Vous êtes sur la page 1sur 14

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

Artificial Noise Aided Secure Cognitive


Beamforming for Cooperative MISO-NOMA Using
SWIPT
Fuhui Zhou, Member, IEEE, Zheng Chu, Member, IEEE, Haijian Sun, Student Member, IEEE,
Rose Qingyang Hu, Senior Member, IEEE, and Lajos Hanzo, Fellow, IEEE

Abstract—Cognitive radio (CR) and non-orthogonal multiple I. I NTRODUCTION


access (NOMA) have been deemed two promising technologies
due to their potential to achieve high spectral efficiency and HE next generation wireless communication systems call
massive connectivity. This paper studies a multiple-input single-
output NOMA CR network relying on simultaneous wireless
T for advanced communication techniques that can achieve
high spectral efficiency (SE) and provide massive connectiv-
information and power transfer (SWIPT) conceived for sup-
porting a massive population of power limited battery-driven ity in support of the escalating high data rate requirements
devices. In contrast to most of the existing works, which use an imposed by the unprecedented proliferation of mobile devices
ideally linear energy harvesting model, this study applies a more [1]. Cognitive radio (CR) and non-orthogonal multiple access
practical non-linear energy harvesting model. In order to improve (NOMA) constitute promising techniques of achieving high
the security of the primary network, an artificial-noise-aided
cooperative jamming scheme is proposed. The artificial-noise- SE [2]-[4]. Specifically, CR enables the secondary users (SUs)
aided beamforming design problems are investigated subject to to exploit the frequency bands of the primary users (PUs)
the practical secrecy rate and energy harvesting constraints. provided that the interference imposed on the PUs from the
Specifically, the transmission power minimization problems are SUs is below a certain level. NOMA has a higher information-
formulated under both perfect channel state information (CSI) theoretic rate region than orthogonal techniques albeit, which
and the bounded CSI error model. The problems formulated
are non-convex, hence they are challenging to solve. A pair of is achieved by increasing the receiver’s implementation com-
algorithms either using semidefinite relaxation (SDR) or a cost plexity [4]. One of the main ideas for realizing NOMA is to
function are proposed for solving these problems. Our simulation exploit the power domain. Specifically, multiple users’ signals
results show that the proposed cooperative jamming scheme are superimposed by using different power levels and succes-
succeeds in establishing secure communications and NOMA is sive interference cancellation (SIC) is installed at the receiver
capable of outperforming the conventional orthogonal multiple
access in terms of its power efficiency. Finally, we demonstrate for mitigating the mutual interference imposed by using non-
that the cost function algorithm outperforms the SDR-based orthogonal resources [5]. It is envisioned that applying NOMA
algorithm. in CR networks (CRNs) is capable of significantly improving
the SE and the user connectivity [6], [7].
Index Terms—Cognitive radio, non-orthogonal multiple access,
non-linear energy harvesting, physical-layer secrecy. Meanwhile, the next generation wireless communication
systems also need energy-efficient techniques due to the ever-
increasing greenhouse gas emission concerns and explosive
Manuscript received September 15, 2017; revised Feb. 1, 2018 and accepted proliferation of power-limited devices, e.g., sensors and mo-
February 15, 2018. Date of publication ****; date of current version ****. The bile phones. Energy-efficient techniques can be divided into
research of F. Zhou was supported in part by the Natural Science Foundation
of China under Grant 61701214, in part by the Young Natural Science
two broad categories. One of the categories focuses on the
Foundation of Jiangxi Province under Grant 20171BAB212002, in part by techniques that can achieve high energy efficiency (EE) [8],
The Open Foundation of The State Key Laboratory of Integrated Services [9], while the other one aims for recycling energy, where both
Networks under Grant ISN19-08, and in part by The Postdoctoral Science
Foundation of Jiangxi Province under Grant 2017M610400, Grant 2017KY04
wireless charging as well as simultaneous wireless information
and Grant 2017RC17. The research of Prof. R. Q. Hu was supported in part and power transfer (SWIPT) [10] fit. In this paper, we focus
by the US National Science Foundation grants ECCS-1308006 and EARS- on SWIPT since it can simultaneously transmit information
1547312. The research of Prof. L. Hanzo was supported by the ERC for his
Advanced Fellow Grant and the Royal Society for his Research Merit Award.
and achieve energy harvesting (EH). In SWIPT, the radio
F. Zhou is with the Department of Electrical and Computer Engineering frequency (RF) signals carry not only information to the users,
as a Research Fellow at Utah State University, U.S.A. F. Zhou is also with but also transfer energy for the energy harvesting receivers
the School of Information Engineering, Nanchang University, P. R. China,
330031. He is also with State Key Laboratory of Integrated Services Networks,
(EHRs). Compared to the conventional EH techniques, such
Xidian University, Xian, 710071, P. R. China (e-mail: zhoufuhui@ieee.org). as wind charging, SWIPT has an advantage in providing more
Z. Chu is with 5G Innovation Center (5GIC), Institute for Communication stable and controllable amount of power for energy-limited
Systems (ICS), University of Surrey, Guildford GU2 7XH, U.K (e-mail:
z.chu@mdx.ac.uk).
devices. Hence, it is of significant importance to study the
H. Sun and R. Q. Hu are with Electrical and Computer Engineer- application of SWIPT in NOMA CRNs that aim for supporting
ing Department, Utah State University, USA (e-mail: h.j.sun@ieee.org, massive population of battery driven power-limited devices.
rose.hu@usu.edu).
L. Hanzo is with the University of Southampton, Southampton SO17 1BJ, However, due to the broadcast nature of NOMA as well as
U.K. (e-mail: lh@ecs.soton.ac.uk). CR and the dual function of RF signals [11], [12], NOMA

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

CRNs relying on SWIPT are vulnerable to eavesdropping. shown that as anticipated the secrecy rate of the SUs can be
Malicious EHRs may intercept the confidential information significantly improved by using multiple antennas techniques,
transmitted to the PUs and the SUs [13]. Thus, it is vital by contrast it is reduced when the CSI inaccuracy goes up.
to improve the security of NOMA CRNs using SWIPT. By exploiting the relationship between multi-antenna aided
As an alternative to the traditional cryptographic techniques, secure communications and cognitive radio communications,
physical-layer security exploits the physical characteristics the authors of [32] designed an optimal beamforming scheme
(e.g., multipath fading, propagation delay, etc.) of wireless for MISO-aided CRNs. In [33], the authors extended the con-
channels to achieve secure communications [16]-[17]. It was tributions of [31] and [32] into a fading channel and the secure
shown [16]-[17] that the secrecy rate of wireless communica- throughput was maximized by optimizing both the beam-
tion systems directly depends on the accuracy of the channel forming vector and the transmission power. The authors of
state information (CSI). Moreover, the secrecy rate of SUs [34] studied the robust beamforming design problem in MISO
in CRNs is more severely limited [13], [18]-[22] since their CRNs where realistic finite-alphabet inputs are considered. A
transmission power should be controlled in order to protect global optimization approach was proposed for designing an
the PUs’ quality of service. In order to improve the secrecy optimal beamforming scheme for maximizing the secrecy rate.
rate of SUs, multiple antennas, cooperative relaying, jamming Recently, the authors of [35] and [36] studied the beamforming
and artificial noise (AN)-aided techniques have been applied design problems of secure MISO multiuser unicast CRNs and
[18]-[22]. Moreover, the secrecy rate can be further improved of mutlicast CRNs, respectively. Specifically, in [35], an AN-
by designing an optimal resource allocation scheme [18]-[22]. aided beamforming scheme was proposed. It was shown that as
Furthermore, the secure energy efficiency can be enhanced by expected the secrecy rate of SUs can be improved by imposing
using AN-aided techniques and designing the optimal resource artificial noise on malicious SUs. Cooperation between the
allocation schemes [23], [24]. However, the performance gains primary network and the secondary network was proposed in
achieved by using these techniques are significantly influenced [36] where the secrecy rate of SUs was maximized under the
by the accuracy of CSI. What’s worse, it is a challenge to max-min fairness criterion.
obtain accurate CSI, especially for NOMA [25], [26]. Thus, it Since energy harvesting has not been considered in [31]-
is important to design resource allocation schemes under the [36], the beamforming schemes proposed in these works are
imperfect CSI. inappropriate in CRNs using SWIPT. Recently, the authors of
Numerous investigations have been conducted for improv- [13], [22], [38]-[41] studied the resource allocation problems
ing the security of the conventional OMA systems and efforts of various CRNs using SWIPT. In [13], a multi-objective
have been invested into conceiving secure NOMA systems optimization framework was applied in MISO CRNs with
[12], [27]-[30]. However, no contributions have been devoted SWIPT. The beamforming scheme, the covariance matrix of
to improving the security of NOMA CRNs using SWIPT. AN and energy signals were jointly optimized. It was shown
In this paper, in order to achieve secure communications, that there are several tradeoffs in CRNs using SWIPT, such as
beamforming design problems are studied in multiple-input the tradeoff between the secrecy rate of SUs and the harvested
single-output (MISO) NOMA CRNs using SWIPT where a power of EHRs. The authors of [13] only considered the
practical non-linear EH model is applied as well as different bounded CSI error model. In [22], the authors studied the
CSI models are considered. An AN-aided cooperative scheme robust beamforming design problem both under the bounded
is proposed for improving the security of the primary network. CSI error model and the probabilistic CSI error model. It was
By using this scheme, the secondary network imposes artificial shown that a performance gain can be obtained under the
noise for jamming the malicious EHRs while aspiring to get a probabilistic CSI error model compared to the bounded CSI
chance to access the frequency bands of the primary network. error model. Mohjazi et al. [37] extended the robust beam-
The related work and the motivation of our investigation are forming design problem into a multi-user MISO CRNs using
presented as follows. SWIPT. The transmission power of the cognitive base station
(CBS) was minimized by jointly optimizing the beamforming
of CBS and the power splitting factor of the energy-harvesting
A. Related Work and Motivation SUs. In order to further improve the secrecy rate and the
Beamforming design problems have been extensively stud- harvested power of EHRs, an optimal precoding scheme was
ied both in conventional CRNs [31]-[36] and in conventional designed for multiple-input multiple-output (MIMO) aided
CRNs using SWIPT [13], [22], [38]-[41]. Recently, some CRNs using SWIPT [38]. In [39], a cooperative mechanism
efforts have also been dedicated to designing NOMA resource and a robust beamforming scheme were proposed for im-
allocation schemes for improving their security [12], [27]-[30]. proving the security of CRNs, where the energy signals were
These contributions can be summarized as follows. exploited to jam the malicious EHRs. The authors of [40] have
Due to the broadcast nature of the conventional CRNs, studied robust resource allocation problems in MIMO-aided
malicious SUs may intercept the confidential information CRNs using SWIPT under the probabilistic CSI error model.
transmitted to the legitimate SUs. In order to improve the The contributions of [13], [22], [38]-[40] assumed an ideal
security of CRNs, numerous secure physical-layer techniques linear EH model. However, practical power conversion circuits
have been proposed by using different CSI models [31]-[36]. have a non-linear end-to-end wireless power transfer function.
In [31], a robust beamforming scheme has been proposed for Hence, the robust resource allocation schemes proposed in
MISO CRNs in the face of a bounded CSI error model. It was these treatises would perform difficultly in the face of a

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

realistic non-linear power transfer characteristic. In [41], the 1) The AN-aided cooperative scheme is proposed for
robust beamforming design problem was studied in MISO MISO-NOMA CRNs using SWIPT in order to improve
CRNs using SWIPT, where a non-linear EH model was used. the security of the primary network. By using this
The above-mentioned contributions were made for CRNs scheme, the CBS transmits a jamming signal to cooper-
and CRNs with SWIPT where OMA is applied. However, ate with the primary base station (PBS) for improving
these resource allocation schemes proposed in the above- the security of the PUs. As a reward, the secondary
mentioned works are inappropriate or suboptimal in NOMA network is granted to access the frequency bands of the
systems since NOMA schemes are very different from OMA. primary network and provide SWIPT services both for
The authors of [12], [27]-[30] have studied the optimal re- the SUs and for the EHRs in the secondary network.
source allocation problems in NOMA systems in order to Moreover, the covariance matrix of the jamming signals
achieve secure communications. In [12], an optimal power transmitted at CBS and the beamforming of the CBS
allocation scheme was proposed for maximizing the secrecy and the PBS are jointly optimized.
sum rate of a single-input single-output (SISO) NOMA sys- 2) Beamforming design problems are studied under both
tem, where only an eavesdropper was assumed and a constant the perfect CSI model and the bounded CSI error model.
decoding order was applied. In [27], the authors considered a In contrast to the works that only an eavesdropper
more general scenario, where a dynamic decoding order was was considered in the NOMA system [12], [27]-[30],
considered. The sum secrecy rate was maximized by jointly we investigate a more general scenario, where multiple
optimizing the decoding order, the transmission rates and malicious EHRs exist. The total transmission power
the power allocated to users. The secrecy rate maximization is minimized by jointly optimizing the transmission
problems of MISO NOMA systems [28], [29] and MIMO beamforming vectors of both the PBS and the CBS as
NOMA systems [30] were investigated. It was shown that well as the covariance matrix of the jamming signal
the secrecy rate achieved by using NOMA is higher than that transmitted at the CBS, subject to constraints on the
achieved by using OMA, and that the secrecy rate of users can secrecy rates of both the PUs and the SUs as well as
be improved by using multiple antennas-aided techniques. on the energy harvesting requirements of the EHRs.
Although resource allocation problems have indeed been A pair of algorithms are proposed for solving these
conceived for NOMA systems for achieving secure communi- challenging non-convex problems. One of them relies on
cations [12], [27]-[30], resource allocation schemes proposed semidefinite relaxation (SDR) while the other is based
in these contributions operated under the assumption that per- on a carefully conceived cost function.
fect CSI can be obtained. Moreover, these resource allocation 3) Our simulation results show that the proposed AN-
schemes cannot work in NOMA CRNs using SWIPT since the aided cooperative scheme can reduce the transmission
interference between the primary network and the secondary power required in MISO-NOMA CRNs using SWIPT.
network as well as the energy harvesting requirements of Moreover, it is shown that the performance achieved
the EHRs are required to be considered. Furthermore, the by NOMA is proven to be better than that obtained by
robust resource allocation schemes proposed in conventional OMA, even when the CSI is imperfect. Furthermore, our
CRNs using SWIPT are inappropriate for NOMA CRNs using simulation results also show that the algorithm based on
SWIPT due to the differences between NOMA and OMA. the cost function outperforms the algorithm based on
To the best of our knowledge, few investigations have been using SDR.
conduced for improving the security of NOMA CRNs using
SWIPT. Thus, in order to achieve secure communications in
NOMA CRNs using SWIPT, beamforming design problems C. Organization and Notations
are studied both under the perfect CSI model and the bounded The remainder of this paper is organized as follows. The
CSI error model. These problems are challenging but mean- system model is presented in Section II. Our secure beam-
ingful. The reasons are from the following two perspectives. forming design problems are examined under the perfect CSI
On the one hand, a practical non-linear EH model is applied, assumption in Section III. Section IV presents our secure
but the EH form is more complex than the linear form. On beamforming design problems under the bounded CSI error
the other hand, the mutual interference between the primary model while our simulation results are presented in Section V.
network and the secondary network as well as the interference Finally, the paper is concluded in Section VI.
among NOMA SUs have to be considered. Notations: Vectors and matrices are represented by boldface
lower case letters and boldface capital letters, respectively. The
B. Contributions and Organization identity matrix is denoted by I; NP,t and NS,t are the number
In contrast to [12], [27]-[30], this paper studies the beam- of antennas of the PBS and the CBS, respectively; vec(A)
forming design problems of MISO-NOMA CRNs using denotes the vectorization of matrix A and it is obtained by
SWIPT, where multiple malicious EHRs exist and a practical stacking its column vectors. The Hermitian (conjugate) trans-
non-linear EH model is applied. Both the perfect CSI and the pose, trace, and rank of a matrix A are denoted respectively by
bounded CSI error model are considered. In order to improve AH , Tr(A) and Rank(A). x† represents the conjugate trans-
the security of the primary network, an AN-aided cooperative pose of a vector x. CM ×N stands for an M -by-N dimensional
scheme is proposed. The main contributions are summarized complex matrix set. A ≽ 0 (A ≻ 0) represents that A is
as follows: a Hermitian positive semidefinite (definite) matrix. HN and

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

clusters and simultaneously transfers energy to EHRs. In the


secondary network, the CBS provides SWIPT service to Ks
EHRs and to Ns SUs by using NOMA. Due to the constrained
size of devices, the PUs and SUs can only perform information
decoding while the EHRs can only harvest energy from the
RF signals [22], [35]. The primary network coexists with the
secondary network by using the spectrum sharing mode. The
PBS is equipped with Np,t antennas and the CBS is equipped
with Ns,t antennas. All the PUs, SUs and EHRs are equipped
with a single antenna.
Due to the broadcast natures of NOMA and the dual
function of RF signals in SWIPT, the EHR may eavesdrop
and intercept the information transmitted by the PBS and
the CBS. It is assumed that EHRs in each network can only
intercept confidential information from the same network and
the PUs in each cluster are respectively wiretapped by EHRs
in the same cluster [36]. For example, PUs in the mth cluster,

where m ∈ M and M = {1, 2, · · · , M }, are wiretapped
Fig. 1: The system model. by the kth EHR in the mth cluster, where k ∈ M and

Km = {1, 2, · · · , Km } and Km is the number of EHRs
while Np,m is the number of PUs in the mth cluster. In
HN+ represent a N -by-N dimensional Hermitian matrix set order to improve the security of both the primary network
and a Hermitian positive semidefinite matrix set, respectively. and the secondary network, an AN-aided cooperative scheme
∥·∥ denotes the Euclidean norm of a vector. |·| represents is applied. Using this scheme, the CBS of Fig. 1 transmits
the absolute value of a complex scalar. x ∼ CN (u, Σ) a jamming signal to the primary network for improving the
means that x is a random vector, which follows a complex security of the PUs. As a reward, the primary network allows
Gaussian distribution with mean u and covariance matrix Σ. the secondary network to operate on its frequency bands. All
E[·] denotes the expectation operator. Re (a) extracts the real the channels involved are assumed to be flat fading channels.
part of vector a. λmax (A) is the maximum eigenvalue of A. In this paper, both the perfect CSI and imperfect CSI cases are
R+ represents the set of all nonnegative real numbers. [x]+ studied. The performance achieved under the perfect CSI can
denotes the maximum between 0 and x. be used as a bound in our analysis and provides meaningful
insights into the design of MISO NOMA CRNs using SWIPT.
II. S YSTEM M ODEL The assumption has also been used in [11], [12], [18], [19].
In this section, we will describe the network model and
security metrics in the downlink MISO NOMA CRNs using B. Security Metrics
SWIPT under a practical non-linear energy harvesting model.
In [12], [27]-[30], only one eavesdropper has been consid- Let yp,m,i denote the signal received at the ith PU in the
ered in the designed NOMA systems and resource allocation mth cluster, ys,j represent the signal received at the jth SU,
schemes have been proposed. In this paper, the beamforming ye,m,k denote the EH signal received at the kth EHR in the
design problems are studied in a more general scenario, mth cluster and ye,l represent the EH signal received at the lth
where multiple malicious EHRs exist. The detail description EHR in the secondary network, respectively, where i ∈ Np,m ,
is presented in the following subsections. Np,m = {1, 2, · · · , Np,m }; j ∈ Ns , Ns = {1, 2, · · · , Ns } and
l ∈ Ks , Ks = {1, 2, · · · , Ks }. These signals are respectively
expressed as
A. Network Model [ M ]


Our downlink MISO NOMA CR network using SWIPT is yp,m,i =hp,m,i (wp,m sp,m + vp,m )
shown in Fig. 1. In the primary network, unicast-multicast
communications are exploited since they can provide high SE 
m=1


Ns
and massive connectivity. This scenario is widely encountered, †
+ fs,m,i  ws,j ss,j + vs  + np,m,i , (1a)
for example in Internet of Things, wireless sensor networks j=1
and the cellular networks [35], [36]. Specifically, the PB- [ ]
S sends different confidential information-bearing signals to ∑
M
ys,j =q†p,j (wp,m sp,m + vp,m )
the PUs in the different clusters. And the primary users in
each individual multicast cluster receive the same confidential 
m=1

information-bearing signal from the PBS. In the secondary ∑
Ns
+ h†s,j  ws,j ss,j + vs  + ns,j , (1b)
network, the NOMA is applied since it can achieve high power
j=1
transfer efficiency and SUs can perform SIC [6], [7]. In this
case, the PBS broadcasts the information to the PUs in M

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

[ ]


M
hp,m,i h†p,m,i ; Fs,m,i = fs,m,i fs,m,i

; Qp,j = qp,j q†p,j ; Hs,j =
ye,m,k =ge,m,k (wp,m sp,m + vp,m ) † † †
hs,j hs,j ; Ge,m,k = ge,m,k ge,m,k ; Fe,m,k = fe,m,k fe,m,k ;
m=1
  † †
Qe,l = qe,l qe,l and Ge,l = ge,l ge,l . The expressions of

Ns

+ fe,m,k  ws,j ss,j + vs  , (1c) Γp,m,i , Γe,m,k , Γs,j , Λe,l,j , Λs,j,z and Λs,l,j are given in
j=1 (3). Without loss of generality, it is assumed that ∥h1 ∥ ≤
[ ] ∥h2 ∥ ≤ · · · ≤ ∥hNs ∥. Similar to [12], [27]-[28], it is

M
ye,l =q†e,l (wp,m sp,m + vp,m ) assumed furthermore that the EHR in the secondary network
has decoded SU j’s message before it decodes the SU i’s
m=1
  message, j < i. This over-estimates the interception capability

Ns
+ †
ge,l  ws,j ss,j + vs  , (1d) of EHRs and results in the worst-case secrecy rate of the SUs.
j=1 This conservative assumption was also used in [12], [27]-[28].
{[ M ]
where hp,m,i ∈ CNp,t ×1 and fs,m,i ∈ CNs,t ×1 are the ∑
channel vector between the PBS and the ith PU as well as Γp,m,i = Tr (Wp,m + Σp,m ) Hp,m,i
that between the CBS and the ith PU in the mth cluster,  m=1
 
respectively; qp,j ∈ CNp,t ×1 and hs,j ∈ CNs,t ×1 denote the ∑
Ns 
channel vector between the PBS and the jth SU as well as that + Ws,j + Σs  Fs,m,i 2
+ σp,m,i , (3a)

between the CBS and the jth SU, respectively. Furthermore, j=1
{[ ]
ge,m,k ∈ CNp,t ×1 and fe,m,k ∈ CNs,t ×1 are the channel ∑
M

vector between the PBS and the kth EHR and that between Γe,m,k = Tr (Wp,m + Σp,m ) Ge,m,k
the CBS and the jth EHR in the mth cluster, respectively;  m=1
 
qe,l ∈ CNp,t ×1 and ge,l ∈ CNs,t ×1 represent the channel ∑
Ns 
vector between the PBS and the lth EHR and that between the + Ws,j + Σs  Fe,m,k , (3b)

CBS and the lth EHR in the secondary network, respectively. j=1
{[ ]
Still regarding to (1a), sp,m ∈ C1×1 and wp,m ∈ CNp,t ×1 ∑
M
are the confidential information-bearing signal for the PUs in Γs,j = Tr (Wp,m + Σp,m ) Qp,j
the mth cluster and the corresponding beamforming vector, m=1
respectively. Furthermore, ss,j ∈ C1×1 and ws,j ∈ CNs,t ×1 + (Ws,j + Σs ) Hs,j } + σs,j
2
, (3c)
represent the confidential information-bearing signal delivered [[ M ]

for the jth SU and the corresponding beamforming vector, Λe,l,j = Tr (Wp,m + Σp,m ) Qe,l
respectively. Additionally, vp,m and vs denote the noise m=1
vector artificially generated by the PBS and the CBS. It is 2
+ (Ws,j + Σs ) Ge,l ] + σe,l , (3d)
2 2
assumed that E[|sp,m | ] = 1 and E[|ss,j | ] = 1. It is also {[ M ]

assumed that vp,m ∼ CN (0, Σp,m ) and vs ∼ CN (0, Σs ), Λs,j,z = Tr (Wp,m + Σp,m ) Hp,z
where Σp,m and ( Σ s are
) the AN covariance ( matrix.
) In (1),
 m=1
 
np,m,i ∼ CN 0, σp,m,i2
and ns,j ∼ CN 0, σs,j 2
respectively ∑
Ns 
denote the complex Gaussian noise at the ith PU in the mth + Ws,u + Σs  Hs,z 2
+ σs,z , (3e)
cluster and the lth SU. 
u=j
The secrecy rate of the ith PU in the mth cluster and {[ ]

M
the secrecy rate of the jth SU, denoted by Rp,m,i and Rs,j , Λs,l,j = Tr (Wp,m + Σp,m ) Qe,l
respectively, can be expressed as
[ ( )  m=1
 
Γp,m,i ∑
Ns 
Rp,m,i = log + Ws,υ + Σs  Ge,l 2
Γp,m,i − Tr (Wp,m Hp,m,i ) 
+ σe,l . (3f)
( )]+ υ=j
2
Γe,m,k + σe,m,k
− max log ,
k∈Km Γe,m,k − Tr (Wp,m Ge,m,k ) + σe,m,k 2

(2a) C. Non-linear Energy Harvesting Model


 [ ( )
 Γs,j In this paper, a practical non-linear EH model is adopted.


= log2 Γs,j −Tr(W s,j Hs,j )

 ( )] + According to [41]-[43], the harvesting power of EHRs, denot-



 − max log
Λe,l,j
, if j = Ns , ed by ΦE,A , can be formulated as:
 2 Λe,l,j −Tr(Ws,j Ge,l )
[ l∈L ( ) ( )
Rs,j ψe,A − Pe,A
max
Ψe,A

 = min log2 Λs,j,z −Tr(W
Λs,j,z
Φe,A = , (4a)

 H ) 1 − Ψe,A
 z∈{j,j+1,Ns } s,j s,z


 ( )] + max

 − max log2 Λ −Tr(W Λs,l,j
, otherwise.
Pe,A
s,j Ge,l )
ψe,A = , (4b)
l∈L s,l,j
1 + exp [−ae,A (Γe,A − be,A )]
(2b) 1

Ψe,A = , (4c)
† 1 + exp (ae,A be,A )
where Wp,m = wp,m wp,m ; Ws,j = ws,j ws,j ; Hp,m,i =

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

where A is the set of EHRs in the primary network and the sec- rates of the PUs and SUs, respectively; the constraints C3 and
ondary network, namely, A = A1 ∪ A2 , and A1 = ∪ Km , C4 are the constraints that can satisfy the harvested power
m∈M
m ∈ M, A2 = Ks ; ae,A and be,A represent parameters requirements of the EHRs in both the primary and secondary
that reflect the circuit specifications, such as the resistance, networks; and the constraint C5 is the rank-one constraint
the capacitance and diode turn-on voltage [42]. Furthermore, required for obtaining rank-one beamforming. Note that the
max
Pe,A is the maximum harvested power of EHRs when the EH optimization objective of P1 can be identified as the weight
circuit is saturated. In (4b), Γe,A is the RF power received at objective of a multiple-objective optimization problem that has
EHRs. Furthermore, Γe,A = Γe,m,k when the EHRs are in two optimization objectives (e.g., the transmission power of
the primary network and Γe,A = Λs,l,1 − σe,l 2
when the EHRs the PBS and the CBS) with the same weight. Due to the
are in the secondary network. Note that the noise power is constraints C1, C2 and C5, P1 is non-convex and difficult
ignored, since it is small compared to the RF signal power to solve. In order to solve this problem, a pair of suboptimal
[41]-[43]. schemes are proposed as follows.

III. AN- AIDED B EAMFORMING D ESIGN U NDER P ERFECT


CSI
In this section, an AN-aided beamforming design problem B. Suboptimal Solution Based on SDR
is formulated in MISO NOMA CRNs using SWIPT under the
perfect CSI. The CSI between the PBS and PUs as well as the To address the constraint C1, an auxiliary variable τm , m ∈
CSI between the CBS and the SUs can be obtained through the M, is introduced. Then, the constraint C1 can be equivalently
feedback from the corresponding transmitters and the receivers expressed as
[11], [12], [18], [19]. The CSI between the two networks
can be obtained with the cooperation between the primary { }
network and the secondary network [18], [33], [34]. The total Γp,m,i
log ≥ γp,m,i , (6a)
transmission power is minimized subject to the constraints [Γ − Tr (Wp,m Hp,m,i )] τm
 p,m,i 
on both the secrecy rates of PUs and SUs as well as on  2
Γe,m,k + σE,m,k 
the harvested power of EHRs in both the primary and the log [ ] ≤ 1,
 Γ 
e,m,k − Tr (Wp,m Ge,m,k ) + σe,m,k τm
2
secondary networks. In order to solve the challenging non-
convex problem, again, a pair of suboptimal algorithms are (6b)
proposed. One is based on SDR and the other is based on a
cost function.
where k ∈ Km and m ∈ M. Using successive convex
approximation (SCA), the constraints given by (6a) and (6b)
A. AN-aided Beamforming Design Problem
can be approximated as (7) and (8)
In order to minimize the sum of the transmission power
of the PBS and CBS, the beamforming weights and the AN
covariance of the PBS and the CBS are jointly optimized under exp (αp,m,i + βm − λp,m,i ) ≤ 2−γp,m,i , (7a)
constraints of the secrecy rate of PUs as well as SUs and under Γp,m,i − Tr (Wp,m Hp,m,i )
the EH requirements of the EHRs. The power minimization
≤ exp (e
αp,m,i ) (αp,m,i − α
ep,m,i + 1) , (7b)
problem is formulated as follows: ( )( )
  τm ≤ exp βem βm − βem + 1 , (7c)
∑M ∑Ns
P1 : min Tr  (Wp,m + Σp,m ) + Ws,j + Σs  Γp,m,i ≥ exp (λp,m,i ) , (7d)
Wp,m ,Σp,m
Ws,j ,Σs m=1 j=1
(5a)
s.t.
C1 : Rp,m,i ≥ γp,m,i , i ∈ Np,m , m ∈ M, (5b) exp (µe,m,k − ρe,m,k − δm ) ≤ 1, (8a)
2
C2 : Rs,j ≥ γs,j , j ∈ Ns , (5c) Γe,m,k + σe,m,k
C3 : Φe,A1 ≥ ζe,A1 , k ∈ Km , m ∈ M, (5d) ≤ exp (e
µe,m,k ) (µe,m,k − µ
ee,m,k + 1) , (8b)
C4 : Φe,A2 ≥ ζe,A2 , l ∈ Ks , (5e) Γe,m,k − Tr (Wp,m Ge,m,k ) + 2
σe,m,k ≥ exp (ρe,m,k ) , (8c)
C5 : Rank (Wp,m ) = 1, Rank (Ws,j ) = 1 (5f) τm ≥ exp (δm ) , (8d)
C6 : Wp,m ≽ 0, Ws,j ≽ 0. (5g)
In (5), γp,m,i and γs,j are the minimum secrecy rate re- where αp,m,i , βm , λp,m,i , µe,m,k , ρe,m,k , and δm are auxiliary
quirements of the ith PU in the mth cluster and of the ep,m,i , βem and µ
variables. Furthermore, α ee,m,k are approximate
jth SU; ζe,A1 and ζe,A2 are the minimum EH requirements values, and they are equal to αp,m,i , βm and µe,m,k , respec-
of EHRs in the primary and the secondary network. The tively, when the constraints are tight. Similarly, the constraint
constraints C1 and C2 are imposed to guarantee the secrecy C2 can be approximated as (9) and (10). When j = Ns , the

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

secrecy rate constraint of the Ns th SU can be formulated as TABLE I: The SCA-based algorithm
−γs,Ns
exp (αs,Ns + βs,Ns − λs,Ns ) ≤ 2 , (9a) Algorithm 1: The SCA-based algorithm for P1
Γs,Ns − Tr (Ws,Ns HS,Ns ) 1: Setting:
≤ exp (e
αs,Ns ) (αs,Ns − α
es,Ns + 1) , (9b) γp,m,i , γs,j Υm,i , ζe,A1 , ζe,A1 , i ∈ Np,m , k ∈ Km , m ∈ M
( )( ) l ∈ Ks and the tolerance error ϖ;
τs,Ns ≤ exp βes,Ns βs,Ns − βes,Ns + 1 , (9c) 2: Initialization:
The iterative number n = 1, α en en en en en
p,m,i , βm , µ e,m,k , α s,Ns , βs,Ns ,
Γs,Ns ≥ exp (λs,Ns ) , (9d) µ n es,j,z , ξes,j and µ
ee,l , αn n ee,l,j and Popt ;
n n

exp (µe,l − ρs,l − ωs,Ns ) ≤ 1, l ∈ Ks , (9e) 3: Repeat:


solve P2 by using CVX for the given approximate values;
Λe,l,Ns ≤ exp (eµe,l ) (µE,l − µ
ee,l + 1) , (9f) obtain αen+1 en+1 en+1 , α
p,m,i , βm , µ
n+1 en+1
e,m,k es,Ns , βs,Ns , µ en+1
e,l ,
n+1 en+1 n+1 n+1
Λe,l,Ns − Tr (Ws,Ns Ge,l ) ≥ exp (ρs,l ) , (9g) es,j,z , ξs,j and µ
α ee,l,j and Popt ;
if Rank (Wp,m ) = 1 and Rank (Ws,j ) = 1
τs,Ns ≥ exp (ωs,Ns ) , (9h) Obtain optimal Wp,m and Ws,j ;
else
where αs,Ns , βs,Ns , λs,Ns , µe,l , ρs,l , and ωs,Ns are auxiliary Obtain suboptimal Wp,m and Ws,j ;
variables. Furthermore, αes,Ns , βes,Ns and µ ee,l are approximate end
update the iterative number n = n + 1;
values, and they are equal to αs,Ns , βs,Ns and µe,l , respective- calculate the total transmit power Poptn ;

ly, when the constraints are tight. When j = 1, 2, · · · , Ns − 1, n (n−1)
if Popt − Popt ≤ ϖ
the secrecy rate constraint of the jth SU can be formulated as break;
end;
κj − ωj 2γs,j ≥ 0, (10a) 4: Obtain resource allocation:
exp (αs,j,z + ξs,j − λs,j,z ) ≤ 1, z ∈ {j, j + 1, Ns } , (10b) Wp,m , Ws,j , Σp,m and Σs .

Λs,j,z − Tr (Ws,j Hs,z )


≤ exp (e
αs,j,z ) (αs,j,z − α
es,j,z + 1) , (10c)
( )( ) Ws,j are of rank-one, the optimal beamforming scheme can
κj ≤ exp ξes,j ξs,j − ξes,j + 1 , (10d) be obtained by the eigenvalue decomposition and the obtained
Λs,j,z ≥ exp (λs,j,z ) , (10e) eigenvectors are optimal beamforming. If Wp,m and Ws,j are
not of rank-one, the suboptimal beamforming vectors can be
exp (µe,l,j − ρs,l,j − τs,j ) ≤ 1, (10f)
obtained by using the Gaussian randomization procedure [22].
Λs,l,j ≤ exp (e
µe,l,j ) (µe,l,j − µ
eE,l,j + 1) , (10g)
Λs,l,j − Tr (Ws,j Ge,l ) ≥ exp (ρs,l,j ) , C. Suboptimal Solution Based on Cost Function
(10h)
ωj ≥ exp (τs,j ) , Since Wp,m and Ws,j are semi-positive definite, the
(10i)
ranks of Wp,m and Ws,j are equal to 1 when their
where κj , ωj , αs,j,z , ξs,j , λs,j,z , µe,l,j , ρs,l,j and τs,j de-
maximum eigenvalues are equal to its trace, name-
note auxiliary variables. Furthermore, α es,j,z , ξes,j and µ
ee,l,j ly, when we have Rank (W ) = λ
p,m max (Wp,m )
are approximate values and equal to αs,j,z , ξs,j and µe,l,j , and Rank (W ) = λ
s,j max (Ws,j ); Otherwise, we have
respectively, when the constraints are tight. Constraints C3 Rank (W ) > λ
p,m max (Wp,m ) and Rank (Ws,j ) >
and C4 can be equivalently expressed as λ (W ). Thus, the rank-one constraint can be equivalent
{ } max s,j
1 max
Pe,A ∑
M ∑
Ns
Γe,A ≥ be,A − ln − 1 . to [Tr (Wp,m ) − λmax (Wp,m )] + [Tr (Ws,j ) − λmax
ae,A ζe,A (1 − Ψe,A ) + Pe,A max Ψ
e,A m=1 j=1
(Ws,j )] ≤ 0. From this insight, we can see that the smaller
(11)

M

Based on (7) and (11), using SDR, P can be solved by [Tr (Wp,m ) − λmax
1 m=1
iteratively solving P2 , given as ∑
Ns
  (Wp,m )] + [Tr (Ws,j ) −λmax (Ws,j )] is, the more likely

M ∑
Ns j=1
P2 : min Tr  (Wp,m + Σp,m ) + Ws,j + Σs  that the rank-one constraints can be satisfied. By exploiting a
Ξ cost function based approach, P2 is reformulated into P3 as
m=1 j=1
 
(12a) ∑M ∑Ns
s.t. C6, (7) − (11) , (12b) P3 : min Tr  (Wp,m + Σp,m ) + Ws,j + Σs 
Ξ
m=1 j=1
where Ξ is the set including all optimization variables and {
auxiliary variables. P2 is convex and can be efficiently solved ∑
M
+ℓ [Tr (Wp,m ) − λmax (Wp,m )]
by using the software CVX [22]. Algorithm 1 calculates the
solution of P1 . The details of Algorithm 1 are provided in
m=1

n ∑
Ns 
Table 1, where Popt denotes the minimum total transmission
+ [Tr (Ws,j ) − λmax (Ws,j )] (13a)
power at the nth iteration. 
j=1
Algorithm 1 does not guarantee that the optimal beam-
forming weights wp,m , ws,j can be obtained. If Wp,m and s.t. (12b) , (13b)

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

where ℓ > 0 is a cost factor. It may be readily shown that the TABLE II: The cost function-based algorithm

M
minimum value [Tr (Wp,m ) − λmax (Wp,m )] Algorithm 2: The cost function-based algorithm for P1
m=1

Ns 1: Setting:
+ [Tr (Ws,j ) − λmax (Ws,j )] can be obtained by using γp,m,i , γs,j Υm,i , ζe,A1 , ζe,A1 , i ∈ Np,m , k ∈ Km , m ∈ M
j=1 l ∈ Ks and the tolerance error ϖ;
a large ℓ value. Since λmax (Wp,m ) and λmax (Ws,j ) are 2: Initialization:
convex, P3 is non-convex. The following lemma is applied The iterative number n = 1, α en en en en en
p,m,i , βm , µ e,m,k , α s,Ns , βs,Ns ,
n n
to solve the non-convex problem P3 . Wp,m and Ws,j ;
Lemma 1 [45]: Let λmax (X) and λmax (Y) denote the 3: Repeat:
solve P4 by using CVX for the given approximate values;
maximum eigenvalue of X and Y, respectively. If X and en+1 en+1 en+1 , α n+1 en+1
en+1
obtain α p,m,i , βm , µ e,m,k es,Ns , βs,Ns , µ e,l ,
Y are semi-positive definite, then we have λmax (X) − (n+1)
Wp,m and Ws,j ;
(n+1)

λmax (Y) ≥ ymax (X − Y) ymax , where ymax is the eigen- set ℓ = 2ℓ;
vector related to the maximum eigenvalue of Y. end if
Using Lemma 1, P3 can be approximated as P4 . update the iterative number n = n + 1;
calculate
( the total
) transmit power P n ;
( n+1 ) n+1 ( n ) opt
P4 : min f Wp,m n+1
, Ws,j (14a) if Tr Wp,m − λmax Wp,m ≤ ϖ
Ξ ( ) ( )
n+1
and Tr Ws,j − λmax Ws,j n ≤ϖ
s.t. (12b) , (14b) break;
( n+1 n+1
) end;
where n is the iteration index and f Wp,m , Ws,j is given 4: Obtain resource allocation:
n , wn , Σ
wp,m
by s,j p,m and Σs .

( n+1 n+1
)
f Wp,m , Ws,j
 
∑M
( n+1 ) ∑
Ns
= Tr  Wp,m + Σp,m + Ws,jn+1
+ Σs  back to the CBS through a feedback link, which assumed error-
m=1 j=1 free in this simplified model [22], [44]. Under the bounded
{ error model, a robust AN-aided beamforming design problem

M
[ ( n+1 ) ( n )
+ℓ Tr Wp,m − λmax Wp,m is formulated. The non-convex problem is solved by using
SCA and the S-Procedure [45].
) n ]
m=1
( )† ( n+1
− n
wp,m Wp,m − Wp,m
n
wp,m

Ns
[ ( n+1 ) ( n )
+ Tr Ws,j − λmax Ws,j A. Robust AN-aided Beamforming Design Problem Formula-
j=1 tion
( n )† ( n+1 ) n ]}
− ws,j Ws,j − Ws,j
n
ws,j , (15) Under the bounded error model, the channel vector qe,l can
n n be formulated as
where wp,m and ws,j are the eigenvectors related to the
n n
maximum eigenvalue of Wp,m and Ws,j , respectively. It is
seen that P4 is convex and can be solved by using CVX. By qe,l = qe,l + ∆qe,l , l ∈ Ks , (16a)
{ }
Ψe,l = ∆qe,l ∈ C Np,t ×1 : ∆q†e,l ∆qe,l ≤ £2e,l ,
solving P4 , the iterative Algorithm 2 can be designed to solve ∆
(16b)
P1 . The details of Algorithm 2 are presented in Table 2.
while the channel vector fe,m,k can be expressed as
IV. AN- AIDED B EAMFORMING D ESIGN U NDER
I MPERFECT CSI fe,m,k = f e,m,k + ∆fe,m,k , m ∈ M, k ∈ Km , (17a)
{ }
In this section, the AN-aided beamforming design problem ∆ Ns,t ×1 †
Ψe,m,k = ∆fe,m,k ∈ C : ∆fe,m,k ∆fe,m,k ≤ £e,m,k ,
2
is studied in a more practical MISO NOMA CRN using
SWIPT, where the CSI between the CBS and the EHRs in the (17b)
primary network and the CSI between the PBS and the EHRs
in the secondary network are imperfect due to the limited where qe,l and f e,m,k are the estimates of qe,l and fe,m,k ,
cooperation between the primary and the secondary network. respectively; Ψe,l and Ψe,m,k represent the uncertainty region-
Since the bounded CSI error can be readily applied to model s of the channel vectors qe,l and fe,m,k , respectively; ∆qe,l
the estimating errors [13], [25], [35]-[38], it is opted for this and ∆fe,m,k denote the channel estimation errors of qe,l and
paper. Moreover, the CSI between the PBS and SUs and the fe,m,k ; £e,l and £e,m,k are the radii of the uncertainty regions
CSI between the CBS and the PBS can be obtained through the Ψe,l and Ψe,m,k , respectively.
cooperation between the primary network and the secondary Based on the bounded error models for qe,l and fe,m,k ,
network, or can be obtained from a third party such as a the power minimization problem subject to the constraints on
manager center [8]. Furthermore, the CSI of the secondary link the secrecy rates of the PUs and the SUs as well as on the
can be obtained by estimating it at the CBS and then sending it harvested power requirements of EHRs in both the primary

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

and secondary networks can be formulated as P5 , given as


 

M ∑Ns
min Tr  (Wp,m + Σp,m ) + Ws,j + Σs 
Wp,m ,Σp,m
Ws,j ,Σs m=1 j=1
(18a)
s.t.
C6 : Rp,m,i ≥ γp,m,i , i ∈ Np,m , m ∈ M, ∀∆fe,m,k ∈ Ψe,m,k ,
(18b)
C7 : Rs,j ≥ γs,j , j ∈ Ns , ∀∆qe,l ∈ Ψe,l , (18c)
C8 : Φe,A1 ≥ ζe,A1 , k ∈ Km , m ∈ M, ∀∆fe,m,k ∈ Ψe,m,k ,
(18d)
C9 : Φe,A2 ≥ ζe,A2 , l ∈ Ks , ∀∆qe,l ∈ Ψe,l , (18e)
C10 : Rank (Wp,m ) = 1, Rank (Ws,j ) = 1 (18f)
C11 : Wp,m ≽ 0, Ws,j ≽ 0. (18g)

The problem P5 is more challenging to solve due to the infinite Fig. 2: The minimum transmission power versus the number of
inequality constraints imposed by the uncertain regions, Ψe,l EHRs in the secondary network under the perfect CSI scenario.
and Ψe,m,k and owing to the non-convex constraints C6- C10.
V. S IMULATION R ESULTS
B. Suboptimal Solution Based on Cost Function In this section, simulation results are provided for com-
In order to make P5 tractable, the S-Procedure of [45] is paring the performance obtained by using NOMA to that
applied. achieved by using OMA. Simulation results are also presented
† to evaluate the performance of the proposed algorithms. The
Lemma
{ }2 (S-Procedure) [45]: Let fi (z) = z Ai z +

2Re bi z + ci , i ∈ {1, 2}, where z ∈ CN ×1 , Ai ∈ HN , time division multiple access (TDMA) scheme is selected as
the OMA scheme for comparison. The simulation settings
bi ∈ CN ×1 and ci ∈ R. Then, the expression f1 (z) ≤ 0 ⇒
are based on those used in [42] and [43]. All the channels
f2 (z) ≤ 0 holds if and only if there exists a ς ≥ 0 such that
involved are assumed to be Rayleigh flat fading. The number
we have:
[ ] [ ] of channel realizations is 104 . The variance of noise at all
ς
A1 b1

A2 b2
≽ 0, (19) users and EHRs is −120 dBm. The channel distributions are
b†1 c1 b†2 c2 set as: hp,m,i ∼ CN (0, 2I), fs,m,i ∼ CN (0, 0.5I), qp,j ∼
CN (0, 0.5I), hs,j ∼ CN (0, 2I), ge,m,k ∼ CN (0, 1.5I),
provided that there exists a vector b
z so that we have fi (b
z) < 0. fe,m,k ∼ CN (0, 0.5I), qe,l ∼ CN (0, 0.5I) and ge,l ∼
Using Lemma 2 and SCA, the constraint C6 of P5 can be CN (0, 1.5I). The detailed simulation settings are given in
approximated as (20) at the top of the next page. In (20), Table III.
λe,m,k ≥ 0 and ue,m,k ≥ 0 are slack variables while θe,m,k
and oe,m,k are auxiliary variables. Similarly, the constraint C7
TABLE III: Simulation Parameters
is approximated as follows.
Parameters Notation Typical Values
When j = Ns , one has (21) at the top of the next page.
and when j = 1, 2, · · · , Ns − 1, the secrecy rate constraint Numbers of antennas of the PBS Np,t 10
of the jth SU can be approximated as (22) at the top of the Numbers of antennas of the CBS Ns,t 5
Numbers of the clusters M 2
next page. where ω e,l ≥ 0, κe,l ≥ 0, τ ≥ 0 and η e,l ≥ 0 are Numbers of PUs Np,m 2
slack variables. The constraints C8 and C9 can be equivalently Numbers of SUs Ns 3
The maximum harvested power max
Pe,A 24 mW
expressed as (23) at the top of the next two pages. In (23),
Circuit parameter ae,A 1500
χe,m,k ≥ 0 and φe,l ≥ 0 are slack variables. By using (20)- Circuit parameter be,A 0.0022
(23), P5 can be solved by iteratively solving P6 , given as The minimum secrecy rate of PUs γp,m,i 2 bits/s/Hz
( n+1 n+1
) The minimum secrecy rate of SUs γs,j 1 bits/s/Hz
P6 : min f Wp,m , Ws,j (24a) The minimum EH of EHRs in set A1 ζe,A1 15 mW
Ξ1 The minimum EH of EHRs in set A2 ζe,A2 5 mW
s.t. C11, (20) − (23) , (24b) The tolerance error ϖ 10−4
( n+1 n+1
) The radiuses of the uncertainty regions £e,l 10−2
where f Wp,m , Ws,j is given by (15), and Ξ1 denotes the £e,m,k 10−2
set including all optimization variables, auxiliary variables and
slack variables. Since P6 is convex, it can be readily solved by
using CVX. Similar to P1 , Algorithm 2 can be used to solve Fig. 2 shows the minimum transmission power versus the
P5 . The procedure is the same and it is omitted due to space number of EHRs in the secondary network under the perfect
limitation. CSI scenario. Since Algorithm 2 is capable of obtaining

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

10

 ( ) ( ) 

Ns ∑
Ns
 λe,m,k I − Ws,j + Σs − Ws,j + Σs f e,m,k 
 
 ( j=1 ) ( j=1
)  ≽ 0, (20a)
 ∑ ∑ 
 † Ns † Ns

−f e,m,k Ws,j + Σs θe,m,k − f e,m,k Ws,j + Σs f e,m,k − λe,m,k £2e,m,k
j=1 j=1
 ( ) ( ) 

Ns ∑
Ns
 ue,m,k I + Ws,j + Σs Ws,j + Σs f e,m,k 
 
 ( j=1 ) ( j=1
)  ≽ 0, (20b)
 ∑ ∑ 
 † Ns † Ns

f e,m,k Ws,j + Σs f e,m,k Ws,j + Σs f e,m,k − λe,m,k £2e,m,k − oE,m,k
j=1 j=1

(7) , (8a) and (8d) . (20c)

 

M ∑
M
 ω e,l I − (W p,m + Σ p,m ) − (W p,m + Σ p,m )qe,l 
 m=1 m=1  ≽ 0, (21a)
 † ∑
M
† ∑M 
−qe,l (Wp,m + Σp,m ) −Λe,l − qe,l (Wp,m + Σp,m ) qe,l − ω e,l £e,l
2
m=1 m=1
 
∑M ∑
M
κ
 e,l I + (W p,m + Σ p,m ) (W p,m + Σ )q
p,m e,l 
 m=1 m=1  ≽ 0, (21b)
 ∑ 
e e,l + q† ∑ (Wp,m + Σp,m ) qe,l − κe,l £2
M M
q†e,l (Wp,m + Σp,m ) Λ e,l e,l
m=1 m=1
2
Λe,l = Tr [(Ws,Ns + Σs ) Ge,l ] + σe,l − exp (e
µe,l ) (µe,l − µ
ee,l + 1) , (21c)
e
Λe,l = Tr (Σs Ge,l ) + σe,l − exp (ρs,l ) ,
2
(21d)
(9a) − (9d) , (9e) and (9h) , (21e)

 

M ∑
M
 τ e,l I − m=1 (Wp,m + Σp,m ) − (Wp,m + Σp,m )qe,l 
 m=1  ≽ 0, (22a)
 † ∑
M
† ∑M 
−qe,l (Wp,m + Σp,m ) −Λs,l,j − qe,l (Wp,m + Σp,m ) qe,l − τ e,l £e,l 2
m=1 m=1
 
∑M ∑M
η
 e,l I + (W p,m + Σ p,m ) (W p,m + Σ p,m )q e,l 
 m=1 m=1  ≽ 0, (22b)
 † ∑
M
† ∑M 
qe,l (Wp,m + Σp,m ) Λs,l,j + qe,l (Wp,m + Σp,m ) qe,l − η e,l £e,l 2
m=1
   m=1
 ∑ N s 
Λs,l,j = Tr  Ws,υ + Σs  Ge,l + σe,l 2
− exp (e
µe,l,j ) (µe,l,j − µ ee,l,j + 1) , (22c)
 
υ=j
  

Ns
e s,l,j = Tr 
Λ Ws,υ + Σs  Ge,l  + σe,l 2
− exp (ρs,l,j ) , (22d)
υ=j+1

(10a) − (10f) , and (10i) , (22e)

rank-one solutions, it is used to obtain the results of CRNs The reason is that our proposed cooperative jamming scheme
using OMA, OMA with cooperation and NOMA without is beneficial for the PUs to achieve a high secrecy rate and
cooperation. Fig. 2 also compares the performance obtained that a low transmission power is required for guaranteeing
by using Algorithm 1 to that achieved by using Algorithm the secrecy rate of the PUs. This indicates that our proposed
2. It is seen that the minimum transmission power consumed cooperative jamming scheme is eventually helpful for secure
without the cooperative jamming scheme is higher than that communications. It is also seen that the transmission power
consumed with our proposed cooperative jamming scheme. consumed by using NOMA is lower than that consumed by

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

11

 ( ) ( ) 

Ns ∑
Ns
 χe,m,k I + Ws,j + Σs Ws,j + Σs f e,m,k 
 
 ( j=1 ) ( j=1 )  ≽ 0, (23a)
 ∑ ∑ 
 † Ns † Ns

f e,m,k Ws,j + Σs Γe,m,k + f e,m,k Ws,j + Σs f E,m,k − χe,m,k £2e,m,k
j=1 j=1
 

M ∑
M
 φe,l I + m=1 (Wp,m + Σp,m ) (Wp,m + Σp,m )qe,l 
 m=1  ≽ 0, (23b)
 † ∑M
† ∑M 
qe,l (Wp,m + Σp,m ) Γe,l + qe,l (Wp,m + Σp,m ) qe,l − ϕe,l £E,l
2
m=1 m=1
{[ M ] }

Γe,m,k = Tr (Wp,m + Σp,m ) Ge,m,k − be,m,k m
m=1
{ }
max
1 Pe,m,k
+ ln −1 , (23c)
ae,m,k ψe,m,k (1 − Ψe,m,k ) + Pe,m,k
max Ψ
E,m,k
   { }
 ∑ Ns  1 max
Pe,l
Γe,l = Tr  Ws,j + Σs  GE,l − be,l + ln −1 . (23d)
  ae,l ψe,l (1 − Ψe,l ) + Pe,l
max Ψ
e,l
j=1

TABLE IV: Comparison of the number of rank-one solutions


achieved
``by Algorithms 1 and 2
``` (Ks )
Algorithm ````
(1) (2) (3) (4) (5)
Algorithm 1 0 0 0 0 0
Algorithm 2 259 207 248 301 196

using TDMA both with and without cooperation between the


primary network and the secondary network. This can be
explained by the fact that NOMA provides a higher secrecy
rate for SUs than TDMA [12], [27]. Thus, the transmission
power required for guaranteeing the secrecy rate of SUs can
be decreased. Moreover, it is interesting to note that the
transmission power achieved by using Algorithm 1 is higher
than that obtained by using Algorithm 2. It indicates that
Algorithm 2 outperforms Algorithm 1 in terms of transmission Fig. 3: The minimum transmission power versus the secrecy
power minimization. The reason is that Algorithm 2 can obtain rate of PUs under perfect CSI scenario.
rank-one solutions, while Algorithm 1 cannot achieve rank-one
solutions.
Table 3 is given to show the number of rank-one solutions transmission power is required for guaranteeing the increased
achieved by using Algorithm 1 and Algorithm 2 under perfect secrecy rate requirement of PUs. As shown in Fig. 3, the
CSI. The results are obtained for 1000 channel realizations. minimum transmission power required when the proposed
It is seen from Table 3 that Algorithm 2 can provide rank- cooperative jamming scheme is applied is lower than that
one solutions, while Algorithm 1 cannot. The reason is that in the absence of cooperation between the primary and the
a cost function related to rank-one solutions is applied in secondary network. This phenomenon further demonstrates
Algorithm 2. When rank-one solutions are achieved, the op- the efficiency of our proposed cooperative jamming scheme
timal beamforming vectors can be obtained for the CBS and for achieving secure communications. It is also seen from
PBS; otherwise, the optimal beamforming vectors cannot be Fig. 3 that NOMA outperforms TDMA and that Algorithm
obtained. This is the reason why Algorithm 2 performs better 2 performs better than Algorithm 1 in terms of the required
than Algorithm 1. transmission power.
Fig. 3 is provided for further verifying the efficiency of Fig. 4 shows the minimum transmission power versus the
the proposed cooperative jamming scheme. The number of number of iterations required by using Algorithm 1 and
EHRs in the secondary network is 2. It is observed that Algorithm 2. The number of EHRs in the secondary network is
the transmission power increases with the secrecy rate of set to 2 or 3. It is seen from Fig. 4 that both Algorithm 1 and
PUs. It can be readily explained by the fact that a high Algorithm 2 require only a few iterations to converge. This

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

12

Fig. 4: The minimum transmission power versus the number Fig. 6: The minimum transmission power versus the minimum
of iterations required by using Algorithms 1 and 2. harvested power requirement of EHRs in the secondary net-
work for both perfect and imperfect CSI scenarios.

minimum harvested power requirement of EHRs in the sec-


ondary network for both perfect and imperfect CSI scenarios.
The number of EHRs in the secondary network is set to 3.
The simulation results are obtained by Algorithm 2. It is seen
that the transmission power consumed by using TDMA is
higher than that consumed by using NOMA, for both perfect
and imperfect CSI scenarios. The reason is that the secrecy
rate of SUs achieved by using NOMA can be higher than
that obtained by using TDMA [12]. A lower transmission
power is required to grantee the secrecy rate of SUs. It is
also seen that the imperfect CSI has a significant effect on the
minimum transmission power. Moreover, as shown in Fig. 6,
the minimum transmission power increases with the harvested
Fig. 5: The empirical CDF of the minimum transmission power power of EHRs in the secondary network. It can be readily
of the CBS for both perfect and imperfect CSI scenarios. explained by the fact that a higher transmission power is
required to satisfy the increased harvesting power of EHRs
in the secondary network.
confirms the efficiency of our proposed algorithms. It is also
seen that the number of iterations required by Algorithm 2 is VI. C ONCLUSIONS
larger than that required by Algorithm 1. The reason is that the Secure communication was studied for MISO NOMA CRNs
complexity of Algorithm 2 is higher than that of Algorithm using SWIPT, where a practical non-linear EH model was
1. From Fig. 3 and Fig. 4, it is seen that there is a tradeoff applied. To enhance the security of the primary network, an
between the complexity of algorithms and the transmission artificial-noise-aided cooperative jamming scheme was pro-
power obtained by using these algorithms. posed. The transmission beamforming vectors and AN-aided
Fig. 5 shows the empirical cumulative distribution function covariance matrix were jointly optimized to minimize the
(CDF) of the minimum transmission power under the perfect total transmission power of the network, while the secrecy
CSI and the bounded CSI error model. The number of EHRs rates of both PUs and SUs as well as the EH requirement
in the secondary network is 2. It is seen from Fig. 5 that the of EHRs were satisfied. The beamforming design problems
transmission power consumed under the bounded CSI error were investigated under both the perfect CSI and the bounded
model is higher than that consumed under the perfect CSI, for CSI error model. A pair of algorithms were proposed to solve
both Algorithm 1 and Algorithm 2. The reason is that a higher these challenging non-convex problems. It was shown that
transmission power is required for guaranteeing the secrecy the performance achieved by using NOMA is better than that
rates of the PUs and the SUs, when the CSI is imperfect. It is obtained by using OMA. Simulation results also show that
also seen that Algorithm 2 is superior to Algorithm 1 in terms the algorithm based on the cost function is superior to the
of the transmission power minimization, even when the CSI algorithm based on SDR. Moreover, our proposed cooperative
is imperfect. jamming scheme is efficient to improve the security of MISO
Fig. 6 shows the minimum transmission power versus the NOMA CRNs using SWIPT.

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

13

R EFERENCES [24] D. Wang, P. Ren, Q. Du, J. Chen, and Y. Wang, “Achieving full secrecy
rate with energy-efficient transmission control,” IEEE Trans. Commun.,
[1] J. G. Andrews, S. Buzzi, W. Choi, et al., “What will 5G be?,” IEEE J. vol. 65, no. 12, pp. 5386-5400, Dec. 2017.
Sel. Areas Commun., vol. 32, no. 6, pp. 1065-1082, Jun. 2014. [25] Q. Zhang, Q. Li, and J. Qin, “Robust beamforming for nonorthongonal
[2] Y. C. Liang, K. C. Chen, G. Y. Li, and P. Mahonen, “Cognitive muliple-access systems in MISO channels,” IEEE Trans. Veh. Technol.,
radio networking and communications: an overview,” IEEE Trans. Veh. vol. 65, no. 12, pp. 10231-10236, Dec. 2016.
Technol., vol. 60, no. 7, pp. 3386-3407, Sept. 2011. [26] F. Alavi, K. Cumanan, Z. Ding, and A. G. Burr, “Robust beaforming
[3] Z. Zhang, H. Sun, and R. Q. Hu, “Downlink and uplink non-orthogonal techniques for non-orthogonal multiple access systems with bounded
multiple access in a dense wireless network,” IEEE J. Sel. Areas channel uncertainties,” IEEE Commun. Lett., to be published, 2017.
Commun., to be published, 2017. [27] B. He, A. Liu, N. Yang, V. K. N. Lau, “On the design of secure non-
[4] Z. Ding, X. Lei, G. K. Karagiannidis, R. Schober, J. Yuan, and orthogonal multiple access systems,” IEEE J. Sel. Areas Commun., to
V. Bhargava, “A survey on non-orthognonal multiple access for 5G be published, 2017.
networks: Research chanllenges and future trends,” IEEE J. Sel. Areas [28] Y. Li, M. Jiang, Q. Zhang, Q. Li, and J. Qin, “Secure beamforming in
Commun., to be published, 2017. downlink MISO nonorthogonal multiple access systems,” IEEE Trans.
[5] B. Wang, L. Dai, Z. Wang, N. Ge, and S. Zhou, “Spectrum and energy ef- Veh. Technol., vol. 66, no. 8, pp. 7563-7567, Aug. 2017.
ficient beamspace MIMO-NOMA for millimeter-wave communications [29] J. Lei, J. Zhang, K. H. Park, et al., “On secure NOMA systems with
using lens antenna array,” IEEE J. Sel. Areas Commun., to be published, tranmit antennas selection schemes,” IEEE Access, to be published,
2017. 2017.
[6] Y. Liu, Z. Ding, M. Elkashlan, and J. Yuan, “Nonorthogonal multiple [30] M. Tian, Q. Zhang, S. Zhao, Q. Li, and J. Qin, “Secrecy sum rate opti-
access in large-scale underlay cognitive radio networks,” IEEE Trans. mization for downlink MIMO non-orthogonal multiple access systems,”
Veh. Technol., vol. 65, no. 12, pp. 10152-10157, Dec. 2016. IEEE Signal Process. Lett., vol. 24, no. 8, pp. 1113-1117, Aug. 2017.
[7] L. Lv, J. Chen, Q. Ni, and Z. Ding, “Design of cooperative non- [31] Y. Pei, Y. C. Liang, K. Teh, and K. H. Li, “Secure communication
orthogonal multicast cognitive multiple access for 5G systems: user in multiantenna cognitive radio networks with imperfect channel state
scheduling and performance analysis,” IEEE Trans. Commun., vol. 65, information,” IEEE Trans. Signal Process., vol. 59, no. 4, pp. 1683-1693,
no. 6, pp. 2641-2656, June 2017. Apr. 2011.
[8] F. Zhou, N. C. Beaulieu, Z. Li, J. Si, and P. Qi, “Energy-efficient [32] L. Zhang, R. Zhang, Y. C. Liang, Y. Xin, and S. Cui, “On the relationship
optimal power allocation for fading cognitive radio channels: ergodic between the multi-antenna secrecy communications and cognitive radio
capacity, outage capacity and minimum-rate capacity,” IEEE Trans. communications,” IEEE Trans. Commun., vol. 58, no. 6, pp. 1877-1886,
Wireless Commun., vol. 15, no. 4, pp. 2741-2755, Apr. 2016. June 2010.
[9] D. Feng, C. Jiang, G. Lim, L. J. Cimini, G. Feng, and G. Y. Li, “A survey [33] C. Wang, and H. Wang, “On the secrecy throughput maximization for
of energy-eficient wireless communications,” IEEE Commun. Surveys MISO cognitive radio network in slow fading channels,” IEEE Trans.
Tuts., vol. 15, no. 1, pp. 167-178, Feb. 2013. Inf. Forensics Security, vol.9, no. 11, pp. 1814-1827, Nov. 2014.
[10] X. Lu, P. Wang, D. Niyato, D. I. Kim, and Z. Han, “Wireless networks [34] W. Zeng, Y. Zheng, and C. Xiao, “Multiantenna secure cognitive radio
with RF energy harvesting: A contemporary survey,” IEEE Commun. networks with finite-alphabet inputs: A global optimization approach
Surveys Tuts., vol. 17, pp. 757-789, Second Quarter, 2015. for precoder desing,” IEEE Trans. Wireless Commun., vol. 15, no. 4,
[11] Y. Wu, R. Schober, D. W. K. Ng, C. Xiao, and G. Caire, “Secure pp. 3044-3057, Apr. 2016.
massive MIMO transmission with an active eavesdropper,” IEEE Trans. [35] D. W. K. Ng, M. Shaqfeh, R. Schober, and H. Alnuweiri, “Robust
Inf. Theory, vol. 62, no. 7, pp. 3880-3900, 2016. layered transmission in secure MISO multiuser unicast cognitive radio
[12] Y. Zhang, H. Wang, Q. Yang, and Z. Ding, “Secrecy sum rate maxi- systems,” IEEE Trans. Veh. Technol., vol. 65, no. 10, pp. 8267-8282,
mization in non-orthogonal multiple access,” IEEE Commun. Lett., vol. Oct. 2016.
20, no. 5, pp. 930-933, 2016. [36] V. Nguyen, T. Duong, O. Shin, A. Nallanathan, and G. K. Karagiannidis,
[13] D. W. K. Ng, E. S. Lo, and R. Schober, “Multi-objective resource “Enhancing PHY security of cooperative cognitive radio multicast
allocation for secure communication in cognitive radio networks with communications,” IEEE Trans. Cogn. Net., to be published, 2017.
wireless information and power transfer,” IEEE Trans. Veh. Technol., [37] L. Mohjazi, I. Ahmed, S. Muhaidat, M. Dianati, M. Al-qutayri, “Down-
vol. 20, no. 2, pp. 328-331, Feb. 2016. link Beamforming for SWIPT multi-user MISO underlay cognitive radio
[14] Y. Wu, R. Schober, D. W. K. Ng, C. Xiao, and G. Caire, “Secure massive networks” IEEE Commun. Lett., vol. 21, no. 2, pp. 434-437, Feb. 2017.
MIMO transmission in the presence of an active eavesdropper,” in Proc. [38] C. Xu, Q. Zhang, Q. Li, Y. Tan, and J. Qin, “Robust transceiver design
IEEE ICC 2015, London, UK, 2015. for wireless information and power transmission in underlay MIMO
[15] Y. Zou, B. Champagne, W. Zhu, and L. Hanzo, “Relay-selection cognitive radio networks,” IEEE Commun. Lett., vol. 18, no. 9, pp. 1665-
improves the security-reliablity trade-off in cognitive radio systems,” 1668, Sept. 2014.
IEEE Trans. Commun., vol. 63, no. 1, pp. 215-228, Jan. 2015. [39] H. Zhang, P. Sun, C. Li, Y. Huang, and L. Yang, “Cooperative precoding
[16] X. Chen, D. W. K. Ng, W. H. Gerstacker, and H. H. Chen, “A survey on for wireless energy transfer and secure cognitive radio coexisence
multiple-antenna techniques for physical layer security,” IEEE Commun. systems,” IEEE Signal Process. Lett., vol. 24, no. 5, pp. 540-544, May
Surveys Tuts., vol. 19, no. 2, pp. 1027-1053, Second Quarter, 2017. 2017.
[17] Y. Wu, J. Wang, J. Wang, R. Schober, and C. Xiao, “Secure transmission [40] Y. Yuan and Z. Ding, “Outage constrained secrecy rate maximization
with large numbers of antennas and finite alphabet inputs,” IEEE Trans. design with SWIPT in MIMO-CR systems,” IEEE Trans. Veh. Technol.,
Commun., vol. 65, no. 8, pp. 3614-3628, June 2017. to be published, 2017.
[18] Y. Pei, Y. C. Liang, K. C. Teh, and K. H. Li, “Secure communication [41] Y. Huang, Z. Li, F. Zhou, and R. Zhu, “Robust AN-aided beamforming
over MISO cognitive radio channels,” IEEE Trans. Wireless Commun., design for secure MISO cognitive radio based on a practical nonlinear
vol. 9, no. 4, pp. 1494-1502, Apr. 2010. EH model,” IEEE Access, vol. 5, pp. 14011-14019, July 2017.
[19] Y. Wu, R. Schober, D. W. K. Ng, C. Xiao, and G. Caire, “Secure massive [42] E. Boshkovska, D. Ng, N. Zlatanov, and R. Schober, “Practical non-
MIMO transmission in the presence of an active eavesdropper,” in Proc. linear energy harvesting model and resource allocation for SWIPT
IEEE ICC 2015, London, UK, 2015. systems,” IEEE Commun. Lett., vol. 19, pp. 2082-2085, Dec. 2015.
[20] Y. Zou, X. Li, and Y. Liang, “Secrecy outage and diversity analysis of [43] E. Boshkovska, D. W. K. Ng, N. Zlatanov, A. Koelpin, and R. Schober,
cognitive radio systems,” IEEE J. Sel. Areas Commun., vol. 32, no. 11, “Robust resource allocation for MIMO wireless powered communication
pp. 2222-2236, Nov. 2014. networks based on a non-linear EH model,” IEEE Trans. Commun., vol.
[21] Y. Zou, J. Zhu, X. Li, and L. Hanzo, “Relay selection for wireless 65, no. 5, pp. 1984-1999, May 2017.
communications against eavesdropping: A security-reliablity trade-off [44] X. Kang, R. Zhang, Y. C. Liang, and H. K. Garg, “Optimal power
persepective,” IEEE Network, vol. 30, no. 5, pp. 74-79, May, 2016. allocation strategies for fading cognitive radio channels with primary
[22] F. Zhou, Z. Li, J. Cheng, Q. Li, and J. Si, “Robust AN-Aided Beamform- user outage constraint,” IEEE J. Sel. Areas Commun., vol. 29, no. 2, pp.
ing and Power Splitting Design for Secure MISO Cognitive Radio With 374-383, Feb. 2011.
SWIPT,” IEEE Trans. Wireless Commun., vol. 16, no. 4, pp. 2450-2464, [45] S. P. Boyd and L. Vandenberghe, Convex Optimization. Cambridge,
April 2017. U.K.: Cambridge Univ. Press, 2004.
[23] D. Wang, P. Ren, Q. Du, L. Sun, and Y. Wang, “Security provisioning
for vehicular relay networks via cooperative jamming and signal super-
position,” IEEE Trans. Veh. Technol., vol. 66, no. 12, pp. 10732-10747,
Dec. 2017.

0733-8716 (c) 2018 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.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JSAC.2018.2824622, IEEE Journal
on Selected Areas in Communications

14

Fuhui Zhou received the Ph. D. degree from Xidian Lajos Hanzo (http://www-mobile.ecs.soton.ac.uk)
University, Xian, China, in 2016. He is an associate is FREng, FIEEE, FIET, Fellow of EURASIP. He
Professor with School of Information Engineering, received his degree in electronics in 1976 and his
Nanchang University. He is now a Research Fellow doctorate in 1983. In 2009, he was awarded an
at Utah State University. He has worked as an honorary doctorate by the Technical University of
international visiting Ph. D student of the University Budapest and in 2015 by the University of Edin-
of British Columbia from 2015 to 2016. His research burgh. In 2016, he was admitted to the Hungarian
interests focus on cognitive radio, green communi- Academy of Science. During his 40-year career in
cations, edge computing, machine learning, NOMA, telecommunications he has held various research
physical layer security, and resource allocation. He and academic posts in Hungary, Germany and the
has published more than 40 papers, including IEEE UK. Since 1986, he has been with the School of
Journal of Selected Areas in Communications, IEEE Transactions on Wireless Electronics and Computer Science, University of Southampton, UK, where
Communications, IEEE Wireless Communications, IEEE Network, IEEE he holds the chair in telecommunications. He has successfully supervised 111
GLOBECOM, etc. He has served as Technical Program Committee (TPC) PhD students, co-authored 18 John Wiley/IEEE Press books on mobile radio
member for many International conferences, such as IEEE GLOBECOM, communications totalling in excess of 10 000 pages, published 1701 research
IEEE ICC, etc. He serves as an Associate Editor of IEEE Access. contributions at IEEE Xplore, acted both as TPC and General Chair of IEEE
conferences, presented keynote lectures and has been awarded a number of
distinctions. Currently he is directing a 60-strong academic research team,
working on a range of research projects in the field of wireless multimedia
communications sponsored by industry, the Engineering and Physical Sciences
Research Council (EPSRC) UK, the European Research Council’s Advanced
Fellow Grant and the Royal Society’s Wolfson Research Merit Award. He
Zheng Chu received the Ph.D. degree from Newcas- is an enthusiastic supporter of industrial and academic liaison and he offers
tle University, Newcastle upon Tyne, U.K., in 2016. a range of industrial courses. He is also a Governor of the IEEE ComSoc
He was with the Faculty of Science and Technology, and VTS. During 2008 - 2012, he was the Editor-in-Chief of the IEEE Press
Middlesex University, London, U.K., from 2016 to and a Chaired Professor also at Tsinghua University, Beijing. For further
2017. He is currently with the 5G Innovation Center, information on research in progress and associated publications, please refer
Institute of Communication Systems, University of to http://www-mobile.ecs.soton.ac.uk. He has 30000+ citations and an H-index
Surrey, Guildford, U.K. His current research inter- of 73.
ests include physical layer security, wireless coop-
erative networks, wireless power transfer, convex
optimization techniques, and game theory.

Haijian Sun received the B.S. and M.S. degrees


from Xidian University, Xi’an, China, in 2011 and
2014, respectively. He is currently pursuing the
Ph.D. degree with the Department of Electrical and
Computer Engineering, Utah State University, Lo-
gan, UT, USA. His research interests include MIMO,
non-orthogonal multiple access, SWIPT, wearable
communications, and 5G PHY.

Rose Qingyang Hu is a Professor of Electrical


and Computer Engineering Department at Utah S-
tate University. She received her B.S. degree from
University of Science and Technology of China,
her M.S. degree from New York University, and
her Ph.D. degree from the University of Kansas.
She has more than 10 years of R&D experience
with Nortel, Blackberry and Intel as a technical
manager, a senior wireless system architect, and
a senior research scientist, actively participating in
industrial 3G/4G technology development, standard-
ization, system level simulation and performance evaluation. Her current
research interests include next-generation wireless communications, wireless
system design and optimization, green radios, Internet of Things, Cloud
computing/fog computing, multimedia QoS/QoE, wireless system modeling
and performance analysis. She has published over 180 papers in top IEEE
journals and conferences and holds numerous patents in her research areas.
Prof. Hu is an IEEE Communications Society Distinguished Lecturer Class
2015-2018 and the recipient of Best Paper Awards from IEEE Globecom
2012, IEEE ICC 2015, IEEE VTC Spring 2016, and IEEE ICC 2016.

0733-8716 (c) 2018 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.

Vous aimerez peut-être aussi