of wireless energy links
from primary transmitter (PT) to CRs, we respectively propose
non-energy aware relay selection (NEARS) and energy aware relay
selection (EARS) schemes for protecting the transmission from CS
via CRs to CD from leaking to Es. We consider the use of the
security-reliability tradeoff (SRT) to measure the performance of
NEARS and EARS schemes. For the purpose of comparison, we
also carry out the analysis of SRT performance for conventional
direct communication (DC). Furthermore, the closed-form expres-
sions of outage probability and intercept probability for NEARS
and EARS schemes are derived and numerical results show that the
SRT performance of EARS is better than that of DC and NEARS
schemes.
Index Terms—Physical layer security, security-reliability
tradeoff, energy harvest, relay selection, cognitive radio.
I. INTRODUCTION
A. Background
ITH the explosive growth of mobile communications,
W the wireless devices such as smart phones and unmanned
aerial vehicles (UAVs), have penetrated into many aspects of
our life, gradually [1]. As we all know, these wireless terminals
Manuscript received August 10, 2019; revised November 4, 2019, December
25, 2019, and February 12, 2020; accepted March 2, 2020. Date of publication
March 9, 2020; date of current version May 14, 2020. This work was supported
in part by the National Natural Science Foundation of China under Grants
61671253, 61631020, and 91738201 and in part by the Natural Science Foun-
dation of Jiangsu Province under Grant BK20171446. The review of this article
was coordinated by Prof. D. B. da Costa. (Corresponding authors: Peishun Yan;
Yulong Zou; Xiaojin Ding; and Jia Zhu.)
Peishun Yan, Yulong Zou, and Jia Zhu are with the School of Telecommuni-
cations and Information Engineering, Nanjing University of Posts and Telecom-
munications, Nanjing 210003, China (e-mail: 2019010215@njupt.edu.cn;
yulong.zou@njupt.edu.cn; jiazhu@njupt.edu.cn).
Xiaojin Ding is with the Jiangsu Engineering Research Center of Communi-
cation and Network Technology, Nanjing University of Posts and Telecommu-
nications, Nanjing 210003, China (e-mail: dxj@njupt.edu.cn).
Digital Object Identifier 10.1109/TVT.2020.2979267
0018-9545 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
5115
are powered by the battery with finite capacity. Consequently,
the development and deployment of mobile networks have
been seriously subject to unsustainable power supply [2]. In
recent years, energy-harvesting (EH) technique [2]–[4] has been
attracted much attention and considered as a fundamentally
promising technology to improve the energy efficiency of wire-
less devices. The relaying protocols based on EH are divided
into time-switching relaying (TSR) protocol and power-splitting
relaying (PSR) protocol [5]–[7]. More specifically, the relay first
harvests energy from the ambient surroundings within given
time and then the harvested energy is used for forwarding
source information in TSR protocol. While in PSR protocol,
the received power is split into two parts, which are used for
information transmission and energy storage. However, the en-
ergy efficiency have been enhanced aided with EH technique, the
private transmission still faces the serious threaten from vicious
attackers due to the broadcast nature of wireless mediums.
B. Related Works
Physical layer security [8] and [9] has been seen as a promis-
ing method to prevent the private information from leaking to
the eavesdroppers by making use of physical characteristics of
wireless links. In [10], Wyner first proposed notion of the secrecy
capacity which is the difference between the capacity of main
channel (from legitimate source to legitimate destination) and
that of wiretap channel (from legitimate source to eavesdropper).
Whilst, [10] certified that the intercept event will happen if the
secrecy capacity is less than zero. A large number of existing
works have paid attention to increasing the secrecy capacity
such as cooperative relay selection [11]–[13], multi-input and
multi-output (MIMO) [14]–[16] and multiuser scheduling [17]
and [18].
In order to improve physical layer security, [12] proposed
optimal single relay selection and two-step single relay selec-
tion as well as optimal dual relay selection to decrease the
secrecy outage probability in a non-orthogonal multiple access
system under Nakagami-m channels. While [13] theoretically
analyzed the physical layer security in cooperative cognitive
radio systems. In [14], MIMO was employed to combat against
the eavesdropping attack in a non-cognitive radio system, where
optimal antenna selection (OAS) and suboptimal antenna selec-
tion (SAS) were proposed to decrease the intercept probability.
[15] extended the results of [14] into cognitive radio systems
 5116
over Nakagami-m fading channels. Moreover, [17] examined
the physical layer security in a cognitive radio system which
is composed of one common cognitive base station and mul-
tiple cognitive users while an eavesdropper may overhear the
legitimate transmission. The authors in [18] proposed random
jammer selection aided multiuser scheduling and optimal jam-
mer selection aided multiuser scheduling to enhance the secrecy
outage performance. Moreover, the numerical results indicated
that multiuser scheduling aided with friendly jammer selection
outperforms the conventional multiuser scheduling without jam-
mer selection in terms of secrecy outage performance.
Recently, EH combined with physical layer security has
became a hot topic [19]–[24]. In [20], the authors studied a
PSR system in the view of secrecy transmission. The system
outage probability for the linear and nonlinear EH models
were derived. Additionally, the analysis of effect of power
split ratio and different relay‘s location as well as different
eavesdropper’s location on secrecy outage probability were
carried out. Differing from [20], the EH untrusted relay was
employed to assist the confidential transmission between
source and destination for PSR and TSR protocols in [21].
For the purpose of improving the secrecy transmission, [22]
and [23] examined the secrecy outage performance jointly
aided with relay and jammer selection. The authors in [23]
employed best-relay-best-jammer (BRBJ) selection to enhance
the secrecy performance for cooperative communications. The
results indicated that the secrecy outage performance of BRBJ
is superior to BRRJ [22] and RRBJ [22] schemes. The secrecy
outage performance for dual-hop mixed radio frequency-free
space optical downlink simultaneous wireless information and
power transfer system were analyzed in [24]. While [25]–[27]
exploited the interference alignment to fight against attacks by
eavesdroppers in wireless power transfer systems.
As discussed in [19]–[27], EH non-cognitive radio systems
were exploited. While [28]–[32] examined confidential trans-
mission in EH cognitive radio (CR) systems. For example, Lei
et al. investigated the secrecy outage performance for MIMO-
CR systems with EH technology [28]. Depending on the channel
state information of wiretap channels is available or unavailable
for cognitive users, optimal antenna selection and suboptimal
antenna selection schemes were separately employed to improve
the cognitive secrecy transmission. In [29], the energy-aware
multiuser scheduling was investigated to improve the cognitive
secrecy transmission from cognitive users to cognitive common
base station. In [30], the optimal relay selection (ORS) with EH
was proposed to ameliorate the wireless physical layer security.
It was shown that the ORS with EH achieves better perfor-
mance than ORS without EH in terms of the tradeoff between
secrecy outage probability of primary transmission and ergodic
rate of cognitive transmission. Additionally, [32] studied the
non-convex problem for maximizing the secrecy rate of primary
communication. To elaborate, the primary network took use of
two-way mechanism and the optimal cognitive user forwarded
the primary signals with harvested energy from ambient radio
frequency signals. Furthermore, this problem was separated
into three tractable problems because of the non-convexity and
the two-tiered iterative algorithm was obtained to optimize the
secrecy performance for primary users.
Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO. 5, MAY 2020
C. Motivation and Contributions
To the best of our knowledge, less existing works were devoted
to enhancing the physical layer security in cooperative cognitive
radio (CCR) systems combining with energy-harvesting (EH)
technique. In this context, we explore the secrecy performance
for CCR-EH systems in terms of security-reliability tradeoff
(SRT), where a cognitive source (CS) communicates with cogni-
tive destination (CD) assisted by multiple cognitive relays (CRs).
Meanwhile, multiple eavesdroppers (Es) may tap the legitimate
transmission over wiretap channels. The main contributions are
descried as follows.
r We present energy-aware relay selection (EARS) and non-
energy aware relay selection (NEARS) schemes to protect
cognitive transmission between CS and CD from overhear-
ing by multiple Es in a CCR-EH system. The difference
between EARS and NEARS is whether or not dependent
on channel state information (CSI) of energy channels
from PT to CRs. More specifically, the CR maximizing
the capacity of main channels spanning from CRs to CD
will be selected for forwarding source signals in EARS
scheme. Contrarily, the NEARS scheme only relies on the
CSI of CRs-CD.
r We provide closed-form expressions of outage probability
(OP) and intercept probability (IP) for EARS and NEARS
schemes. For comparison, the analysis of classical direct
communication (DC) is carried out in terms of OP and IP.
It is shown that our proposed EARS scheme achieves the
best performance among EARS, NEARS and DC in the
perspective view of SRT.
r It needs to be pointed that we analyze physical layer
security in a CCR-EH system, while [4] only considered
the outage performance for EH cognitive communica-
tion. Differing from [19] and [20] which only considered
physical layer security in non-cognitive radio systems, we
examine the secrecy transmission in cognitive radio sys-
tems. Additionally, the authors in [21]–[23] and [28] only
studied the secure transmission with EH technique, while
the security and reliability of wireless communications are
mutually constrained. As a result, this paper discusses the
SRT performance to reveal constraint mechanism between
reliability and security.
The rest of this paper is organized as follows. We provide
the system model for our CCR-EH system and the criteria of
relay selection in Section II. Section III presents the closed-
form expressions of OP and IP for our proposed EARS scheme
and NEARS scheme as well as DC and the SRT performance
is analyzed over Rayleigh fading channels. In Section IV, the
numerical results are shown to validate our theoretical results
of EARS, NEARS and DC schemes. Finally, the concluding
remarks are drawn in Section V.
II. SYSTEM MODEL AND PROBLEM FORMULATION
A. System Model
As shown in Fig. 1, we present DC as a bench-marking scheme
where CS transmits confidential information to CD directly. Due
to the broadcast nature of wireless mediums, Es can tap the
 YAN et al.: ENERGY-AWARE RELAY SELECTION IMPROVES SECURITY-RELIABILITY TRADEOFF 5117

TABLE I
NOTATIONS AND REPRESENTATIONS

Fig. 1. A Cognitive Radio System with Energy Harvesting technique which

consists of a CS, a CD in the face of multiple Es.

private transmission from CS via CRs to CD. We consider a

Rayleigh fading model for characterizing all the channels of
Fig. 1. We assume that hsd , hse , hps , hsp , hJ1 and hJ2 are
the channel gains of CS-CD, CS-Ee (e = 1, 2, . . . , Nv ), PT-CS,
CS-PD, PT-CD and PT-Ee respectively. Hence |hsd |2 , |hse |2 ,
|hps |2 , |hsp |2 , |hJ1 |2 and |hJ2 |2 are subject to independently
2 2 2
exponential distribution with respective means of σsd , σse , σps ,
2 2 2
σsp , σJ1 and σJ2 . Furthermore, this paper considers the case that
Es adopt maximum ratio combining [28] and [29] to overhear where 0 ≤ η ≤ 1 and Pp denote the efficiency of energy con-
the confidential information, so even if an E fails to tap the version and the transmit power of PT, respectively. Then, the
legitimate information, the cognitive secrecy transmission still harvested energy will be utilized to transmit information by CS
be threatened by eavesdropping attack. We assume that all nodes at the remaining (1 − α)T . Hence, the transmit power of CS is
in this paper equip single antenna [33]. Following our previous described as
works [29], [34] and other literatures [35], [36] (We have carried  
2 I
out the SRT performance for SIMO [29] and the similar results Ps = min aPp |hps | ,
dc
, (2)
can be got as SISO [34]. Additionally, the secrecy performance |hsp |2
in SIMO [35] was extended into MIMO [36] and the similar where a = 1−ααη
. Upon the theory of Shannon capacity, the
performance can be got in terms of secrecy outage probability), capacity of main channel from CS to CD is written as
the similar SRT results of SISO can be got in MIMO. Moreover,  
to ensure the quality-of-service of primary transmission, the Psdc |hsd |2
Cd = (1 − α)T log2 1 + , (3)
transmit power of CS shall be limited by the tolerable inter- NJ1
ference temperature I at PD. The interference at CD and E from
PT can be respectively represented as Pp |hJ1 |2 and Pp |hJ2 |2 . where NJ1 = N0 + Pp σJ1 2
. Meanwhile, the confidential infor-
For computational simplicity, we consider that the average in- mation will be intercepted by Ee results from the broadcast and
terference at CD and an E from PT in the transmitting phase. openness nature of wireless communications. Thus, the capacity
This can be explained that CD merely pays attention to the signal of wireless channel from CS-Ee is obtained as
 
transmitting channel without focusing on the interfered channel Psdc |hse |2
from PT in the transmission phase. Additionally, Fig. 3 in Cse = (1 − α)T log2 1 + , (4)
NJ2
Section IV shows that the performance of average interference
from PT is compared with that of instantaneous interference where NJ2 = N0 + Pp σJ2 2
. Here, we consider that Es adopt
from PT and the performance gap between them does exist. maximal ratio combining technique to handle the information
However, the performance gap has no significant effect on our from CS. As a result, the overall capacity of wiretap channels is
analysis for outage probability and intercept probability. Finally, given by
the thermal noise received at CD and Ee is assumed to be AWGN   v 
with zero mean and variance N0 . Psdc N e=1 |hse |
2
Ce = (1 − α)T log2 1 + . (5)
NJ2
B. Direct Communication
The whole communication process is divided into two time C. Non-Energy Aware Relay Selection
slots which are called energy harvesting phase and information Fig. 2 shows that CS transmits confidential information to CD
transmitting phase. To be specific, CS first harvests energy from via Nr CRs. Moreover, Es are beyond the coverage of CS due to
PT in αT , where 0 ≤ α ≤ 1 is time allocated ratio and T is the deep fading and obstacle blocking effects [8] and [13]. Hence,
time needed to the whole communication. Therefore, the energy Nv Es who are distributed around CD can overhear the legitimate
harvested at CS in the first phase is given by transmission from CRs. We assume that hsi , hid , hie , hpi , and
hip are the channel gains of CS-CRi (i = 1, 2, . . . , Nr ), CRi -
Esdc = αηT Pp |hps |2 , (1) CD, CRi -Ee , PT-CRi and CRi -PD, respectively. Hence |hsi |2 ,

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 5118 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO. 5, MAY 2020

Additionally, the overall capacity of wireless channels from CRi

to Es is written as
  v 
(1 − α)T Pi N e=1 |hie |
2
Cie = log2 1 + . (12)
2 NJ2
The research findings in [37] indicated that DF protocol
has a higher probability of successful decoding and flawless
retransmission from a shorter distance than AF protocol if the
relay is closer to the source. Moreover, the theoretical results
illustrated that the secrecy outage capacity of DF performs better
than that of AF at low and medium signal-to-noise ratio (SNR)
regimes [38]. Consequently, we assume that all CRs employ DF
relaying protocol to deal with the source signal. Specifically, if
Fig. 2. A Cooperative Cognitive Radio System with Energy Harvesting tech- CRi fails to decode the received signal, it will not participate
nique, which is made up of a CS, multiple CRs and a CD in the face of multiple in forwarding the source signal. For simplicity, let D to be a
Es. set which is composed of CR who can succeed in decoding the
signal from CS. By using the law of probability, the sample space
of D can be written as
|hid |2 , |hie |2 , |hpi |2 and |hip |2 are subject to independently
exponential distribution with respective means of σsi 2 2
, σid 2
, σie ,  = {φ, D1 , D2 , . . . , D2Nr −1 } , (13)
2 2
σpi and σip . In addition, the average interference received at CRi where φ means the empty set that all CRs decode received signal
is E(Pp |hJi |2 ) = Pp σJi 2
, where hJi is the CSI of interference unsuccessfully and Dn is the n-th null set. According to the law
channel of PT to CRi . The whole transmission includes three of Shannon coding, if the channel capacity is lower than the
phases. In the first phase αT , CS and CRi harvest energy from transmission rate R0 , the receiver is incapable of recovering
radio frequency signals from PT. The energy collected at CS and the source signal. Hence, the case D = φ and D = Dn can be
CRi are obtained as further separately described as
Es = αηT Pp |hps |2 (6) D = φ, {Csi < R0 |i = 1, 2, 3, . . . , Nr } (14)
and and
 
Ei = αηT Pp |hpi |2 . (7) D = Dn , Csi > R0 , Csj < R0 |i ∈ Dn , j ∈ Dn , (15)
Therefore, the transmit powers at CS and CRi are given by where Dn denotes the complementary set of Dn . During the
  last remaining phase (1−α)T , only one relay is chosen forward
I 2
2
Ps = min bPp |hps | , (8) information from CS to CD. The selection criterion of NEARS
|hsp |2 is defined as
and r = arg max |hid |2 , (16)
  i∈Dn
I
Pi = θ min bPp |hpi |2 , , (9) Combining (11) and (16), the capacity of the main channel is
|hip |2 rewritten as
 
where b = 1−α
2αη
and θ is the remaining ratio of total harvested en- (1 − α)T Pr |hrd |2
ergy for transmitting information. Thus, 1 − θ is the consumed Crd = log2 1 + , (17)
2 NJ1
ratio for decoding, demodulating, re-coding and re-modulating
by cognitive relay. The capacity of CS-CRi can be expressed as where hrd means the wireless channel gain of CRr -CD. Mean-
  while, the capacity of wiretap channel is simply formulated as
(1 − α)T Ps |hsi |2   v 
Csi = log2 1 + . (10) (1 − α)T Pr N 2
2 NJi e=1 |hre |
Cre = log2 1 + , (18)
2 NJ2
where NJi = N0 + Pp σJi 2
.
(1−α)T where hre denotes the fading coefficient of CRr -Ee .
In the next phase 2 , the energy harvested at CS in the
αT is employed to transmit cognitive source signal to CRs. In
D. Energy Aware Relay Selection
the last remaining time (1−α)T
2 , CRi forwards the confidential
information from CS to CD. Therefore, the capacity of main link This subsection provides our proposed EARS scheme to
spanning from CRi to CD can be formulated as enhance physical layer security. The CRi having the maximal
  channel capacity of CRi -CD is considered to be an optimal
(1 − α)T Pi |hid |2 relay to recover and forward the confidential information. This
Cid = log2 1 + . (11)
2 NJ1 differs from NEARS which only depends on the CSI of channel

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 YAN et al.: ENERGY-AWARE RELAY SELECTION IMPROVES SECURITY-RELIABILITY TRADEOFF 5119

R0
spanning from CRi to CD. Thus, the criterion of EARS is defined where λ0 = 2 (1−α)T − 1. After some manipulations, Pout
dc
can be
as further rewritten as
   
2 γI1 2
o = arg max min bγp1 |hpi | , |hid | , (19) dc
Pout = Pout,I
dc
+ Pout,II
dc
, (25)
i∈Dn |hip |2
P
where
where γp1 = NJ1p
, γI1 = NIJ1 . Combining (19) and (11), we can  
obtain the channel capacity of CRo -CD 2 2 2 γI1
 
dc
Pout,I = Pr aγp1 |hps | |hsd | < λ0 , aγp1 |hps | <
|hsp |2
(1 − α)T Po |hod |2
Cod = log2 1 + , (20) (26)
2 NJ1 and
 
where Po and hod represent the transmit power of CRo and the γI1 γI1
fading coefficient of CRo -CD. Similarly to (18), the channel dc
Pout,II = Pr |hsd |2 < λ0 , aγp1 |hps |2 > .
capacity for Es can be formulated as |hsp |2 |hsp |2
  (27)
 v
(1 − α)T Po N |h oe | 2
Using the results of Appendix A, the closed-form expressions
Coe = log2 1 + e=1
, (21) dc
of Pout,I dc
and Pout,II can be respectively expressed as
2 NJ2
 
where hoe represents the fading coefficient of channel from CRo ε 1 γ I1 ε 1 γ I1 ε1 λ 0
to Ee . As shown in (16) and (19), both the criteria of NEARS and
dc
Pout,I =1− 2
K1 2
− 2
σsp σsp σsd
EARS schemes are independent on the CSI of wiretap channel
 
due to the passive eavesdrop by Es. In other words, these two ε1 λ 0 √ √
relay selection schemes merely aim to enhance the reliability of × K1 2
+ ε1 ε2 K 1 ( ε1 ε2 ) (28)
σsd
cognitive transmission without being beneficial for the wireless
capacity of wiretap link. and
 
III. SECURITY AND RELIABILITY ANALYSIS OVER ε1 γI1 ε1 γI1 ε1 √ γI1
dc
Pout,II = 2
K1 2
K1 ( ε1 ε2 ),
− 2
RAYLEIGH FADING CHANNELS σsp σsp ε2 σsp
In this section, we carry out the analysis of SRT performance (29)
for DC, NEARS and EARS schemes over Rayleigh fading where Kn (x) represents modified bessel function of order n
dc
channels. As discussed in [39] and [40], the initial message [41]. Therefore, substituting (28) and (29) into (25), Pout can be
with a secrecy rate Rs is encoded by a secrecy encoder, which rewritten as
generates transmission rate R0 . The difference between R0 and  
ε 1 λ 0 ε 1 λ 0 γI1
Rs is extra redundancy to against eavesdropping attacks. We use dc
Pout =1− 2
K1 2
− 2
the OP and IP to respectively evaluate two important metrics: σsd σsd σsp
reliability and security. Exactly, the outage event occurs when ε1 √ √ √
the capacity of the main channel is lower than transmission data × K1 ( ε1 ε2 ) + ε1 ε2 K1 ( ε1 ε2 ). (30)
ε2
rate R0 , and eavesdropping event happens if the wiretap channel
capacity becomes higher than R0 − Rs . Therefore, the OP and Additionally, combining (5) with (23), we can obtain the corre-
IP are separately given by sponding IP of DC scheme as
Pout = Pr(Cd < R0 ) (22) dc
Pint = Pr(Ce > R0 − Rs )
and    Nv

2 γI2 2
Pint = Pr(Ce > R0 − Rs ), (23) = Pr min aγp2 |hps | , |hse | > λs .
|hsp |2 e=1
where Cd and Ce denote the capacities received at destination (31)
and eavesdropper. R0 −Rs
P
where λs = 2 (1−α)T − 1, γp2 = NJ2
p
and γI2 = NIJ2 . After some
A. Direct Communication dc
mathematical manipulations, Pint is reformulated as
As a baseline, this subsection first analyzes the SRT perfor- dc
Pint = Pint,I
dc
+ Pint,II
dc
, (32)
mance of DC scheme. Using (2), (3) and (22), the OP of DC can
be written as wherein,
dc
Pout = Pr(Cd < R0 ) N 
v
2 λs 2 γI2
γI1
dc
Pint,I = Pr |hse | > , |hsp | <
= Pr[min(aγp1 |hps |2 , 2
)|hsd |2 < λ0 ], (24) e=1 aγp2 |hps |2 aγp2 |hps |2
|hsp | (33)

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 5120 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO. 5, MAY 2020

and from CS to CD. This means that Crd = 0. Hence, we can write
N  (38) as
v
λs γI2
dc
Pint,II = Pr |hse |2 > |hsp |2 , |hps |2 > .
γI2 aγp2 |hsp |2
NEARS
Pout = Pr (D = φ) + Pr (Crd < γ0 , D = Dn ) . (39)
e=1
(34) Relying on (14), (15) and (17), (39) can be further simplified as
dc dc
Referring to the results of Appendix B, we get Pint,I and Pint,II
as 
Nr 2Nr−1  
Nv −1
  NEARS
Pout = ψI + (1−ψI ) NEARS
ψII Pout,I , (40)
2Λ1 g 2 1−g Λ1
Pint,I =
dc
2
(Λ1 σps ) 2 K1−g 2 2
i=1 n=1 i∈Dn j∈Dn

g=0
g!σps σps
NEARS
wherein, the expressions of ψI , ψII and Pout,I can be written
Nv−1 as
2Λ1 g Λ1 +Λ2
− 2 1−g
[(Λ1 +Λ2 )σps ] 2 K1−g (2    
2
g!σps 2
σps 2 γIi 2
g=0
ψI = Pr min bγpi |hps | , |hsi | < γ0 , (41)
(35) |hsp |2
   
and γIj
2 2
  g+1   ψII = Pr min bγpj |hps | , |hsj | < γ0 (42)
Nv −1
2Λ3 g Λ4
2
Λ4 |hsp |2
dc
Pint,II = K1+g 2 Λ4 Λ3 + 2 .
g=0
2
g!σsp Λ3 + σ12 σsp and
   
sp
(36)
dc 2 γ I1 2
Thus, we can obtain the closed-form expression of Pint by sub- NEARS
Pout,I = Pr min bγp1 |hpr | , |hrd | < λ1 ,
stituting (35) and (36) into (32), which can be further computed |hrp |2
as (43)
Pp Pp
Nv −1
  where λ1 = γ0
θ , γpi = NJi , γIi = I
NJi , γpj = N0 +Pp σJj
2 and
2Λ1 g 2 1−g Λ1
Pint =
dc
2
(Λ1 σps ) 2 K1−g 2 2
γIj = N0 +PI p σ2 . Carefully observing from (41) and (42) as well
g=0
g!σps σps Jj
as (24), we can readily reformulate (41) and (42), respectively.
Nv −1
  Additionally, by using the law of total probability, we arrive at
2Λ1 g 2 1−g Λ1 + Λ2
− [(Λ1 + Λ2 )σps ] 2 K1−g 2  NEARS 
g=0
2
g!σps 2
σps NEARS
Pout,I = Pout,I_1 + Pout,I_2
NEARS
, (44)
i∈Dn
Nv−1
  g+1  
g −1 2
2Λ3 1 1 NEARS NEARS
+ 2
Λ4 Λ3 + 2 K1+g 2 Λ4 Λ3 + 2
. wherein, Pout,I_1 and Pout,I_2 are separately expressed as (45)
g=0
g!σsp σsp σsp and (46)
(37)
NEARS
Pout,I_1 = Pr bγp1 |hpi |2 |hid |2 < λ1 , bγp1 |hpi |2
B. Non-Energy Aware Relay Selection 
In this subsection, we consider NEARS scheme to against γI1 2 2
< , max |hkd | < |hid | (45)
eavesdropping attacks from the perspective view of physical |hip |2 k∈Dn ,k=i

layer security, where only the CR maximizing |hid |2 is selected

and
to decode and forward source signal in the corresponding time 
slot. The OP of NEARS is given by γI1 |hid |2
NEARS
Pout,I_2 = Pr < λ1 , bγp1 |hpi |2
NEARS
Pout = Pr (Crd < γ0 , D = φ)+Pr (Crd < γ0 , D = Dn ) , |hip |2
(38) 
2R0 γI1 2 2
where γ0 = 2 − 1 and Crd is written as (17). In the case of
(1−α)T > , max |hkd | < |hid | . (46)
D = φ, there is no CR being selected to retransmit information |hip |2 k∈Dn ,k=i

 
   
γI1 Λ7 γI1 Λ7 −2 −2 −2 −2
NEARS
Pout,I_1 =1− Λ 5 λ 1 K1 ( Λ 5 λ1 ) − 2
K1 2
+ Λ7 (γI1 σip + λ1 σid )K1 Λ7 (γI1 σip + λ1 σid )
σip σip

2|Dn |−1 −1      
(−1)|Dn (m)| Λ6
+ 2
1 − 4λ 1 (bγ Λ σ 2 )K ( 4λ (bγ Λ σ 2 ))
p1 6 pi 1 1 p1 6 pi
m=1
σid
|Dn |−1
−1 2|Dn |−1 −1
(−1)|Dn (m)| Λ6   (−1)|Dn (m)| Λ6  
2
− 2
γI1 Λ5 K1 ( γI1 Λ5 ) + 2
Λ 8 K1 ( Λ 8 ) (47)
m=1
σid m=1
σid

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 YAN et al.: ENERGY-AWARE RELAY SELECTION IMPROVES SECURITY-RELIABILITY TRADEOFF 5121

Referring to the results of Appendix C, the final expressions of where Dn (m) is the m-th non-empty subset of Dn . Therefore,
(45) and (46) can be derived as (47) shown at the bottom of we have
the previous page, and (48) shown at the bottom of this page. NEARS
Pint,I_2
Using (14), (15), (18) and (23), the IP of NEARS scheme can
⎡ ⎛ ⎞⎤
be expressed as  ∞ 2|Dn|−1−1
⎣1+ x ⎠⎦
= (−1)|Dn (m)| exp⎝− fX (x)dx
NEARS
Pint = Pr (Cre > γs , D = φ)+Pr (Cre > γth , D = Dn ) , 0 m=1 k∈Dn (m)
2
σkd
(49) ⎛ ⎞−1
2(R0 −Rs )
where γs = 2 (1−α)T − 1. Similarly to (39), if D = φ, all relays 2|Dn |−1−1
(−1)|Dn (m)| ⎝ 1 1 ⎠
fail to recover and forward the source signal, which causes Cre = = 1+ + . (54)
NEARS
σ2id σ2kd σ2id
0. As a result, we can reformulate Pint as m=1 k∈Dn (m)

NEARS NEARS
2Nr −1 Hence, substituting ψI , ψII and Pint,I into (50), Pint can
 
NEARS
Pint = (1 − ψI ) NEARS
ψII Pint,I , (50) be finally obtained.
n=1 i∈Dn j∈Dn
C. Energy Aware Relay Selection
NEARS
wherein, Pint,I can be given by This subsection presents our proposed EARS scheme to
NEARS
improve the cognitive secrecy transmission in terms of SRT
Pint,I performance. Using (14), (15), (20) and (22), the OP of EARS
    is written as
2 γI2 Nv 2
= Pr min bγp2 |hpi | , |hie | > λ2 EARS
Pout = Pr (Cod < γ0 , D = φ)+Pr (Cod < γ0 , D = Dn ) ,
i∈Dn
|hip |2 e=1
   (55)
NEARS
Pint,I_1 which is similar to (38) and can be further obtained as
  2Nr −1
× Pr maxk∈Dn ,k=i |hkd |2 < |hid |2 . (51) 
Nr  
EARS
Pout = ψI + (1 − ψI ) EARS
ψII Pout,I , (56)
  
NEARS i=1 n=1 i∈Dn j∈D n
Pint,I_2
where ψI and ψII are derived as (41) and (42). Additionally,
where λ2 = γθs . It is clearly for us to see that Pint,I_1
NEARS
can be got EARS
Pout,I can be expressed as
by substituting b, σpi , σip , σie and λ2 into (31) by replacing a,
2 2 2
 
2
σps 2
, σsp 2
, σse and λs , respectively. In addition, considering X = Pi |hid |2
Pout,I = Pr max
EARS
< γ0
|hid | to be a random variable followed exponential distribution
2 i∈Dn N0
with mean σid 2 NEARS
. Pint,I_2 can be computed as    
 γI1
 ∞    = Pr min bγp1 |hpi | , 2 2
|hid | < λ1 .
x |hip |2
Pint,I_2 =
NEARS
1 − exp − 2 fX (x)dx, i∈Dn
0 σkd
k∈Dn ,k=i (57)
(52)
where fX (x) is the PDF of X, which can be de- Thus, according to the derivation of ψI , we can readily get (57).
 Moreover, the closed-form expression of Pout EARS
can be readily
scribed as fX (x) = σ12 exp(− σx2 ). Moreover, k∈Dn ,k=i
id id got. We then carry out the analysis of IP for EARS scheme.
[1 − exp(− σx2 )] can be expanded by using the binomial the-
kd Similarly to (50), the IP of EARS scheme can be expressed as
orem
   2Nr −1  
x EARS
Pint = (1 − ψI ) EARS
ψII Pint,I , (58)
1 − exp − 2
σkd n=1 i∈Dn
k∈Dn ,k=i j∈Dn
⎛ ⎞ where
2|Dn |−1 −1  
x ⎠  Nv
=1+ (−1) |Dn (m)|
exp ⎝− , (53) Po e=1 |hoe |
2
2
σkd EARS
Pint,I = Pr > γs . (59)
m=1 k∈Dn (m) NJ2

  2|Dn |−1 −1
γI1 Λ7 γI1 Λ7 −2 γI1 Λ7  (−1)|Dn (m)| Λ6  
NEARS
Pout,I_2 = K1 − σip K1 ( γI1 Λ7 Λ9 ) + γI1 Λ7 K1 ( γI1 Λ7 )
2
σip 2
σip Λ9 m=1
2
σid
|Dn |−1
−1
 
(−1)|Dn (m)| Λ6
2
γI1 Λ7 λ1 Λ 7 γI1 Λ7 λ1 Λ 7
− + K1 + (48)
m=1
2 σ2
σip id
2
σip Λ6 2
σip Λ6

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 5122 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO. 5, MAY 2020

λ2 λ2
Relying on the theory of total probability, we can further write where Δ3 = 2 , Δ4
bγp2 σie
= γI2
2 , Δ5
bγp2 σip
= 2 , Δ6
γI2 σie
= γI2
2 ,
bγp2 σpi
EARS
Pint,I as
Ξ = {(x, y)|x > 0, y > 0} and ζ = {(y, z)|y > 0, z > 0}. Ad-
ditionally, with the help of FXk (xk ), Θ(x, y) and T(y, z) can
EARS
Pint,I = (Pint,I_1
EARS
+ EARS
Pint,I_2 ), (60)
be computed as
i∈Dn
 
−2 −2
EARS
which is composed of Pint,I_1 EARS
and Pint,I_2 and they are respec- Θ(x, y) = 1 − aγp2 Δ1 θ−1 σkd xyK1 aγp2 Δ1 θ−1 σkd
tively reformulated as
   
 −2
Pk |hkd |2 + Δ1 Δ2 K1 ( Δ1 Δ2 ) − γI2 σkp Δ1 Δ−1
2 K1 ( Δ1 Δ2 )
Pint,I_1 = Pr
EARS
max < bγp2 |hpi |2 |hid |2 , (67)
k∈Dn ,k=i NJ2
Nv
 and
2 λ2 2 γI2
|hie | > , |hip | < , γI2 y γI2 y
bγp2 |hpi |2 bγp2 |hpi |2 T(y, z) = 1 − Δ1 θ−1 2
K1 Δ1 θ−1 2
e=1 zσkd zσkd
(61) 
  
−2
and + Δ1 Δ2 K1 ( Δ1 Δ2 ) − γI2 σkp Δ1 Δ−1
2 K1 ( Δ1 Δ2 ).
 (68)
Pk |hkd |2 γI2
EARS
Pint,I_2 = Pr max < |h |2 ,
2 id EARS
It is challenging to get the closed-form expressions of Pint,I_1
k∈Dn ,k=i NJ2 |hip | EARS
 and Pint,I_2 with our existing knowledge. However, the optimal
Nv CR merely relies on the CSI of energy harvesting channels from
λ2
2 γI2
|hie | > |hip |2 , |hpi |2 > . PT to CRs and main channels from CRs to CD. This results in
e=1
γI2 bγp2 |hip |2
that only the outage probability of legitimate transmission sig-
(62) nificantly decreases without enhancing the secrecy transmission
obviously.
 v As 2mentioned above, the PDFs of random
Nvariables
In (61) and (62), Pk
= θ min(bγp2 |hpk |2 , |hγI2|2 ). For nota-
NJ2 kp of N e=1 |hoe | , |hop | and |hpo | are the same as
2 2 v 2
e=1 |hie | ,
tional convenience, let Xk and Yi respectively denote NPJ2
k
|hkd |2 |hip |2 and |hpi |2 . Hence, formula (60) can be reformulated as
N v
and e=1 |hie |2 . And the CDF of Xk and Yi can be respectively Nv −1 
computed [28] as 2Δm 2 1−m −2
EARS
Pint = 3
2
(Δ3 σpi ) 2 K1−m 2 Δ3 σpi
  m=0
m!σpi
−2 −2
FXk (xk ) = 1 − Δ1 θ−1 σkd x k K1 Δ1 θ−1 σkd xk Nv−1
 
2Δm −2 2 1−m Δ3 +Δ7 σ−2
  − 3
[(Δ3 +Δ7 σie )σpi ] 2 K1−m 2 ie
+ Δ1 Δ2 K1 ( Δ1 Δ2 ) m=0
m!σ2pi σ2pi
   
Nv−1
−2
− γI2 σkp Δ1 Δ−1 2Δm (Δ5 +σ−2ip )Δ7
2 K1 ( Δ1 Δ2 ) (63)
+ 5 −2 2 1+m
[(Δ5 +σip )Δ7 σpi ] 2 K1+m 2 ,
m=0
m!σ2ip σ2pi
and
(69)
Nv −1 g
yi 1 yi
FYi (yi ) = 1 − exp − 2 2
, (64) where Δ7 = γI2
aγp2 .
σie g=0
g! σie
IV. NUMERICAL RESULTS AND DISCUSSIONS
where Δ1 = bγp24σ2 and Δ2 = σx2k + σγ2I2 . Aided with (63) and
pk kd kp
(64), we then separately rewrite (61) and (62) as This section shows the numerical results and discussions
about the analysis of DC, NEARS and EARS schemes. The
 Nv −1
Δ3 Δg3 Δ4 correctness of theoretical OPs and IPs for DC, NEARS and our
EARS
Pint,I_1 = exp − 1 − exp − proposed EARS schemes are verified by Monte-Carlo simula-
Ξ x g!xg x
g=0 tions. Let us denote γI = I/N0 and γp = Pp /N0 . Nst , Nrt , Ndt
 and Net respectively represent the antenna numbers at CS, CR,
× Θ(x, y)f|hpi |2 (x)f|hid |2 (y)dxdy (65)
k∈Dn ,k=i
CD and E. For simplicity, let “T” represent the theoretical result,
“S” represent the simulated result.
and Fig. 3 shows that OPs and IPs versus γp for EARS and NEARS
 Nv −1 as well as DC schemes. One can clearly see that the performance
Δg5 z g Δ6
EARS
Pint,I_2 = exp(−Δ5 z) exp − gap of solid lines and discrete dots does exist but it has no obvious
ζ g=0
g! z impact on our analysis, which indicates that the feasibility of
 our consideration for average interference at cognitive users
× T(y, z)f|hid |2 (y)f|hip |2 (z)dydz, (66) from PT. In Fig. 4, the theoretical results match well with our
k∈Dn ,k=i simulated results, certifying the correctness of theoretical OP

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 YAN et al.: ENERGY-AWARE RELAY SELECTION IMPROVES SECURITY-RELIABILITY TRADEOFF 5123

Fig. 3. IPs and OPs of DC and NEARS as well as EARS schemes versus SNR Fig. 5. SRT of DC and NEARS as well as EARS schemes versus η
with α = 0.4, η = 0.6, θ = 0.8, γI = 20 dB, σsd
2 = σ 2 = σ 2 = 0.5, σ 2 =
ps
with α = 0.4, θ = 0.8, γI = 20 dB, σsd
2 = σ 2 = σ 2 = 0.5, σ 2 = σ 2 =
si id ps pi
si id
σpi
2 = 0.2, σ 2 = σ 2 = σ 2 = 0.2, σ 2 = σ 2 = 0.2, MER = 10 dB, T =
J1 J2 Ji ip sp
0.4, σip = σsp = 0.2, σse = σie = 0.1, σJ1
2 2 2 2 2 = σ 2 = σ 2 = 0.4, T = 1s,
J2 Ji
1s, Nr = 4, Nv = 4. Nr = 4, Nv = 1 and γp = 5 dB.

Fig. 6. SRT of DC and NEARS as well as EARS schemes versus θ

Fig. 4. IPs and OPs of DC and NEARS as well as EARS schemes versus γp with α = 0.4, η = 0.8, γI = 20 dB, σsd
2 = σ 2 = σ 2 = 0.5, σ 2 = σ 2 =
si id ps pi
with α = 0.4, η = 0.6, θ = 0.8, γI = 20 dB, σsd
2 = σ 2 = σ 2 = 0.5, σ 2 =
si id ps
2 = σ 2 = 0.2, σ 2 = σ 2 = 0.1, σ 2 = σ 2 = σ 2 = 0.4, T = 1s,
0.4, σip sp se ie J1 J2 Ji
σpi = 0.2, σip = σsp = 0.2, σJ1 = σJ2 = σJi
2 2 2 2 2 2 = 0.2, MER=10 dB, T =
Nr = 4, Nv = 1 and γp = 5 dB.
1s, Nr = 5, Nv = 8.

IP of EARS is the same as NEARS. This can be explained that the

and IP. Moreover, the performance of OP for EARS is better
than that of NEARS and DC. Whilst, the OP performance of
these three schemes increases and IP performance decreases
with an increasing γp . It is shown that a tradeoff exists between
security and reliability for cognitive transmission. Additionally,
Fig. 4 shows that if γp is lower than 15 dB, the received
signal to interference noise ratio (SINR) at destination of DC
|2 |hsd |2
is αηγγPP|hσsd
2 +1 , which shows an increasing γp will lead to an
J1
enhancement in OP performance. However, if γp increases to a
2 2
| |hsd |
certain degree, the SINR tends to αη|hsd 2
σJ1
, thereby causing
the OP arrive at performance floor. The same analysis can be
carried out for IP. And the effect of γp on NEARS and EARS in
terms of OP and IP can be analyzed as DC scheme. Moreover, the
CSIs of wiretap channels from CRs to Es is unavailable for CRs
in NEARS and our proposed EARS schemes. Consequently, the
IP performance for both two relay selection schemes achieves
no benefits with an increasing number of CRs.
In Fig. 5, we depict OPs and IPs for DC, NEARS and EARS
schemes versus η. It is seen from Fig. 5 that the efficiency of
energy conversion η increases from 0 to 1 results in the reduction
of OPs for all schemes. While, the IP performance of these three
schemes becomes to be deteriorated as η increases. This phe-
nomenon can be explained that more harvested energy will be
used for cognitive transmission with an increasing η, enhancing
the received SINR at CD and Es, separately. Additionally, Fig. 6
shows that OPs and IPs versus remaining ratio θ of total energy
harvested from PT at each CR for transmitting information. One

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 5124 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO. 5, MAY 2020

Fig. 7. SRT of DC and NEARS as well as EARS schemes for differ- Fig. 9. SRT of DC, NEARS and EARS schemes for different MER with α =
ent Nr with α = 0.4, η = 0.6, θ = 0.8, γI = 20 dB, σsd
2 = σ2 = σ2 =
si id 0.4, η = 0.6, θ = 0.8, γI = 20 dB, σsd
2 = σ 2 = σ 2 = 0.5, σ 2 = σ 2 =
si id ps pi
0.5, σps
2 = σ 2 = 0.2, σ 2 = σ 2 = 0.2, σ 2 = σ 2 = σ 2 = 0.2, T = 1s,
pi ip sp J1 J2 Ji
2 = σ 2 = 0.1, σ 2 = σ 2 = σ 2 = 0.1, T = 1s, N = 6, N = 1
0.1, σip sp J1 J2 Ji r v
Nv = 8, MER = 10 dB and γp = [−10,30]dB. and γp = [−10,30]dB.

Fig. 8. SRT of DC and NEARS as well as EARS schemes for different

Nv with α = 0.3, η = 0.8, θ = 0.8, γI = 20 dB, σsd
2 = σ 2 = σ 2 = 0.5, Fig. 10. SRT of DC, NEARS and EARS schemes for different MER with α =
si id
0.4, η = 0.6, θ = 0.8, γI = 20 dB, σsd
2 = σ 2 = σ 2 = 0.5, σ 2 = σ 2 =
σps
2 = σ 2 = 0.2, σ 2 = σ 2 = 0.2, σ 2 = σ 2 = 0.1, σ 2 = σ 2 = σ 2 =
pi ip sp ie se J1 J2 Ji
si id ps pi

0.2, T = 1s, Nr = 10 and γp = [−10,30]dB. 0.2, σip = σsp = 0.2, σJ1 = σJ2 = σJi
2 2 2 2 2 = 0.2, T = 1s, N = 2, N = 1,
r v
Net = 2, MER = 10 dB and γp = [−10,20]dB.

can clearly see from Fig. 6 that the OP performance of EARS

and NEARS is worse than that of DC when a large number of
energy at selected CR are used to decode, demodulate, re-code
and re-modulate. While with an increasing θ, the more energy is
employed to transmit information, which significantly enhances
OP performance for EARS and NEARS and leads to EARS and
NEARS achieve better performance than DC in terms of OP.
Fig. 7 shows that the SRT of DC, NEARS and EARS schemes
versus different Nr . It is clear to see that our proposed EARS
scheme performs better than DC and NEARS in terms of SRT
performance. Moreover, the number of CRs increases from 8 to
24 causes the SRT performance of EARS and NEARS improve,
apparently. This evidently illustrates that SRT performance ben-
efits from multiple relays in cognitive transmission. In addition,
the SRT performance gap of EARS is more notable than that of
NEARS as the number of CRs increases from 8 to 24, which
also further indicates that the superiority of our proposed EARS
scheme in terms of SRT. Fig. 8 shows that the SRT performance
of these three schemes becomes deteriorated as Nv increases
from 1 to 4. As a result, the more private information of cognitive
radio systems leaks to Es.
Fig. 9 shows that the SRT of DC and NEARS as well
2 2
as EARS schemes versus different MER (σid /σie ) which is
main-to-eavesdropper ratio [8], [14] and [17]. It is shown from
Fig. 9 that MER increases from 0 dB to 10 dB causes the SRT
performance of these three schemes improve, accordingly. To
be more specific, if an IP is given by a specific value, the OPs of
these three schemes significantly reduces as MER increases. In
other words, the security can be achieved at the cost of reliability
for cognitive wireless transmission. From Fig. 10, we can see
that the similar SRT performance of DC, NEARS and EARS
in SISO can be got in MIMO. Moreover, the SRT performance
of MIMO performs better than that of SISO and the increasing
antenna numbers at CS, CR and CD will significantly enhance
the SRT performance for cognitive transmission.

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 YAN et al.: ENERGY-AWARE RELAY SELECTION IMPROVES SECURITY-RELIABILITY TRADEOFF 5125

V. CONCLUSION and
 
We have studied the physical layer security in terms of SRT ε1 γI1 ε1 γI1 γI1 ε1 √
performance for a CCR-EH system, where a CS having energy
dc
Pout,II = 2
K1 2
− 2
K1 ( ε1 ε2 ),
σsp σsp σsp ε2
harvester communicates with a CD assisted by multiple CRs (A.6)
who are equipped with energy harvesters. Meanwhile, multiple where ε1 = 4
and ε2 = λ0
+ σγ2I1 . Then substituting
2
aγp1 σps 2
σsd
Es were assumed to tap the confidential CS-CD transmission dc
sp

resulted from the broadcast nature of wireless communications. (A.5) and (A.6) into (25), the closed-form expression of Pout
To fight against eavesdropping attacks, we proposed NEARS can be easily formulated.
and EARS schemes depending on whether the CSIs of energy
links spanning from PT to CRs are unknown or known to CRs. APPENDIX B
Additionally, we compared the SRT performance of EARS with DERIVATIONS OF (35) AND (36)
that of DC and NEARS schemes. Closed-form expressions of Depending on the PDF of X1 and Y1 , we can rewrite (33) and
OP and IP for all aforementioned three schemes were derived. (34) as
Numerical results showed that our proposed EARS scheme has  ∞  Nv 
obvious advantages over DC and NEARS in terms of SRT. 2 λs
Pint,I =
dc
Pr |hse | >
0 e=1
aγp2 x1
APPENDIX A
DERIVATIONS OF (28) AND (29) γI2
× Pr |hsp |2 < fX1 (x1 )dx1
aγp2 x1
Let denote X1 = |hps | and Y1 = |hsp | , and consider that
2 2
⎡ ⎤
they are independently and exponentially distributed with re-  ∞
⎣1 − F Nv λs ⎦
2
spective means σps 2
and σsp . Thus, the PDFs of X1 and Y1 are =
0 |hse |2 aγp2 x1
separately expressed as e=1

1 x1 γI2
fX1 (x1 ) = 2
exp − 2 (A.1) × F|hsp |2 fX1 (x1 )dx1 (B.1)
σps σps aγp2 x1
and and
 N 
1 y1 ∞ v
λs
fY1 (y1 ) = 2 exp − 2 . (A.2) dc
Pint,II = Pr |hse | > 2
y1
σsp σsp γ
0 e=1 I2
Then according to (A.1) and (A.2), we can respectively compute
(26) and (27) as γI2
× Pr |hps |2 > fY1 (y1 )dy1
 ∞  bγp2 y1
λ0 ⎡ ⎤
Pout,I =
dc
1 − exp − 2 x 
aγp1 σsd ∞
0 1
⎣1 − F Nv λs
  = y1 ⎦
γI1 |hse |2 γI2
× 1 − exp − 2 x
fX1 (x1 )dx1 (A.3) 0
e=1
aγp1 σsp 1  
γI2
and × 1 − F|hps |2 fY1 (y1 )dy1 . (B.2)
 ∞   bγp2 y1
λ0
dc
Pout,II = 1 − exp(− y1 ) N v
0
2
γI1 σsd According to [25], we have the CDF of e=1 |hse |2
γI1 Nv −1
× exp − fY1 (y1 )dy1 , (A.4) z zg
2 y
aγp1 σps 1 F Nv (z) = 1 − exp − 2 2g
. (B.3)
|hse | 2 σse g=0 g!σse
e=1
which can be further respectively given by
  Substituting (B.3) into (B.1) and (B.2), we can respectively get
ε1 γI1 ε1 γI1 ε1 λ 0 dc
Pint,I dc
and Pint,II as (B.4) shown at the bottom of this page, and
Pout,I = 1 −
dc
2
K1 2
− 2 λs
σsp σsp σsd (B.5) shown at the bottom of the next page, where Λ1 = aγp2 σ2 , se
  Λ2 = aγγp2I2σ2 , Λ3 = γI2λσs 2 and Λ4 = aγγp2I2σ2 .
ε1 λ 0 √ √ sp se ps
× K1 2
+ ε1 ε2 K1 ( ε1 ε2 ) (A.5) Thus, combining (B.4) and (B.5) as well as (32), we can obtain
σsd dc
the closed-form expression of Pint .
 Ne −1
∞
λs λgs x−g γI2
dc
Pint,I = exp − 2 x
1
2 )g
1 − exp − 2 x
fX1 (x1 )dx1
0 aγp2 σse 1 g=0
g!(aγp2 σse aγp2 σsp 1

Ne −1
2Λ1 g  Ne −1
2Λ1 g   
2 1−g −2 ) − 2 1−g −2 .
= 2
(Λ 1 σ ps ) 2 K
1−g (2 Λ1 σps 2
[(Λ 1 + Λ 2 )σ ps ] 2 K
1−g 2 (Λ 1 + Λ 2 )σ ps (B.4)
g=0
g!σps g=0
g!σps

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 5126 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO. 5, MAY 2020

⎛ ⎞
APPENDIX C 2|Dn |−1 −1 
y2 ⎠
DERIVATIONS OF (47) AND (48) ψIII_2 = (−1)|Dn (m)| exp ⎝− 2
Φ σkd
We consider that |hpi | , |hid | and |hip | are independently
2 2 2 m=1 k∈Dn (m)

and exponential distributed random variables with respective × fX2 ,Y2 (x2 , y2 )dx2 dy2 , (C.8)
means σpi 2 2
, σid and σip 2
. For notational simplicity, denoting X2 =  

|hpi | , Y2 = |hid | and Z2 = |hip |2 . Hence, the joint PDF of
2 2 γI1
ψIII_3 = exp − fX2 ,Y2 (x2 , y2 )dx2 dy2
(X2 , Y2 ) and (Y2 , Z2 ) can be expressed as Φ bγ 2
p1 σip x2
 
1 x2 y2 (C.9)
fX2 ,Y2 (x2 , y2 ) = 2 2 exp − 2 − 2 (C.1)
σpi σid σpi σid and
2|Dn |−1 −1   
and |Dn (m)| γI1
  ψIII_4 = (−1) exp − 2 x
bγp1 σip
1 y2 z2 Φ 2
fY2 ,Z2 (y2 , z2 ) = 2 2 exp − 2 − 2 . (C.2) ⎛
m=1
⎞
σid σip σid σip
y2 ⎠
Therefore, substituting (C.1) and (C.2) into (45) and (46), sep- × exp ⎝− 2
fX2 ,Y2 (x2 , y2 )dx2 dy2 . (C.10)
σkd
arately, yields k∈Dn (m)

γI1 Combining (C.1) and (C.7), we have
NEARS
Pout,I_1 = Pr |hip |2 <  
Φ bγ p1 x2
 ψIII_1 = 1 − Λ5 λ1 K1 ( Λ5 λ1 ), (C.11)
× Pr(|hkd |2 < y2 )fX2 ,Y2 (x2 , y2 )dx2 dy2 (C.3)
where Λ5 = bγp1 σ42 σ2 . Similarly, with the aid of fX2 ,Y2 (x2 , y2 ),
k∈Dn ,k=i id pi
(C.8) can be further computed as
and
 ψIII_2
γI1
NEARS
Pout,I_2 = Pr |hpi |2 >
bγp1 z2 2|Dn |−1 −1 2|Dn |−1 −1

τ
(−1)|Dn (m)| Λ6 (−1)|Dn (m)| Λ6
= −
× Pr(|hkd |2 < y2 )fY2 ,Z2 (y2 , z2 )dy2 dz2 , (C.4) m=1
2
σid m=1
2
σid
k∈Dn ,k=i  
4λ1 4λ1
where Φ = {(x2 , y2 )|bγp1 x2 y2 < λ1 , x2 > 0, y2 > 0} and × K1 (C.12)
τ = {(y2 , z2 )|γI1 y2 < λ1 z2 , y2 > 0, z2 > 0}. In addition, bγp1 Λ6 σpi
2 bγp1 Λ6 σpi
2

Dn represents the n-th non-subset of CRs. Moreover,

  1 −1
2 where Λ6 = ( σ12 + 2 ) . Besides, ψIII_3 can also
k∈Dn ,k=i Pr(|hkd | < y2 ) can be further got with the id
k∈Dn (m) σkd
aid of binomial theorem, yielding be formulated as

Pr(|hkd |2 < y2 ) ψIII_3
k∈Dn ,k=i   &
'  
⎛ ⎞ γI1 Λ7 γI1 Λ7 ' γ I1 λ1
2|Dn (m)|−1 −1 = k1 (
− Λ7 + 2
y2 ⎠ 2
σip 2
σip 2
σip σid
=1+ (−1)|Dn (m)| exp ⎝− 2
.
σkd ⎛&
m=1 k∈Dn (m) '  ⎞
' γ λ
K1 ⎝(Λ7
(C.5)
+ 2 ⎠,
I1 1
× 2
(C.13)
Then combining (C.5) and (C.3), we obtain σip σid
NEARS
Pout,I_1 = ψIII_1 + ψIII_2 − ψIII_3 − ψIII_4 , (C.6) where Λ7 = 4
2 . Moreover, (C.10) can be rewritten as (C.14)
bγp1 σpi
where ψIII_1 , ψIII_2 , ψIII_3 and ψIII_4 are formulated as
 2|Dn |−1 −1
(−1)|Dn (m)| Λ6  
ψIII_1 = fX2 ,Y2 (x2 , y2 )dx2 dy2 , (C.7) ψIII_4 = 2
γI1 Λ5 K1 ( γI1 Λ5 )
σid
Φ m=1

 ∞ Ne −1
λs y1 λgs y1g γI2
dc
Pint,II = exp − g exp − fY1 (y1 )dy1
0
2
γI2 σse g=0
g!(γI2 σse )
2 2 y
aγp2 σps 1

Ne −1
2Λ3 g ) * g+1   
−2 −1 2 −2 ) .
= 2
Λ 4 (Λ 3 + σ sp ) K 1+g 2 Λ 4 (Λ 3 + σ sp (B.5)
g=0
g!σsp

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 YAN et al.: ENERGY-AWARE RELAY SELECTION IMPROVES SECURITY-RELIABILITY TRADEOFF 5127

2|Dn |−1 −1
(−1)|Dn (m)| Λ6   [3] X. Tang, Y. Cai, Y. Deng, Y. Huang, W. Yang, and W. Yang, “Energy-
− 2
Λ8 K1 ( Λ8 )], (C.14) constrained SWIPT networks: Enhancing physical layer security with
σid FD self-jamming,” IEEE Trans. Inf. Forensics Secur., vol. 14, no. 1,
m=1 pp. 212–222, Jan. 2019.
2
σid γI1 Λ5
[4] J. Men, J. Ge, and C. Zhang, “A joint relay-and-antenna selection scheme
wherein, Λ8 = 2
σip
+ 4λ1
2 Λ .
bγp1 σpi 6
Finally, substituting in energy-harvesting MIMO relay networks,” IEEE Signal Process. Lett.,
NEARS
vol. 23, no. 4, pp. 532–536, Apr. 2016.
(C.11), (C.12), (C.13) and (C.14) into (C.6), Pout,I_1 can be [5] A. A. Nasir, X. Zhou, S. Durrani, and R. A. Kennedy, “Relaying protocols
expressed as (47). for wireless energy harvesting and information processing,” IEEE Trans.
Wireless Commun., vol. 12, no. 7, pp. 3622–3636, Jul. 2013.
Similarly to (C.3), we can reformulate (C.4) as [6] S. K. Yadav and T. Acharaya, “Outage analysis in cooperative cognitive ra-
dio networks with simultaneous wireless information and power transfer,”
NEARS
Pout,I_2 = ψIII_5 + ψIII_6 , (C.15) in Proc. IEEE Conf. Commun., 2016, pp. 1–5.
[7] D. Wang, R. Zhang, X. Cheng, and L. Yang, “Capacity-enhancing full-
where ψIII_5 and ψIII_6 are given by duplex relay networks based on power-splitting (PS-)SWIPT,” IEEE
   Trans. Veh. Technol., vol. 66, no. 6, pp. 5445–5450, Jun. 2017.
γI1 [8] Y. Zou, X. Wang, and W. Shen, “Optimal relay selection for physical-layer
ψIII_5 = exp − 2 z
fY2 ,Z2 (y2 , z2 )dy2 dz2 security in cooperative wireless networks,” IEEE J. Sel. Areas Commun.,
τ aγp1 σpi 2 vol. 31, no. 10, pp. 2099–2111, Oct. 2013.
(C.16) [9] D. Wang, B. Bai, W. Zhao, and Z. Han, “A survey of optimization
approaches for wireless physical layer security,” IEEE Commun. Surv.
and Tut., vol. 21, no. 2, pp. 1878–1911, Apr.–Jun. 2019.
2|Dn |−1 −1    [10] A. D. Wyner, “The wire-tap channel,” Bell Syst. Tech. J., vol. 54, no. 8,
Dn (m) γI1 pp. 1355–1387, 1975.
ψIII_6 = (−1) exp − 2 z [11] H. Guo, Z. Yang, L. Zhang, J. Zhu, and Y. Zou, “Power-constrained
m=1 τ aγp1 σpi 2 secrecy rate maximization for joint relay and jammer selection assisted
⎛ ⎞ wireless networks,” IEEE Trans. Commun., vol. 65, no. 5, pp. 2180–2193,
y2 ⎠ May 2017.
× exp ⎝− 2
fY2 ,Z2 (y2 , z2 )dy2 dz2 . [12] H. Lei et al., “Secrecy outage analysis for cooperative NOMA systems
σkd with relay selection scheme,” IEEE Trans. Commun., vol. 67, no. 9,
k∈Dn (m)
pp. 6282–6298, Sep. 2019.
(C.17) [13] H. Lei et al., “On secrecy outage of relay selection in underlay cognitive
radio networks over Nakagami-m fading channel,” IEEE Trans. Cogn.
Then, by using the result of (C.2), (C.16) and (C.17) can be Commun. Netw., vol. 3, no. 4, pp. 614–627, Dec. 2017.
respectively rewritten as [14] J. Zhu, Y. Zou, G. Wang, Y. Yao, and G. K. Karagiannidis, “On se-
crecy performance of antenna-selection-aided MIMO systems against
  eavesdropping,” IEEE Trans. Veh. Technol., vol. 65, no. 1, pp. 214–225,
γI1 Λ7 γI1 Λ7 Jan. 2016.
ψIII_5 = 2
K1 2 [15] H. Lei et al., “Secrecy outage performance of transmit antenna se-
σip σip
lection for MIMO underlay cognitive radio systems over nakagami-m
1 γI1 Λ7   channels,” IEEE Trans. Veh. Technol., vol. 66, no. 3, pp. 2237–2250,
− K1 γI1 Λ9 Λ7 (C.18) Mar. 2017.
2
σip Λ9 [16] S. Asaad, A. Bereyhi, A. M. Rabiei, R. R. Mller, and R. F. Schaefer,
“Optimal transmit antenna selection for massive MIMO wiretap chan-
and nels,” IEEE J. Sel. Areas Commun., vol. 66, no. 9, pp. 1518–1531,
Sep. 2017.
2|Dn |−1 −1
  [17] Y. Zou, X. Wang, and W. Shen, “Physical-layer security with multiuser
(−1)|Dn (m)| Λ6 γI1 Λ7 γI1 Λ7 scheduling in cognitive radio networks,” IEEE Trans. Commun., vol. 61,
ψIII_6 = 2 2
K1 2 no. 12, pp. 5103–5113, Dec. 2013.
m=1
σid σip σip
[18] B. Li, Y. Zou, J. Zhou, F. Wang, W. Cao, and Y. Yao, “Secrecy outage prob-
&  
2|Dn |−1 −1 ' ability analysis of friendly jammer selection aided multiuser scheduling for
(−1) |Dn (m)|
Λ6 '
(γI1 Λ7 1 λ 1
wireless networks,” IEEE Trans. Commun., vol. 67, no. 5, pp. 3482–3495,
− + May 2019.
m=1
σ 2 σ2
ip id
2
σip γI1 Λ6 [19] H. Xing, L. Liu, and R. Zhang, “Secrecy wireless information and power
transfer in fading wiretap channel,” IEEE Trans. Veh. Technol., vol. 65,
⎛&
'  ⎞ no. 1, pp. 180–190, Jan. 2016.
' 1 λ1 [20] X. Liu, Z. Li, and C. Wang, “Secure decode-and-forward relay SWIPT
× K1 ⎝(γI1 Λ7 + ⎠, (C.19) systems with power splitting schemes,” IEEE Trans. Veh. Technol., vol. 67,
2
σip γI1 Λ6 no. 8, pp. 7341–7354, Aug. 2018.
[21] S. S. Kalamkar and A. Banerjee, “Secure communication via a wireless
λ1 energy harvesting untrusted relay,” IEEE Trans. Veh. Technol., vol. 66,
where Λ9 = 1
2
σip
+ 2 .
γI1 σid
Hence, the final expression of no. 3, pp. 2199–2213, Mar. 2017.
NEARS
Pout,I_2
can be got as (48) using the results of (C.18) and (C.19), [22] T. M. Hoang, T. Q. Duong, N. Vo, and C. Kundu, “Physical layer security
in cooperative energy harvesting networks with a friendly jammer,” IEEE
which completes the proof of (47) and (48). Wireless Commun, Lett., vol. 6, no. 2, pp. 174–177, Apr. 2017.
[23] V. N. Vo, T. G. Nguyen, C. So-In, Z. A. Baig, and S. Sanguanpong, “Secrecy
outage performance analysis for energy harvesting sensor networks with
REFERENCES a jammer using relay selection strategy,” IEEE Access, vol. 6, pp. 23406–
[1] Y. Wu, A. Khisti, C. Xiao, G. Caire, K. Wong, and X. Gao, “A survey 23419, 2018.
of physical layer security techniques for 5G wireless networks and chal- [24] H. Lei, Z. Dai, K. Park, W. Lei, G. Pan, and M. Alouini, “Secrecy
lenges ahead,” IEEE J. Sel. Areas Commun., vol. 36, no. 4, pp. 679–695, outage analysis of mixed RF-FSO downlink SWIPT systems,” IEEE Trans.
Apr. 2018. Commun., vol. 66, no. 12, pp. 6384–6395, Dec. 2018.
[2] H. Yu, S. Guo, Y. Yang, L. Ji, and Y. Yang, “Secrecy energy efficiency [25] N. Zhao, F. R. Yu, and V. C. M. Leung, “Wireless energy harvesting in
optimization for downlink two-user OFDMA networks with SWIPT,” interference alignment networks,” IEEE Commun. Mag., vol. 53, no. 6,
IEEE Syst. J., vol. 13, no. 1, pp. 324–335, Mar. 2019. pp. 72–78, Jun. 2015.

Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
 5128
[26] N. Zhao, Y. Cao, F. R. Yu, Y. Chen, M. Jin, and V. C. M. Leung,
“Artificial noise assisted secure interference networks with wireless power
transfer,” IEEE Trans. Veh. Technol., vol. 67, no. 2, pp. 1087–1098,
Feb. 2018.
[27] N. Zhao, S. Zhang, F. R. Yu, Y. Chen, A. Nallanathan, and V. C. M. Leung,
“Exploiting interference for energy harvesting: A survey, research issues,
and challenges,” IEEE Access, vol. 5, pp. 10403–10421, 2017.
[28] H. Lei, M. Xu, I. S. Ansari, G. Pan, K. A. Qaraqe, and M. Alouini, “On
secure underlay MIMO cognitive radio networks with energy harvesting
and transmit antenna selection,” IEEE Trans. Green Commun. Netw., vol. 1,
no. 2, pp. 192–203, Jun. 2017.
[29] X. Ding, Y. Zou, G. Zhang, X. Chen, X. Wang, and L. Hanzo, “The
security-reliability tradeoff of multiuser scheduling aided energy harvest-
ing cognitive radio networks,” IEEE Trans. Commun., vol. 67, no. 6,
pp. 3890–3904, Jun. 2019.
[30] M. Li, H. Yin, Y. Huang, Y. Wang, and R. Yu, “Physical layer security
in overlay cognitive radio networks with energy harvesting,” IEEE Trans.
Veh. Technol., vol. 67, no. 11, pp. 11274–11279, Nov. 2018.
[31] X. Chen, L. Guo, X. Li, C. Dong, J. Lin, and P. T. Mathiopoulos, “Secrecy
rate optimization for cooperative cognitive radio networks aided by a wire-
less energy harvesting jammer,” IEEE Access, vol. 6, pp. 34127–34134,
2018.
[32] M. Xu, T. Jing, X. Fan, Y. Wen, and Y. Huo, “Secure transmission solutions
in energy harvesting enabled cooperative cognitive radio networks,” in
Proc. IEEE Conf. Commun., 2018, pp. 1–6.
[33] S. Gautam, T. X. Vu, S. Chatzinotas, and B. Ottersten, “Cache-aided
simultaneous wireless information and power transfer (SWIPT) with relay
selection,” IEEE J. Sel. Areas Commun., vol. 37, no. 1, pp. 187–201,
Jan. 2019.
[34] P. Yan, Y. Zou, and J. Zhu, “Energy-aware multiuser scheduling for
physical-layer security in energy-harvesting underlay cognitive radio sys-
tems,” IEEE Trans. Veh. Technol., vol. 67, no. 3, pp. 2084–2096, Mar. 2018.
[35] H. Alves, R. D. Souza, M. Debbah, and M. Bennis, “Performance of
transmit antenna selection physical layer security schemes,” IEEE Signal
Process. Lett., vol. 19, no. 6, pp. 372–375, Jun. 2012.
[36] N. Yang, P. L. Yeoh, M. Elkashlan, R. Schober, and I. B. Collings,
“Transmit antenna selection for security enhancement in MIMO wiretap
channels” IEEE Trans. Commun., vol. 61, no. 1, pp. 144–154, Jan. 2013.
[37] J. N. Laneman, D. N. C. Tse, and G. W. Wornell, “Cooperative diversity in
wireless networks: Efficient protocols and outage behavior,” IEEE Trans.
Inf. Theory, vol. 50, no. 12, pp. 3062–3080, Dec. 2004.
[38] X. Chen, L. Lei, H. Zhang, and C. Yuen, “Large-scale MIMO relaying
techniques for physical layer security: AF or DF?” IEEE Trans. Wireless
Commun., vol. 14, no. 9, pp. 5135–5146, Sep. 2015.
[39] X. Tang, R. Liu, P. Spasojevic, and H. V. Poor, “On the throughput of
secure hybrid-ARQ protocols for Gaussian block-fading channels,” IEEE
Trans. Inf. Theory, vol. 55, no. 4, pp. 1575–1591, Apr. 2009.
[40] Y. Zou, “Intelligent interference exploitation for heterogeneous cellular
networks against eavesdropping,” IEEE J. Sel. Areas Commun., vol. 36,
no. 7, pp. 1453–1464, Jul. 2018.
[41] I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and Products,
7th ed. New York, NY, USA: Academic, 2007.
Peishun Yan received the B.Eng. degree in electronic
and information engineering from Nantong Univer-
sity, Nantong, China, in 2015 and the M.S. degree in
signal and information processing from the Nanjing
University of Posts and Telecommunications, Nan-
jing, China, in 2018, where he is currently working
toward the Ph.D. degree. His research interests in-
clude cognitive radio, cooperative communications,
and wireless security and quantum communications.
Authorized licensed use limited to: International Institute of Information Technology Bangalore. Downloaded on June 29,2022 at 18:41:05 UTC from IEEE Xplore. Restrictions apply.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 69, NO. 5, MAY 2020
Yulong Zou (Senior Member, IEEE) received the
B.Eng. degree in information engineering from the
Nanjing University of Posts and Telecommunications
(NUPT), Nanjing, China, in July 2006, the first Ph.D.
degree in electrical engineering from the Stevens
Institute of Technology, Hoboken, NJ, USA, in 2012,
and the second Ph.D. degree in signal and informa-
tion processing from NUPT, Nanjing, China, in July
2012. He is a Full Professor and Doctoral Super-
visor NUPT, Nanjing, China. Dr. Zou was awarded
the 9th IEEE Communications Society Asia-Pacific
Best Young Researcher in 2014 and a co-receipt of the Best Paper Award
at the 80th IEEE Vehicular Technology Conference in 2014. He has served
as an Editor for the IEEE COMMUNICATIONS SURVEYS & TUTORIALS, IEEE
COMMUNICATIONS LETTERS, EURASIP Journal on Advances in Signal Pro-
cessing, IET Communications, and China Communications. In addition, he has
acted as TPC members for various IEEE sponsored conferences, e.g., IEEE
ICC/GLOBECOM/WCNC/VTC/ICCC, etc.
Xiaojin Ding received the Ph.D. degree in informa-
tion and communication engineering from the Na-
tional Mobile Communication Research Laboratory,
Southeast University, Nanjing, China. He is currently
a Lecturer with the Nanjing University of Posts and
Telecommunications, Nanjing, China. His research
interests include space information networks, coop-
erative communications, and physical-layer security.
Jia Zhu received the B.Eng. degree in computer
science and technology from the Hohai University,
Nanjing, China, in July 2005, and the Ph.D. de-
gree in signal and information processing from the
Nanjing University of Posts and Telecommunications
(NUPT), Nanjing, China, in April 2010. She is a Full
Professor with NUPT, Nanjing, China. From June
2010 to 2012, she was a Postdoctoral Research Fellow
with the Stevens Institute of Technology, NJ, USA.
Since November 2012, she has been a Full-Time
Faculty Member with the Telecommunication and
Information School of NUPT, Nanjing, China. Her general research interests
include the cognitive radio, physical-layer security, and communications theory.