Content uploaded by Gaofeng Pan
Author content
All content in this area was uploaded by Gaofeng Pan on Nov 02, 2021
Content may be subject to copyright.
IEEE SYSTEMS JOURNAL, VOL. 12, NO. 3, SEPTEMBER 2018 2129
Relay Cooperation and Outage Analysis in Cognitive
Radio Networks With Energy Harvesting
Jing He, Songtao Guo , Member, IEEE,GaofengPan , Member, IEEE, Yuanyuan Yang , Fellow, IEEE,
and Defang Liu
Abstract—Cooperative cognitive radio (CCR) is a novel
paradigm to improve both radio spectrum efficiency and commu-
nication quality. In this paper, we consider a CCR network (CCRN)
with energy harvesting in which multiple secondary transmitters
(STs) are able to harvest energy from the received signals to serve
their own receivers and primary transmitters (PTs). Besides, not
only PT’s transmitting signal but also interferers, i.e., RF signals in
ambient environment received at STs can be used for energy har-
vesting. Moreover, we aim to analyze the outage probability (OP)
for two relay cooperation schemes including single-relay coopera-
tion (SC) and multirelay cooperation (MC) in CCRNs while con-
sidering energy harvesting. In addition, we consider Nakagami-m
fading rather than Rayleigh fading channels as it is more practi-
cal and general for characterizing the fading effect over wireless
channels. Most importantly, we compare the outage performance
of SC and MC with that of direct transmission and investigate the
tradeoff between the primary and secondary users’ performance.
Simulation results are presented to validate our analysis, which
show the exact performance for different parameter settings.
Index Terms—Cognitive radio networks (CRNs), energy har-
vesting, Nakagami-mfading, outage probability (OP), relay
cooperation.
I. INTRODUCTION
IN RECENT years, it has been particularly noticed that spec-
trum resource scarcity is one of the obstacles that block
some beneficial effects of wireless communications. However,
according to federal communication commission, most licensed
bands remain unoccupied for a large part of time [1], [2]. There-
fore, an emerging wireless technology, cognitive radio (CR), has
attracted a wide public concern due to its remarkable improve-
ment in spectrum utilization efficiency. Normally, users in CR
Manuscript received April 16, 2016; revised August 19, 2016; accepted
November 2, 2016. Date of publication December 1, 2016; date of cur-
rent version August 23, 2018. This work was supported in part by the
National Natural Science Foundation of China (No. 61373179, 61373178,
61402381), in part by the Natural Science Key Foundation of Chongqing
(cstc2015jcyjBX0094), in part by the Fundamental Research Funds for the
Central Universities (XDJK2013C094, XDJK2013A018, XDJK2015C010,
XDJK2015D023, XDJK2016A011), and in part by the Science and Technology
Leading Talent Promotion Project of Chongqing (cstc2013kjrc-ljrccj40001).
(Corresponding author: Songtao Guo.)
J. He, S. Guo, and G. Pan are with the College of Electronic and Informa-
tion Engineering, Southwest University, Chongqing 400715, China (e-mail:
984546045@qq.com; stguo@swu.edu.cn; gfpan@swu.edu.cn).
Y. Yang is with the Department of Electrical and Computer Engi-
neering, Stony Brook University, Stony Brook, NY 11794 USA (e-mail:
yuanyuan.yang@stonybrook.edu).
D. Liu is with the College of Chemistry and Chemical Engineering, Southwest
University, Chongqing 400715 China (e-mail: liudf2009@swu.edu.cn).
Digital Object Identifier 10.1109/JSYST.2016.2628862
networks (CRNs) are classified into two groups: primary users
(PUs) and secondary users (SUs). PUs are authorized to use the
licensed spectrum, while no corresponding licensed spectrum
being assigned to SUs. Most importantly, PUs could share the
licensed spectrum with SUs in CRNs based on some given rules.
In other words, SUs could transmit information on PUs’ licensed
spectrum only if the secondary usage does not cause any harmful
effects on PUs. Compared with other wireless communication
systems, this exactly innovates wireless communication and is
an efficient way to make the most of precious spectrum resource.
There exist three common paradigms in CRNs: interweave,
underlay, and overlay [3], [4]. In addition, SUs in different
paradigms are constrained by different rules to access the li-
censed spectrum. In interweave CRN, secondary transmitter
(ST) detects whether the licensed spectrum is occupied by pri-
mary transmitter (PT) or not at first to find a spectrum hole
which means the licensed spectrum has not been occupied by
PU. Then, ST makes an access decision on the basis of the de-
tection result to avoid collision between PU and SU, i.e., PUs
and SUs access the licensed spectrum at different times and PU
has priority over SU in accessing the licensed spectrum [5]–[7].
In underlay CRN, SUs transmit concurrently with PUs only if
the interference temperature constraints imposed by PUs are
met to avoid their performance degradation [8]–[10]. However,
both PUs in interweave and underlay CRNs do not take SU’s
existence and needs into account.
In overlay CRN, STs cooperate with PT as relays and aim
to ask for a right to access the licensed spectrum to trans-
mit their own information. It is obvious that overlay CRN is
cooperative CRN (CCRN). In fact, relay communication has
been regarded as a promising scheme to improve the through-
put and coverage of wireless communication systems [11], [12]
and it also has gotten much application in CR systems recently
[13], [14]. This is attributed to the use of intermediate relay
nodes, which are used to assist transmission from the source
to the destination. In CCRNs, no additional relay nodes are
needed and the CCRN is an emerging paradigm for the spec-
trum efficiency improvement while taking the SU’s existence
and needs into account. In CCRNs, STs help PTs with data trans-
mission as relays in exchange for an access to the licensed spec-
trum to transmit their own information. It is shown that CCRN
can improve the capacity and coverage of wireless networks
while improving the spectrum efficiency by relay cooperation
[15]–[17].
1937-9234 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
2130 IEEE SYSTEMS JOURNAL, VOL. 12, NO. 3, SEPTEMBER 2018
Two of the most common relay cooperation strategies in the
CCRN are single-ST and multi-ST. In single-ST, only one ST
acts as PTs relay, and often the ST with the best quality of ser-
vice (QoS) could be selected to improve PUs communication
quality. The selected ST processes PTs signal and, then, retrans-
mits it to the primary receiver (PR, receiver of PU). Compared
with multinode cooperative schemes, single-relay cooperation
(SC) requires neither cooperative beamforming nor distributed
spacetime coding (DSTC), where only the “best” relay out of a
set of candidates participates in the data transmission. In multi-
ST, multiple STs are selected to cooperate with PT. All selected
STs receive the same signal transmitted by PT and retrans-
mit them concurrently to PR. That is to say, in multi-ST, PT
can receive multiple replicas of PTs transmitting signal. It is
observed that multi-ST can combat fading by spatial diversity
[18]. In [19], both single relay and multirelay were consid-
ered in an interweave CRN and it was shown that multirelay
outperforms single relay in terms of outage probability (OP),
while multirelay requires high-complexity symbol-level syn-
chronization. Gomez-Cuba et al. and Han et al. in [18] and
[20] considered single-ST cooperation in a CCRN and analyzed
the outage performance for both PU and SU. Besides, Simeone
et al. investigated a multi-ST cooperation strategy and presented
the achievable rate of PU from the viewpoint of game theory
in [21]. Shalmashi and Ben Slimane proposed a single-ST se-
lection strategy in [22]. Although these aforementioned works
studied the relay cooperation strategies in CCRNs, they did
not consider relay selection and make a comparison between
single-ST and multi-ST cooperation strategies in terms of PU’s
OP. Moreover, as a common puzzle in any wireless communi-
cation system, energy supply is another issue in the CCRN. The
new born technique, simultaneously wireless information and
energy transfer (SWIPT), enables the receiver that uses the same
received signal to decode information while harvesting energy.
Moreover, many recent literatures have studied the applica-
tion of energy harvesting to the overlay CRN. Both information
and energy cooperation in CCRNs were considered in [23]–[25].
The ST in [23] was equipped with multiple antennas and as-
sumed to harvest energy from PT’s transmitting signal. Pradha
et al. proposed an iterative algorithm to achieve the optimal
power allocation while maximizing the primary rate [24]. Yin
et al. considered an ST as an energy harvesting node, which
can harvest energy from RF signals in the ambient environment
[25]. They aimed to maximize the SU’s transmission rate and
derive the closed-form solution. In [26], Chen investigated a
novel energy harvesting technique and considered that nodes
can harvest energy from both received signals and ambient ra-
dio signals simultaneously. In fact, it is only an assumption that
the receiver can decode information and harvest energy in the
same received signal until the work in [27] was released. In [27],
two practical receiver architectures were designed for SWIPT.
Therefore, SWIPT can be implemented in practical scenarios.
Compared with the analysis in [18] and [20], we consider
Nakagami-mfading channels due to that it can model prac-
tical channels more comprehensively including the one-sided
Gaussian distribution (m=0.5), Rayleigh (m=1), and other
distribution (m≥1), which are quite closer to the practical
fading channels than one-sided Gaussian and Rayleigh. Be-
sides, it is shown that Nakagami-mmatches with the data ac-
quired from various practical scenarios [28], [29]. Motivated
by above considerations, in this paper, we investigate a CCRN
with energy harvesting. More specifically, we analyze and com-
pare the performance of both single-ST cooperation (SC) and
multi-ST cooperation (MC) in terms of OP, because OP is an
important performance criterion commonly used to quantize
the quality of communication systems [30]. To the best of our
knowledge, there are few literatures having examined CCRNs
with the SWIPT. Our main contributions can be summarized as
follows.
1) We consider that relays harvest energy from both received
signals and interferers in CCRNs to make the full advan-
tage of SWIPT.
2) We analyze the OP of SC and MC strategies in a CCRN
with energy harvesting over Nakagami-mfading chan-
nels.
3) We derive the exact closed-form expressions of OP for SC
and MC strategies.
4) We compare the outage performance of direct transmis-
sion (DT), SC, and MC. Most importantly, we investigate
the tradeoff between the system performance of PU and
SU in detail.
It is worth noting that our study aims to just analyze and com-
pare the OP for two relay cooperation schemes including SC and
multirelay cooperation (MC) in CCRNs rather than propose re-
lay cooperation schemes (algorithms). Therefore, the complex-
ity of our work is how to derive the close-form OP according
to the probability distribution of more practical Nakagami-m
channel fading instead of time complexity.
This paper is organized as follows. Section II depicts the
cooperative system model. Section III presents system model
and signal analysis of DT. Section IV analyzes detailed relay
cooperation schemes. In Section V, we analyze the outage per-
formance of DT and both relay cooperation schemes with energy
harvesting over Nakagami-mfading channels. Then, Section VI
provides simulation results. Finally, Section VII concludes this
paper.
II. COOPERATIVE SYSTEM MODEL
A. System Model
As sketched in Fig. 1, we consider a CCRN where a primary
network (PN) coexists with a secondary network (SN). PN in-
cludes one PU composed of one transmitter and receiver, while
by contrast SN consists of KSUs, i.e., Ktransmitter–receiver
units. We assume that there is no DT between PT and PR because
of PN’s poor channel condition. Moreover, SUs access the li-
censed spectrum in the overlay mode, that is, STs cooperate with
PT as relays in exchange for an access to the licensed spectrum
after the cooperation. Both information and energy cooperation
between PU and SU are considered. Thus, SUs can gain benefits,
meanwhile, PU’s QoS requirements can be satisfied. Besides,
not all STs of SUs are capable of becoming PT’s relays, and
relays are selected based on certain rules. Besides, we assume
HE et al.: RELAY COOPERATION AND OUTAGE ANALYSIS IN CRNs WITH ENERGY HARVESTING 2131
Fig. 1. System model.
Fig. 2. Phase division.
that all nodes are in half-duplex mode and all channel gains are
modeled as identical and independent Nakagami-mfading.
The key idea of CRNs is to let the SU exploit the licensed
frequency bands, which are already assigned to the PUs, without
degrading the performance of the PUs. In a system with cogni-
tive spectrum access, however, it is difficult for SUs to obtain
reliable and fast spectrum usage information in practice. In our
proposed schemes, therefore, SUs adopt phase-division strategy
to access the licensed bands so as to avoid obtaining reliable and
fast spectrum usage information.
B. Phase Division and Energy Harvesting
Let Φ(|Φ|=K) and Θ(|Θ|=N≤K) denote the set of all
STs in SN and the set of all STs having qualifications for coop-
eration, respectively. Without loss of generality, we assume the
transmission duration to be unity, and divide it into three phases
as shown in Fig. 2. In the first phase αβ (0<α<1), PT trans-
mits its data to satisfied relays (i.e., STs having qualifications
for cooperation). In the second phase α(1 −β)(0<β<1),
all STs selected in the first phase forward data for PR. In the
third phase 1−α, each selected ST transmits its own data to the
corresponding receiver. A perfect phase-division transmission
is assumed and the delay caused by transmission switching can
be ignored. Besides, in the second phase, different STs trans-
mit simultaneously to PR. In the third phase, SUs transmit on
orthogonal channels.
In the system with SWIPT, there remain two key challenges
for practical implementations [27]. First, the receiver is difficult
to observe and extract power simultaneously from the same re-
ceived signal. This is because practical circuits for harvesting
energy from radio signals are not yet able to decode the car-
ried information directly. Second, circuit power consumed by
information decoding is a significant design issue for simulta-
neous information and power transfer, since the circuit power
Fig. 3. Energy harvesting protocol.
reduces the net harvested energy that can be stored in the bat-
tery for future use. To solve the two challenges, Zhou et al.
in [27] studied practical receiver designs for a point-to-point
wireless link with simultaneous information and power transfer,
and proposed a general and practical receiver operation scheme,
namely, dynamic power splitting (DPS). Therefore, to improve
the feasibility of the proposed schemes, we employ DPS to
achieve the simultaneous information and power transfer.
In DPS, each ST can dynamically split the received signals
into two power streams with arbitrary power ratio over time as
shown in Fig. 3 : one (i.e., ρP ) for energy harvesting and the
other one (i.e., (1 −ρ)P) for decoding information contained
in PT’s signal. Moreover, STs not only harvest energy from the
received signal from PT, but also from interferers and RF signals
(xd,d∈[1,D]) in ambient environment [31]. Besides, we set
a harvested energy threshold as Q0and suppose STs to adopt
the decode-and-forward (DF) relaying protocol. This is because
the DF protocol can mitigate noise and interference by fully
decoding, re-encoding, and retransmitting the source messages
to destination.
The channel state information (CSI) exchange can be
achieved by the three-way handshake mechanism [32]. More
specifically, in the first phase, PT transmits RTS message before
data transmission so that ST and PR can estimate the PT–ST and
ST–PR channels and get corresponding channel gains. Then, in
the second phase, ST transmits RTS message to PR so that PR
measures the ST–PR channel and obtain its channel gain. At
last, the measured channel gains can be fed back to PT by CTS
message. Moreover, both instantaneous and statistical CSI are
considered. The statistical signal to noise ratio (SNR) can be
denoted as E(x)when the instantaneous SNR is xwhere E(·)
represents the expected value operator.
III. DIRECT TRANSMISSION
Let us first consider the conventional DT as a benchmark.
Differing from the cooperative system model, the benchmark’s
transmission consists of two parts. In the first phase α(0<α<
1), PT transmits directly to PR, while in the second phase 1−α,
ST transmits its own signal to SR. For simplicity, we assume
the expected values of all transmitting signals is 1. Moreover,
we denote PT’s and STs’ transmission powers as Ppd and Psd,
respectively.
Based on previously declared assumptions, the received sig-
nal at PR and SR in the first and second phases can be expressed,
respectively, as
ypd =Ppdhpdxpd +npd (1)
ysd =Psdhsd xsd +nsd (2)
2132 IEEE SYSTEMS JOURNAL, VOL. 12, NO. 3, SEPTEMBER 2018
where hpd and hsd are channel fading coefficients from PT to
PR and ST to SR, respectively. Meanwhile, npd and nsd are the
identical and independent complex Gaussian noise with zero
mean and variance N0, received at PR and SR, respectively.
Thus, according to (1) and (2), the achievable instantaneous
transmission rate between PT and PR in the first phase, and the
achievable instantaneous transmission rate between ST and SR
in the second phase can be represented, respectively, as
Rpd =αlog2(1 + γpd |hpd|2)(3)
Rsd =(1−α) log2(1 + γsd |hsd|2)(4)
where γpd =Ppd/N0and γsd =Psd/N0denote SNRs on PT–
PR and ST–SR links. Obviously, transmission performance gets
better as SNRs and channel fading coefficients get larger.
Meanwhile, the achievable average transmission rate between
PT and PR in the first phase, and the achievable average trans-
mission rate between ST and SR in the second phase can be
denoted, respectively, as
Rpd =αlog2(1 + γpdE(|hpd |2)) (5)
Rsd =(1−α) log2(1 + γsdE(|hsd|2)).(6)
IV. COOPERATION STRATEGIES IN THE CCRN WITH
ENERGY HARVESTING
Relay cooperation has been proven to be a feasible and
promising way to enhance the QoS of communications [33],
[34]. In this section, we will first introduce the cooperative sig-
nal model from the viewpoint of information theory and, then,
analyze the cooperation criteria and the corresponding trans-
mission rates for both PU and SUs.
A. Cooperative Signal Model
The received signal at the kth (0≤k≤K) ST in the first
phase can be given, respectively, as
y1k=Pphpskxp+
D
d=1 Mdkgdkxd+n1k(7)
where Ppis the transmission power of PT, hpsk is the channel
gain from PT to the kth ST, xpis PT’s transmitting signal,
Dis the number of interferers at ST, Mdk is the transmission
power of the dth interferer, gdk is the channel gain from the
dth interferer to the kth ST, xdis the transmitting signal of the
dth interferer, and n1kis the identical and independent complex
Gaussian noise with zero mean and variance N0, received at the
kth ST.
The received signal at PR from the kth ST in the second phase
can be denoted as
y2k=Pshskpxpk +n2k(8)
where Psis the transmitting power of ST, hskp is the channel
gain from the kth ST to PT, xpk is the transmitting signal of
the kth ST, and n2kis the identical and independent complex
Gaussian noise with zero mean and variance N0, received at PR
from the kth ST.
Moreover, we assume all STs in Φcan decode successfully,
i.e., xpk is the same as xp. Since STs adopt the power splitting
method with a ratio ρ(0<ρ<1) in the first phase, the ob-
servation for the kth ST to harvest energy can be represented
as [35]
yk=ρPphpskxp+√ρ
D
d=1 Mdkgdk xd+√ρn1k.(9)
Thus, the harvested energy can be given by
Ek=αβρη Pp|hpsk |2+
D
d=1
Mdk |gdk|2(10)
where ηdenotes the energy conversion ratio when STs store
the harvested energy into their batteries, αβ represents the time
duration used for energy harvesting because STs only harvest
energy in the first phase, and D
d=1 Mdk |gdk|2arises from in-
terferers received at STs. It is obvious that the harvested energy
is proportional to α,β,ρ, and η.
Note that the energy harvested at ST will only be used as the
relay transmission power, and other energy consumption, like
supporting the transmitter and receiver circuits and information
detection circuits has not been considered in this paper. It is
supposed that STs’ harvested energy is enough for the transmis-
sion from STs to PR. Then, STs in Φwhose harvested energy
are no less than Q0(i.e., Ek≥Q0) will be elected from Φand
put into Θ(|Θ|=N≤K) as a ST having qualifications for
cooperation.
B. Single-Relay Cooperation
In this section, we will consider SC. According to (7), (8), and
the Shannon capacity formula, the instantaneous transmission
rate between PT and the nth ST in the first phase, and the
instantaneous rate between the nth ST and PR in second phase
can be obtained, respectively, as
Rpsn =αβ log21+ γp|hpsn|2
D
d=1 γdn |gdn|2+1(11)
Rsnp =α(1 −β) log2(1 + γs|hsnp |2)(12)
where n∈[1,N],N0is the variance of additive white Gaussian
noise at STs and PT. Besides, γdn =Mdn/N0,γp=Pp/N0,
and γs=Ps/N0denote the SNRs of interferer-ST, PT–ST, and
ST–PR links, respectively.
In the meanwhile, the average transmission rate between PT
and the nth ST in the first phase, and the average rate between
the nth ST and PR in second phase can be deduced, respectively,
as
Rpsn =αβ log21+γpE|hpsn|2
D
d=1 γdn |gdn|2+1(13)
Rsnp =α(1 −β) log2(1 + γsE(|hsnp |2)).(14)
Accordingly, since every cooperative transmission link consists
of two hops, PT–ST and ST–PR, the cooperative rate on the nth
(0≤n≤N) relay link must meet both links’ rate requirements.
HE et al.: RELAY COOPERATION AND OUTAGE ANALYSIS IN CRNs WITH ENERGY HARVESTING 2133
That is to say, the achievable cooperative transmission rate of
PN is dominated by the smaller one, therefore, the achievable
cooperative transmission rate of the nth cooperative link can be
expressed as
Rn
c= min(Rpsn,R
snp)(15)
where Rpsn and Rsnp are the transmission rates of cooperative
transmission link’s first and second hops for SC, respectively.
Since Rpsn increases with βwhile Rsnp decreases with it, there
must exist an optimal βwhich induces the largest cooperative
transmission rate for the nth cooperative transmission link.
Specifically, in SC, PT will choose only one ST as its relay,
namely the best ST. Generally, we regard the one which achieves
the best QoS, i.e., maximum transmission rate, as the best relay.
Hence, the achievable cooperative transmission rate of SC can
be obtained as
RSC
c= max Rn
c.(16)
In the third phase, only the best ST who has played the role
of PT’s relay in the former two phases can transmit its own
information xsto its original receiver. It is assumed that the serial
number of the best relay is b(b∈[1,N]). Then, the transmission
rate for the best SU is
Rb=(1−α) log2(1 + γs|hbb|2)(17)
where γs=Ps/N0denotes the third phase’s SNR and we as-
sume that it is equal to the second phase’s SNR, also |hbb|2is
the channel gain from the best ST to its SR. Thus, the SN’s
achievable transmission rate which indicates the performance
oftheSNinSCis
RSC
s=Rb.(18)
As shown in (17), the term (1 −α)results from the third
phase’s transmission time, and the SU’s performance decreases
with α.
C. Multirelay Cooperation
However, unlike SC, in MC, all STs in Θare selected to
join in the cooperation. Especially, in the MC, a kind of coding
method called DSTC is introduced into the cooperation among
multiple STs to achieve spatial diversity and improve the system
performance [36].
With the consideration that all STs cannot cooperate among
themselves for detection, the instantaneous transmission rate in
the first transmission phase is determinated by the worst channel
gain in Θand it reads [21]
Rps =αβ log21 + min
n∈[1,N ]
γp|hpsn|2
D
d=1 γdn |gdn|2+1.(19)
Similar to SC, αβ,γphpsn , and gdn denote the time duration
allocated for the first phase, the first hop’s SNR, and the channel
coefficients of the cooperative transmission link from PT and
the dth interfere to the nth ST, respectively. Also, Rps increases
with αand β.
Moreover, the statistical transmission rate in the first trans-
mission phase can be computed as
Rps =αβ log21 + min
n∈[1,N ]γpE|hpsn|2
D
d=1 γdn |gdn|2+1.
(20)
On account of the DSTC method employed in the cooperation,
the instantaneous and statistical transmission rate between all
target relays and PR can be obtained, respectively, as [37]
Rsp =α(1 −β) log21+γs
N
n=1 |hsnp|2(21)
Rsp =α(1 −β) log21+γsEN
n=1 |hsnp|2 (22)
where α(1 −β)and N
n=1 |hsnp|2arise from the time division
model and the DSTC method adopted in the cooperation, re-
spectively. Additionally, Rsp increases with αand γswhile it
decreases with β.
Therefore, similar to SC, the achievable cooperative trans-
mission rate in the MC can be expressed as
RMC
c= min(Rps,R
sp)(23)
where the two terms in the min(·)operator represent the first
and second hop’s transmission rates as shown in (19) and (21),
respectively.
Likewise, in the third phase, all STs transmit their own infor-
mation xsto corresponding SRs. Since the transmission is the
typical end-to-end communication, the transmission rate for the
nth SU is
Rnn =(1−α) log21+ γs|hnn|2
γsN
i=n|hin|2+1(24)
where γs=Ps/N0denotes the third phase’s common SNR for
all links’ in SN, and it is equal to the SNR of the second phase.
Also, |hnn|2is the channel gain from the nth ST to its re-
ceiver SR while |hin|2is the noise channel gain from the ith
(i=n)STtonth SR. It is obvious that Rnn decreases as α
increases.
Additionally, the SN’s achievable transmission performance
for MC is set to be the sum of transmission rates of STs
who have played the role of the PT’s relay in the former
two phases. Thus, SN’s achievable transmission rate can be
formulated as
RMS
s=
N
n=1
Rnn.(25)
As shown in (24), the term 1−αoriginates from the third
phase’s transmission time, and the SN’s performance decreases
with α.
V. O P A NALYSIS OVER NAKAGAMI-mFADING CHANNELS
This section will illustrate the OP for SC and MC over
Nakagami-mfading channels. We first derive an exact OP ex-
pression, and based on that, the impact of different parameters
2134 IEEE SYSTEMS JOURNAL, VOL. 12, NO. 3, SEPTEMBER 2018
on the outage performance is then studied. As mentioned in
[38], the commonly used probability density function (pdf) for
Nakagami-mfading can be given by
f(x)=mmxm−1
ΩmΓ(m)exp −mx
Ω(26)
where x=|h|2denotes the channel power gains, Ω=E[x]is
the average power, m=E2[x]/V ar[x]represents the fading
severity parameter. To be specific, Γ(·),E[·]and Var[·]denote
the gamma function, the expected value operator and variance
operator, respectively.
Besides, OP can be defined as the probability that instanta-
neous capacity falls below a predefined rate threshold and it can
be characterized as
Pout =Pr{R<R
p}(27)
where Rdenotes the channel capacity, in other words, the max-
imum transmission rate that can be achieved in this paper, while
RPis the minimum rate requirement. Therefore, once the chan-
nel capacity is less than the minimum rate requirement, the
corresponding transmission link cannot be maintained.
A. Direct Transmission
In this part, we will deduce the OP for DT in detail. Combining
PU’s transmission rate of DT in (3) and the definition of OP in
(27), we can get the OP of PU in DT as
PD
out =Pr{Rpd <R
p}(28)
where Rpd is PU’s achievable transmission rate given by (3),
while Rpis the transmission rate requirement in this paper.
Besides, substituting (3) into (27) yields
Pr{Rpd <R
p}=Pr{|hpd |2<Λpd/γpd }(29)
where Λpd =2
RP/α −1,γpd =Ppd/N0denotes the SNR re-
ceived at PR, and hpd is the channel fading coefficient from PT
to PR.
Let x=|hpd|2,θ=m/Ω=E[x]/V ar[x]and set t=θx,
dx =1
θdt. Thus, (29) can be computed as
Pr{|hpd|2<Λpd/γpd}
=Λpd/γpd
0
θmxm−1
Γ(m)exp(−θx)dx
=θΛpd/γpd
0
θtm−1
Γ(m)exp(−t)1
θdt
=1
Γ(m)θΛpd/γpd
0
tm−1exp(−t)dt. (30)
Accordingly, the definition of incomplete Γfunction is
Γ1(y, z)=z
0
ty−1exp(−t)dt. (31)
Being that, (29) can be further expressed as
Pr{|hpd|2<Λpd/γpd}=1
Γ(m)Γ1(m, θΛpd/γpd ).(32)
Most importantly, if min the Nakagami-mdistribution can
only take natural numbers, the incomplete gamma function in
(32) can be computed as
Γ1(m, θΛpd/γpd )= Γ(m)
⎛
⎝1−e−θΛpd/γpd
m−1
j=0
(θΛpd/γpd)j
j!⎞
⎠.
(33)
Thus, the OP of PU in DT can be further represented as
PD
out =1−e−θΛpd/γpd
m−1
j=0
(θΛpd/γpd)j
j!.(34)
In the following, we analyze the relationship between PD
out and
RPand α. It is obvious that PD
out is an implicit function of α
and RPby the influence of Λpd.Welett=θΛpd /γpd and then
(34) can be rewritten as
PD
out =1−e−t
m−1
j=0
(t)j
j!.(35)
Furthermore, we can get that the derivative of PD
out with respect
to tby
dP D
out
dt =e−ttm−1
(m−1)!.(36)
For all t>0,wehavee−ttm−1
(m−1)! >0, i.e., PD
out increases with
t(t>0). Accordingly, PD
out increases with Λpd as well. There-
fore, PD
out increases with RPand decreases as αincreases since
Λpd =2
RP/α −1.
B. Single-Relay Cooperation
In this part, we will deduce the OP for SC in detail. We first
derive an exact OP expression, and based on that, the impact of
several parameters on the outage performance is then studied.
Combining the achievable cooperative transmission rate of
SC in (16) and the definition of OP in (27), we can get the OP
of PU in SC as
PSC
out =Pr{RSC
c<R
p}(37)
where RSC
cis the achievable cooperative transmission rate of SC
scheme given by (27).
Substituting (16) into (27) yields
PSC
out =Pr{max Rn
c<R
p}
=
N
n=1
Pr{Rn
c<R
p}(38)
where the second equality is derived from the fact that there are
navailable relays which can be selected in total, and max Rn
c<
Rpequals that every Rn
c<R
p(1 ≤n≤N)is true. Then, ac-
cording to the independence of Rn
c<R
p(1 ≤n≤N),the
second equality is deduced.
HE et al.: RELAY COOPERATION AND OUTAGE ANALYSIS IN CRNs WITH ENERGY HARVESTING 2135
Combining (15), (37), (38), and the basic probability com-
puting theory, (37) can be further computed as
PSC
out =
N
n=1
(1 −Pr{Rn
c≥Rp})
=
N
n=1
(1 −Pr{Rpsn ≥Rp}Pr{Rsnp ≥Rp})
=
N
n=1
(1 −(1 −Pr{Rpsn <R
p})(1 −Pr{Rsnp <R
p}))
(39)
where the second equality is derived from the fact that there is
amin(·)operator in the presentation of Rn
c, and then Pr{Rn
c≥
Rp}equals Pr{Rpsn ≥Rp}Pr{Rsnp ≥Rp}.
Besides, substituting (11) into (27) yields
Pr{Rpsn <R
p}=Prγp|hpsn|2
D
d=1 γdn |gdn|2+1 <Λ1(40)
where Λ1=2
RP/(αβ)−1.
For simplicity, we assume Dinterferers have the same effect
on ST, thus D
d=1 γdn |gdn|2=Dγ |g|2. Usually, Dγdn |g|2
1, thus 1 can be neglected. We can rewrite (40) as
Pr{Rpsn <R
p}=Prγp|hpsn|2
Λ1Dγdn |g|2<1.(41)
Moreover, based on the deduced pdf in Appendix A, (41) can
be denoted as
Pr {Rpsn <R
p}
=K1F1(m3+m4),m
3;1+m3;m3Ω2a4
m4Ω3a3(42)
where K1=mm3
3m−m4
4Γ(m3+m4)
(γp)m3(Λ1Dγdn)−m3Ωm3
3Ω−m3
4Γ(m3)Γ(m4),m3=
E2[|hpsn|2]/V ar[|hpsn|2], and m4=E2[|g|2]/V ar[|g|2]are
channel fading severity parameters of PT–ST and ST–PR, re-
spectively. Ω3=E[|hpsn|2]and Ω4=E[|g|2].a3=γpand
a4=Λ
1Dγdn.
In accordance with (32), the OP of the nth relay to the PR
can be given by
Pr{Rsnp <R
p}=Pr{|hsnp |2<Λ2/γs}
=1
Γ(m5)Γ1(m5,θ
nΛ2/γs)(43)
where Λ2=2
Rp/α(1−β)−1,θn=E[|hsnp |2]/V ar[|hsnp |2],
m5is the channel fading severity parameter on ST–PR and
γs=Ps/N0denotes the second hop’s SNR in the cooperative
transmission.
Same as DT, if the value of parameter m5in the Nakagami-
mdistribution can only take a natural number, the incomplete
gamma function in (43) can be represented as (33). Therefore,
the OP of SC in (37) can be deduced as
PSC
out =
N
n=11−exp(−θnΛ2/γs)
m5−1
j=0
(θnΛ2/γs)j
j!
1−K1F1(m3+m4),m
3;1+m3;m3Ω2a4
m4Ω3a3.
(44)
Next, we give the relationship between PSC
out and α,βand Rp.
We can observe from (45) that PSC
out is an implicit function of
phase division ratios (αand β) and Rp, which affect PU’s outage
performance by Λ1,Λ2, and a4, respectively. F1(a, b;c, z)is a
Gaussian hypergeometric function and its partial derivative with
respect to zcan be denoted as
∂F1(a, b;c, z)
∂z =∞
n=0
(a, n)(b, n)
(c, n)
zn−1
(n−1)! (45)
where (a, n)=Γ(a+n)
Γ(a)[39]. Besides, if aonly takes positive
integers, then Γ(a)=(a−1)!. Equation (46) can be rewritten
as
∂F1(a, b;c, z)
∂z =∞
n=0
(a+n)!(b+n)!c!
a!b!(c+n)!
zn−1
(n−1)!.(46)
Therefore, if a,b, and ctake positive integers, then ∂F1(a,b;c,z)
∂z
>0. Since m3+m4,m3, and 1+m3in (45) are positive
integers, F1((m3+m4),m
3;1+m3,m3Ω2a4
m4Ω3a3)increases with
m3Ω2a4
m4Ω3a3where a4=Λ
1Dγdn and Λ1=2
RP/(αβ)−1. There-
fore, F1((m3+m4),m
3;1+m3,m3Ω2a4
m4Ω3a3)increases with RP
and decreases with αand β.
Similarly, we can obtain that the term
e−θnΛ2/γsm5−1
j=0
(θnΛ2/γs)j
j!, where Λ2=2
Rp/α(1−β)−1,
increases with αand βand decreases with Rp. Therefore, PSC
out
increases with RPand decreases with α. In addition, we can
take the partial derivative of PSC
out with respect to βand find
that PSC
out decreases with βwhen βtakes values in the interval
[0,1−RP
αlog2(1+ γs(m5−1)!
θn)], however, PSC
out increases with β
when βtakes values in the interval [1 −RP
αlog2(1+ γs(m5−1)!
θn),1].
C. Multirelay Cooperation
For MC, on the basis of the representation in (27), the OP can
be given by
PMC
out =Pr{RMC
c<R
p}(47)
where RMC
cis the cooperative transmission rate of MC given
by (23), while Rpis the transmission rate threshold like SC.
Furthermore, combining (23) with (45), we can get that
PMC
out =Pr{min(Rps ,R
sp)<R
p}
=(1−Pr{min(Rps,R
sp)≥Rp})
=1−Pr{Rps ≥Rp}Pr{Rsp ≥Rp}
=1−(1 −Pr{Rps <R
p})(1 −Pr{Rsp <R
p})
(48)
2136 IEEE SYSTEMS JOURNAL, VOL. 12, NO. 3, SEPTEMBER 2018
where Rps and Rsp denote the first and second hop’s rate in
the cooperative transmission, respectively, as shown in (19) and
(21). The deduction of (48) is similar to that of (39), where
Pr{min(Rps,R
sp)≥Rp}equals Pr{Rps ≥Rp}Pr{Rsp ≥
Rp}.
Meanwhile, we have
Pr{Rps <R
p}=Prmin
n∈[1,N ]
γp|hpsn|2
D
d=1 γdn |gdn|2+1 <Λ1.
(49)
Similar to the analysis in SC, we can transform
D
d=1 γdn |gdn|2+1into Dγdn |g|2,
Pr{Rps <R
p}=Pr
min
n∈[1,N ]
γp|hpsn|2
Dγdn |g|2<Λ1
=1−Pr min
n∈[1,N ]
γp|hpsn|2
Dγdn |g|2≥Λ1
=1−
N
n=1 Pr γp|hpsn|2
Λ1Dγdn |g|2≥1
=1−
N
n=1 1−Pr γp|hpsn|2
Λ1Dγdn |g|2<1
(50)
where the third equality is derived from the fact that there
are total navailable relays which can be selected, and
minn∈[1,N ]γp|hpsn|2
Dγdn|gdn |2≥Λ1equals that every γp|hpsn|2
Dγdn|gdn |2≥Λ1
is true. Then, according to the independence of γp|hpsn |2
Λ1Dγdn|g|2≥1
(0 ≤n≤N), the third equality is deduced.
According to the deduction in Appendix A, (48) can be further
obtained as
Pr{Rps <R
p}
=1−
N
n=1 1−K1F1(m3+m4),m
3;
×1+m3;m3Ω2a4
m4Ω3a3 (51)
where a3=γpand a4=Λ
1Dγdn. Likewise, another part of
probability representation in (45) can be computed as
Pr{Rsp <R
p}=PrN
n=1 |hsnp|2<Λ2/γs.(52)
For simplicity, all channel fading gains in our model are
supposed to be identical and independent random variables with
the same mean. Similar to [40], we can get that
Pr{Rsp <R
p}=Γ
2Λ2
γsE(|hsnp|2),N.(53)
Thus, the OP of MC can be computed as
PMC
out =1−1−Γ2Λ2
γsE(|hsnp|2),N
N
n=11−K1·F1(m3+m4),m
3;1+m3;m3Ω2a4
m4Ω3a3.(54)
In the following, we analyze the changes of PMC
out with phase
division ratios (αand β) and the achievable rate requirement
(Rp). It is not difficult to observe from (56) that PMC
out is an
implicit function of α,β, and Rp, which take effect on PU’s
outage performance by Λ1,Λ2, and a4. The analysis in SC shows
that F1((m3+m4),m
3;1+m3,m3Ω2a4
m4Ω3a3)increases with RP
and decreases with αand β. Besides, Γ2(a, b)is an incomplete
gamma function, which can be defined as
Γ2(a, b)=a
0
tk−1
Γ(k)exp(−t)dt. (55)
Meanwhile, we take the partial derivative of Γ2(a, b)with re-
spect to aand have
∂Γ2(a, b)
∂a =ak−1exp(−a)
Γ(k).(56)
It is shown that ak−1exp(−a)
Γ(k)>0when a>0and ktakes posi-
tive integers. Thus, Γ2(Λ2
γsE(|hsnp |2),N)increases with Λ2where
Λ2=2
Rp/α(1−β)−1, that is to say, Γ2(Λ2
γsE(|hsnp |2),N)in-
creases with RPand βand decreases with α. Therefore, PMC
out
increases with RPand decreases with α.
Due to F1((m3+m4),m
3;1+m3,m3Ω2a4
m4Ω3a3)and
Γ2(Λ2
γsE(|hsnp |2),N)impose different impacts of βon
PMC
out , we further deduce the partial derivative of PMC
out with
respect to βas
∂PMC
out
∂β =−A21−Γ2Λ2
γsE(|hsnp|2),N
−A1
N
n=1 1−K1F1(m3+m4),m
3;1+m3;m3Ω2a4
m4Ω3a3
(57)
where A1=−
Λ2Rpα(Λ2
γsE(|hsnp |2))N−1exp(−Λ2
γsE(|hsnp |2))ln2
Γ(N)γsE(|hsnp |2)α2(1−β)2and
A2=
N
n=1
[−K1
∞
n=0
(m3+m4,n)(m3,n)
(1+m3,n)
(m3Ω2a4
m4Ω3a3)n−1
(n−1)! ]. It is chal-
lenging to obtain a closed-form solution for dP MC
out
dβ =0.Never-
theless, we can get a numerical solution by computer simula-
tions.
VI. SIMULATION RESULTS AND DISCUSSIONS
In this section, we present simulation results to demonstrate
the performance analysis of DT, SC, and MC over Nakagami-m
fading channels. To be specific, the analytic OP can be obtained
by plotting (34), (45), and (56). In our simulations, all channel
HE et al.: RELAY COOPERATION AND OUTAGE ANALYSIS IN CRNs WITH ENERGY HARVESTING 2137
Fig. 4. OP (45) and (56) versus βwhen m=1,2,3.
fading amplitudes (e.g., |hpsk|2,|hskp|2,|hpsn |2, etc.) are as-
sumed to be Nakagami-mfading with the same expected value
and variance and they are generated randomly. In addition, we
fix other parameters before discussing the exact impact of one
parameter in case of combined effects.
Besides, we plot OP curves for different schemes under vari-
ous parameters, like time division ratio (αand β) and the num-
ber of satisfied STs (N). More specifically, we compare the
outage performance among DT, SC, and MC with different fad-
ing parameters m, harvested energy Q, and the minimum rate
requirement Rp. Moreover, in our simulation, channel fading
coefficients mtakes three values as 1, 2, and 3. The harvested
energy requirement takes two values as Q=40(J)or60(J),
and the rate requirements are set as Rp=0.5(Mb/s)or Rp=
0.6(Mb/s). In fact, harvested energy requirement is a condi-
tion set by STs to decide whether to cooperate with PU or not
and it is the minimum of harvested energy required at STs. The
aim of setting harvested energy requirement is to highlight the
win-to-win characteristics of the overlay CRN and show that
the scenario investigated in this paper not only pay attention
to PUs performance but also SUs performance. In addition, the
first hop’s SNR is set as γp=30 dB, while the second hop’s
SNR is γs=25 dB, unless otherwise stated. Colors are used to
denote different scenarios (DT is green, SC is blue, MC is red).
Fig. 4 illustrates the relation between the OP of single-ST
and multi-ST cooperation schemes and phase division ratio β
under three different fading coefficients (m=1,2,3), respec-
tively, by plotting (45) and (56). We can observe from Fig. 4 that
the OP of SC first decreases, and then, increases with β.Thisis
because the derivative of (45) with respect to βis more than 0
when βtakes values in [0,1−RP
αlog2(1+ γs(m5−1)!
θn)], however, it is
lower than 0 when βtakes values in [1 −RP
αlog2(1+ γs(m5−1)!
θn),1].
Meanwhile, the OP of MC also decreases at first, and then,
increases with βin Fig. 4. Although we cannot obtain a
closed-form solution for dP MC
out
dβ =0, we can get a numerical
solution by computer simulations as shown in Fig. 5. To be
Fig. 5. Numerical solution.
Fig. 6. OP (34), (45), and (56) versus αwhen m=1,2,3.
specific, in Fig. 5, y=−A2(1 −Γ2(Λ2
γsE(|hsnp |2),N)) and z=
A1N
n=1[1 −K1F1((m3+m4),m
3;1+m3;m3Ω2a4
m4Ω3a3)].Itis
shown that there exist two cases of PMC
out ’s monotonicity with
respect to βincluding increasing and decreasing. Furthermore,
the outage performance can be improved with the increasing
of m. It can be seen from Fig. 4 that MC outperforms SC in
low βregion, however, there exists a narrow range of βin
which SC outperforms MC. In practical scenarios, MC requires
configuration with high complexity, especially, in transmission
synchronization. By contrast, SC does not require such compli-
cated settings.
In Fig. 6, we compare DT, SC, and MC’s OP with the chang-
ing of αunder three different fading coefficients (m=1,2,3),
respectively, by plotting (34), (45), and (56). As shown in Fig. 6,
the OP of DT decreases with the increase of α, which is because
the derivative of PD
out with respect to αis always lower than
0, and the OPs of SC and MC decrease with the increasing of
α. The latter is because F1((m3+m4),m
3;1+m3,m3Ω2a4
m4Ω3a3)
2138 IEEE SYSTEMS JOURNAL, VOL. 12, NO. 3, SEPTEMBER 2018
Fig. 7. OP versus the energy harvesting ratio.
and Γ2(Λ2
γsE(|hsnp |2),N)decrease with α. In the meanwhile,
e−θnΛ2/γsm5−1
j=0
(θnΛ2/γs)j
j!increases with α. Moreover, the
OP increases as mincreases, because of the improvement in
combating fading. In low αregion, DT performs better than any
relay cooperation schemes, whereas SC outperforms DT and
MC in high αregion. Nevertheless, SC outperforms DT and
MC in the medium region of α.
In Fig. 7, we present the impact of energy harvesting ratio
ρon the outage performance for both SC and MC by plotting
(45) and (56). It can be seen from Fig. 7 that the OP increases
with Rp, because F1((m3+m4),m
3;1+m3,m3Ω2a4
m4Ω3a3)and
Γ2(Λ2
γsE(|hsnp |2),N)have been proven to increase with Rpand
e−θnΛ2/γsm5−1
j=0
(θnΛ2/γs)j
j!decreases with Rp. More specifi-
cally, we can see that the OPs of both SC and MC decrease as ρ
increases. In addition, it is shown that SC outperforms MC when
Rp=0.5and Q=40and 60, MC outperforms when Rp=0.6
and Q=40. Moreover, when Rp=0.6and Q=60, however,
SC works better than MC in low ρregion, while SC works worse
than MC in high ρregion.
Fig. 8 illustrates the tradeoff between the PU’s and SU’s
performance, i.e., the win-to-win performance and presents the
comparison among DT, SC, and MC. As shown in Fig. 8, SC’s
win-to-win performance is the best, and followed by DT and
MC. In other words, SC has the biggest secondary transmission
rate when achieving the same PU’s OP.
VII. CONCLUSION
In this paper, we consider a CCRN with energy harvesting,
which is composed of one PU and multiple SUs. We study
the outage performance over Nakagami-mfading channels by
deriving closed-form expressions for the exact OP and com-
pare the performance difference between single-ST and multi-
ST cooperation under different parameters. It is illustrated that
SU’s achievable transmission rate decreases with the augment
of PU’s outage performance, and there exists a tradeoff between
Fig. 8. Tradeoff between the PN’s outage performance and the SN’s achiev-
able transmission rate.
PU’s and SU’s system performance. It is shown that SC out-
performs MC and DT in terms of the tradeoff between SU’s
achievable transmission rate and PU’s outage performance, that
is SUs can obtain more benefits in SC than MC and DT. Our fu-
ture work will focus on design of relay selection algorithm and
analysis of secrecy OP in overlay CRN with energy harvest-
ing over Nakagami-mfading channels. Besides, we will further
make a comparison between Nakagami-mand Rayleigh fading
channels.
APPENDIX A
pdf DEDUCTION
The deduction of the pdf of q=a1x1
a2x2.Letxi=|hi|2,
mi/Ωi=E[xi]/V ar[xi]. The pdf of xiover Nakagami-mfad-
ing channel can be given by [41]
fX(xi)=mmi
ixmi−1
i
Ωmi
iΓ(mi)exp −mixi
Ωi.(58)
Thus, the pdf of yi=aixi(aiis a constant) can be deduced
as
fY(yi)= 1
|ai|
mmi
i(yi/ai)mi−1
Ωmi
iΓ(mi)exp −miyi
Ωiai
=mmi
i(yi)mi−1
ami
iΩmi
iΓ(mi)exp −miyi
Ωiai.
According to [41], the pdf of z=y1/y2can be denoted as
fZ(z)= ∞
0
y2
mm2
2(y2)m2−1
am2
2Ωm2
2
exp −miyi
Ωiai
×mm1
1(zy2)m1−1
am1
1Ωm1
1
exp −m1zy2
Ω1a1dy2.(59)
On the basis of [39, eq. (3.381.4)], (64) can be represented as
fZ(z)=Kzm1−1m1Ω2a2
m2Ω1a1
z+1
−(m1+m2)
(60)
HE et al.: RELAY COOPERATION AND OUTAGE ANALYSIS IN CRNs WITH ENERGY HARVESTING 2139
where K=mm1
1m−m2
2Γ(m1+m2)
am1
1a−m1
2Ωm1
1Ω−m1
2Γ(m1)Γ(m2).
With the representation of [39, eq. (3.194.1)], the integral of
fZ(z)with the interval [0,1] can be computed as
I=1
0
fZ(z)dz =KF1(m1+m2),m
1;1+m1;m1Ω2a2
m2Ω1a1
(61)
where F1(·,·;·;·)is Gaussian hypergeometric function [39].
Noted that the integral interval is [0,1].
REFERENCES
[1] J. Wang, M. Ghosh, and K. Challapali, “Emerging cognitive radio ap-
plications: A survey,” IEEE Commun. Mag., vol. 49, no. 3, pp. 74–81,
Mar. 2011.
[2] S. Ghafoor, P. Sutton, C. Sreenan, and K. Brown, “Cognitive radio for dis-
aster response networks: survey, potential, and challenges,” IEEE Wireless
Commun., vol. 21, no. 5, pp. 70–80, Oct. 2014.
[3] B. Wang and K. Liu, “Advances in cognitive radio networks: A survey,”
IEEE J. Sel. Topics Signal Process., vol. 5, no. 1, pp. 5–23, Feb. 2011.
[4] Q. Zhang, B. Cao, Y. Wang, N. Zhang, X. Lin, and L. Sun, “On ex-
ploiting polarization for energy-harvesting enabled cooperative cognitive
radio networking,” IEEE Wireless Commun., vol. 20, no. 4, pp. 116–124,
Aug. 2013.
[5] I. Boufrikech, M. Benjillali, and A. Tamtaoui, “On the performance of
energy detection spectrum sensing for interweave cognitive radio,” in
Proc. 2014 5th Workshop Codes, Cryptography Commun. Syst., Nov. 2014,
pp. 45–49.
[6] F. Yuan, Z. Wang, L. Xiang, and L. Zheng, “Outage performance of
cognitive relay networks using interweave-underlay approach,” in Proc.
2011 2nd Int. Conf. Digital Manuf. Autom., Aug. 2011, pp. 833–836.
[7] L. Lopez-Ramos, A. Marques, and J. Ramos, “Jointly optimal sensing and
resource allocation for multiuser interweave cognitive radios,” in Proc.
2012 7th Int. ICST Conf. Cognitive Radio Oriented Wireless Netw. Com-
mun., Jun. 2012, pp. 373–378.
[8] K. Ho-Van, P. Sofotasios, G. Alexandropoulos, and S. Freear, “Bit error
rate of underlay decode-and-forward cognitive networks with best relay
selection,” J. Commun. Netw., vol. 17, no. 2, pp. 162–171, Apr. 2015.
[9] M. Seyfi, S. Muhaidat, and J. Liang, “Relay selection in underlay cognitive
radio networks,” in Proc. 2012 IEEE Wireless Commun. Netw. Conf.,
Apr. 2012, pp. 283–288.
[10] T. T. Duy and H. Y. Kong, “On performance evaluation of hybrid decode-
amplify-forward relaying protocol with partial relay selection in underlay
cognitive networks,” J. Commun. Netw., vol. 16, no. 5, pp. 502–511,
Oct. 2014.
[11] M. Hasna and M.-S. Alouini, “End-to-end performance of transmission
systems with relays over Rayleigh-fading channels,” IEEE Trans. Wireless
Commun., vol. 2, no. 6, pp. 1126–1131, Nov. 2003.
[12] R. Pabst et al., “Relay-based deployment concepts for wireless and mo-
bile broadband radio,” IEEE Commun. Mag., vol. 42, no. 9, pp. 80–89,
Sep. 2004.
[13] H. Suraweera, P. Smith, and N. Surobhi, “Exact outage probability of
cooperative diversity with opportunistic spectrum access,” in Proc. IEEE
Int. Conf. Commun. Workshops, May 2008, pp. 79–84.
[14] J. Lee, H. Wang, J. Andrews, and D. Hong, “Outage probability of cog-
nitive relay networks with interference constraints,” IEEE Trans. Wireless
Commun., vol. 10, no. 2, pp. 390–395, Feb. 2011.
[15] Y. Zou, Y.-D. Yao, and B. Zheng, “Outage probability analysis of cog-
nitive transmissions: Impact of spectrum sensing overhead,” IEEE Trans.
Wireless Commun., vol. 9, no. 8, pp. 2676–2688, Aug. 2010.
[16] M. Vajapeyam and U. Mitra, “Performance analysis of distributed space-
time coded protocols for wireless multi-hop communications,” IEEE
Trans. Wireless Commun., vol. 9, no. 1, pp. 122–133, Jan. 2010.
[17] Y. Zou, J. Zhu, B. Zheng, and Y.-D. Yao, “An adaptive cooperation diver-
sity scheme with best-relay selection in cognitive radio networks,” IEEE
Trans. Signal Process., vol. 58, no. 10, pp. 5438–5445, Oct. 2010.
[18] F. Gomez-Cuba, R. Asorey-Cacheda, F. Gonzalez-Castano, and H. Huang,
“Application of cooperative diversity to cognitive radio leasing: Model
and analytical characterization of resource gains,” IEEE Trans. Wireless
Commun., vol. 12, no. 1, pp. 40–49, Jan. 2013.
[19] Y. Zou, B. Champagne, W. P. Zhu, and L. Hanzo, “Relay-selection im-
proves the security-reliability trade-off in cognitive radio systems,” IEEE
Trans. Commun., vol. 63, no. 1, pp. 215–228, Jan. 2015.
[20] Y. Han, A. Pandharipande, and S. H. Ting, “Cooperative decode-and-
forward relaying for secondary spectrum access,” IEEE Trans. Wireless
Commun., vol. 8, no. 10, pp. 4945–4950, Oct. 2009.
[21] O. Simeone, I. Stanojev, S. Savazzi, Y. Bar-Ness, U. Spagnolini, and
R. Pickholtz, “Spectrum leasing to cooperating secondary ad hoc net-
works,” IEEE J. Sel. Areas Commun., vol. 26, no. 1, pp. 203–213,
Jan. 2008.
[22] S. Shalmashi and S. Ben Slimane, “Performance analysis of relay-
assisted cognitive radio systems with superposition coding,” in Proc. 2012
IEEE 23rd Int. Symp. Pers. Indoor Mobile Radio Commun., Sep. 2012,
pp. 1226–1231.
[23] G. Zheng, Z. Ho, E. A. Jorswieck, and B. Ottersten, “Information and
energy cooperation in cognitive radio networks,” IEEE Trans. Signal Pro-
cess., vol. 62, no. 9, pp. 2290–2303, May 2014.
[24] J. J. Pradha, S. S. Kalamkar, and A. Banerjee, “On information and energy
cooperation in energy harvesting cognitive radio,” in Proc. 2015 IEEE
26th Annu. Int. Symp. Pers., Indoor, Mobile Radio Commun., Aug. 2015,
pp. 943–948.
[25] S. Yin, E. Zhang, Z. Qu, L. Yin, and S. Li, “Optimal cooperation strategy
in cognitive radio systems with energy harvesting,” IEEE Trans. Wireless
Commun., vol. 13, no. 9, pp. 4693–4707, Sep. 2014.
[26] Y. Chen, “Energy-harvesting AF relaying in the presence of interference
and Nakagami-mfading,” IEEE Trans. Wireless Commun., vol. 15, no. 2,
pp. 1008–1017, Feb. 2016.
[27] X. Zhou, R. Zhang, and C. K. Ho, “Wireless information and power trans-
fer: Architecture design and rate-energy tradeoff,” IEEE Trans. Commun.,
vol. 61, no. 11, pp. 4754–4767, Nov. 2013.
[28] O. El Bashir and M. Abdel-Hafez, “Opportunistic spectrum access using
collaborative sensing in Nakagami-mfading channel with real fading
parameter,” in Proc. 2011 7th Int. Wireless Commun. Mobile Comput.
Conf., Jul. 2011, pp. 472–476.
[29] A. Lodhi, F. Said, M. Dohler, and A. H. Aghvami, “Closed-form sym-
bol error probabilities of STBC and CDD MC-CDMA with frequency-
correlated subcarriers over Nakagami-mfading channels,” IEEE Trans.
Veh. Technol., vol. 57, no. 2, pp. 962–973, Mar. 2008.
[30] M. Hankal, I. Eshrah, and H. Tawfik, “Performance of the dual-hop decode
and forward relaying systems over Nakagami-mfading channels,” in Proc.
2014 31st Nat. Radio Sci. Conf., Apr. 2014, pp. 219–227.
[31] Y. Chen, “Energy-harvesting af relaying in the presence of interference
and Nakagami-mfading,” IEEE Trans. Wireless Commun., vol. 15, no. 2,
pp. 1008–1017, Feb. 2016.
[32] S. Guo, F. Wang, Y. Yang, and B. Xiao, “Energy-efficient cooperative for
simultaneous wireless information and power transfer in clustered wireless
sensor networks,” IEEE Trans. Commun., vol. 63, no. 11, pp. 4405–4417,
Nov. 2015.
[33] J. Laneman, D. Tse, and G. W. Wornell, “Cooperative diversity in wire-
less networks: Efficient protocols and outage behavior,” IEEE Trans. Inf.
Theory, vol. 50, no. 12, pp. 3062–3080, Dec. 2004.
[34] J. Laneman and G. W. Wornell, “Distributed space-time coded protocols
for exploiting cooperative diversity in wireless networks,” in Proc. Global
Telecommun. Conf, Nov. 2002, vol. 1, pp. 77–81.
[35] P. Grover and A. Sahai, “Shannon meets tesla: Wireless information and
power transfer,” in Proc. 2010 IEEE Int. Symp. Inf. Theory, Jun. 2010,
pp. 2363–2367.
[36] J. Mietzner and P. Hoeher, “Improving the performance of mobile broad-
casting systems using multiple base stations and distributed space-time
codes,” IET Commun., vol. 1, no. 3, pp. 348–353, Jun. 2007.
[37] G. Foschini and J. Salz, “Digital communications over fading radio chan-
nels,” Bell Syst. Tech. J., vol. 62, no. 2, pp. 429–456, Feb. 1983.
[38] P. Sharma, S. Solanki, and P. Upadhyay, “Outage analysis of cognitive
opportunistic relay networks with direct link Nakagami-mfading,” IEEE
Commun. Lett., vol. 19, no. 5, pp. 875–878, May 2015.
[39] D. Zwillinger, Table of Integrals, Series, and Products.NewYork,NY,
USA: Elsevier, 2014.
[40] Y. Zou, B. Champagne, W.-P. Zhu, and L. Hanzo, “Relay-selection im-
proves the security-reliability trade-off in cognitive radio systems,” IEEE
Trans. Commun., vol. 63, no. 1, pp. 215–228, Jan. 2015.
[41] C. Tang, G. Pan, and T. Li, “Secrecy outage analysis of underlay cognitive
radio unit over Nakagami-mfading channels,” IEEE Wireless Commun.
Lett., vol. 3, no. 6, pp. 609–612, Dec. 2014.
2140 IEEE SYSTEMS JOURNAL, VOL. 12, NO. 3, SEPTEMBER 2018
Jing He received the B.S. degree from the College of
Electronics and Information Engineering, Southwest
University, Chongqing, China, in 2014, where she is
currently working toward the Ph.D. degree.
Her research interests include wireless energy
transfer and outage probability analysis in cognitive
radio networks.
Songtao Guo (A’10–M’12) received the B.S., M.S.,
and Ph.D. degrees in computer software and theory
from Chongqing University, Chongqing, China, in
1999, 2003, and 2008, respectively.
He was a Professor from 2011 to 2012 with
Chongqing University. He is currently a Professor
with Southwest University, Chongqing. He has au-
thored or co-authored more than 80 scientific papers
in leading refereed journals and conferences. His re-
search interests include wireless sensor networks,
ad hoc networks, and parallel and distributed
computing.
Dr. Guo was the recipient of many research grants from the National Science
Foundation of China and Chongqing and the Postdoctoral Science Foundation
of China.
Pan Gaofeng (M’12) received the B.Sc. degree in
communication engineering from Zhengzhou Uni-
versity, Zhengzhou, China, in 2005, and the Ph.D. de-
gree in communication and information systems from
Southwest Jiaotong University, Chengdu, China, in
2011.
He was with The Ohio State University, Colum-
bus, OH, USA, from September 2009 to Septem-
ber 2011, as a Joint-Trained Ph.D. student. In May
2012, he joined the School of Electronic and Informa-
tion Engineering, Southwest University, Chongqing,
China, where he is currently an Associate Professor. Since January 2016, he has
been with the School of Computing and Communications, Lancaster University,
Lancaster, U.K., where he is a Postdoctoral Researcher. His research interests
include special topics in communications theory, signal processing and proto-
col design, including secure communications, CR/cooperative communications,
and MAC protocols.
Yuanyuan Yang (S’91–M’92–SM’98–F’09) re-
ceived the B.Eng. and M.S. degrees in computer
science and engineering from Tsinghua University,
Beijing, China, and the M.S.E. and Ph.D. degrees in
computer science from The Johns Hopkins Univer-
sity, Baltimore, MD, USA.
She is a currently a Professor of computer en-
gineering and computer science with Stony Brook
University, New York, NY, USA. She has authored
or co-authored more than 300 papers in major jour-
nals and refereed conference proceedings and holds
7 U.S. patents in these areas. Her research interests include wireless networks,
data center networks, optical networks, and high-speed networks.
Dr. Yang has served as an Associate Editor-in-Chief and an Associate Editor
for the IEEE TRANSACTIONS ON COMPUTERS and as an Associate Editor for the
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS.
Defang Liu received the B.S., M.S., and Ph.D. de-
grees in bioengineering from Chongqing Univer-
sity, Chongqing, China, in 1999, 2005, and 2013,
respectively.
She is currently a Lecturer with Southwest Uni-
versity, Chongqing. Her research interests include re-
source allocation in wireless networks, cooperative
communications, mathematical modeling of infec-
tious diseases, and application of surfaces modified
of biosensors.