Anda di halaman 1dari 32

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO.

2, SECOND QUARTER 2018 1329

Beamforming in Wireless Energy Harvesting


Communications Systems: A Survey
Yamen Alsaba , Sharul Kamal Abdul Rahim, Senior Member, IEEE, and Chee Yen Leow, Member, IEEE

Abstract—Wireless energy harvesting (EH) is a promising Both stability and the availability of wireless signals (TV
solution to prolong lifetime of power-constrained networks such broadcasting, mobile base stations, etc. . .), in addition to the
as military and sensor networks. The high sensitivity of energy dependence of the natural energy on location, climate and
transfer to signal decay due to path loss and fading, promotes
multi-antenna techniques like beamforming as the candidate time, nominate wireless EH (the ability of transforming the
transmission scheme for EH networks. Exploiting beamforming wireless RF signals into DC voltage to charge the device
in EH networks has gained overwhelming interest, and lot of lit- battery) as the best EH scheme. Yet, this emerging tech-
erature has appeared recently regarding this topic. The objective nology requires a shift in the system architecture and its
of this paper is to point out the state-of-the-art research activity power-information resource allocation strategies to meet its
on beamforming implementation in EH wireless networks. We
first review the basic concepts and architecture of EH wireless new demands [8].
networks. In addition, we also discuss the effects of beamforming Three main transmit schemes can be distinguished in wire-
transmission scheme on system performance in EH wireless com- less EH networks. The first scheme, is Wireless Power Transfer
munication. Furthermore, we present a comprehensive survey (WPT) [9], where a power transmitter simply transfers power
of multi-antenna EH communications. We cover the support- to EH receivers to charge their batteries, without any infor-
ing network architectures like broadcasting, relay, and cognitive
radio networks with the various beamforming deployment within mation exchange. The second scheme is Wireless Powered
the network architecture. We classify the different beamform- Communication Network (WPCN) [10], [11], where the EH
ing approaches in each network topology according to its design receiver uses the power harvested in the downlink in the
objective such as increasing the throughput, enhancing the energy first time slot, to transmit its information in the second time
transfer efficiency, and minimizing the total transmit power, with slot. The third transferring scheme, is Simultaneous Wireless
paying special attention to exploiting the physical layer secu-
rity. We also survey major advances as well as open issues, Information and Power Transfer (SWIPT) [12], in this scheme
challenges, and future research directions in multi-antenna EH a hybrid transmitter is transferring wireless energy and infor-
communications. mation signals using the same waveform to multiple EH and
Index Terms—Energy harvesting, beamforming, wireless pow- ID receivers. SWIPT has attracted increasing interest as it
ered communication network, simultaneous wireless information exploits both information and power concurrently yielding
and power transfer, wireless power transfer, physical layer higher information-energy transmission efficiency. Figure 1,
security. illustrates the different EH transmission schemes.
The receiver architecture should be reconsidered in order to
adopt to the new wireless EH scheme. Two distinct types of
I. I NTRODUCTION
receivers can be found in EH-enabled networks separated and
NERGY harvesting (EH) technique has lately gained a lot
E of attention in both academia and industry, as this new
green communication technique provides a promising solution
collocated receivers. In separated receivers case, the receivers
are either doing EH (EH receivers) or Information Decoding
(Id) (ID receivers). While in the collocated receiver type, the
for prolonging the lifetime of energy-constrained communica- receiver can do both EH and ID processes. Depending on how
tion networks [1]–[4] such as military communications, sensor the receiver coordinates between the EH and the ID processes,
networks [5], submarines, and medical implants [6]. Many the receiver types in the collocated receiver case can be clas-
types of EH schemes according to the energy source have sified into four receiver architectures. Ideal receiver, which
been considered, like solar, piezoelectric, wind, hydroelectric, means the ability to perform both EH and ID processes simul-
and wireless radio frequency (RF) signals EH [7]. taneously. Of course, this architecture has ideal performance,
but due to circuits limitation it cannot be realized. Power
Manuscript received August 22, 2017; revised November 26, 2017 Splitting (PS) receiver [13], where the received signal power
and January 2, 2018; accepted January 22, 2018. Date of publication
January 25, 2018; date of current version May 22, 2018. This work was is divided by a certain ratio referred to as power splitting
supported in part by H2020-MSCA-RISE-2015 under Grant 690750, and in ratio between the EH and ID processes. Time Switching (TS)
part by the Ministry of Higher Education Malaysia and Universiti Teknologi receiver [14], the receiver divides the receiving time between
Malaysia under Grant 4F818, Grant 4J210, Grant 4F901, Grant 4J213, and
Grant 12H35. (Corresponding author: Yamen Alsaba.) EH and ID processes by a time switching factor, where in the
The authors are with the Wireless Communication Centre (WCC), Faculty first time slot the receiver performs EH process, and in the
of Electrical Engineering, Universiti Teknologi Malaysia, 81310 UTM Skudai second time slot it does the ID process. Antenna Selections
Johor, Malaysia (e-mail: yamensaba@yahoo.com; sharulkamal@fke.utm.my;
bruceleow@fke.utm.my). (AS) receiver, in this architecture the receiver has multiple
Digital Object Identifier 10.1109/COMST.2018.2797886 antennas, and a subset of antennas are assigned to decode
1553-877X  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.
1330 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

Fig. 2. Illustration of the described receiver structures. ρ denotes the power


splitting ratio of dividing the signal power in the PS receiver, α denotes the
time switching factor between ID and EH processes in TS receiver.

system with no need to channel condition information at


the transmitter has been introduced in [15]. The principle of
the proposed opportunistic receiver scheme is to start har-
vesting energy when the channel energy exceeds a certain
level by employing random beamforming at the transmit-
ter to generate artificial channel fading. The performance of
the proposed low-complexity receiver is investigated by com-
paring its achievable average information rate and energy
harvested with the conventional periodic TS-based receiver,
the new design shows better rate-energy trade-off, especially
in the case where the harvested energy level is large enough.

A. Beamforming in Energy Harvesting Wireless Networks


In order to be able to harvest energy, the receiver should
have an EH circuit, which is mainly a rectifier [16], [17].
Unlike information decoding circuits, the sensitivity of the
EH process is quite low (−10dBm − 30dBm) [14]. The
need of higher signal energy levels makes the EH pro-
Fig. 1. Illustration of the described EH transmission schemes: in SWIPT, the cess high sensitive to signal decay due to propagation dis-
base station provides the user with information and power simultaneously, in
WPCN, users harvest the power signal in downlink to send information in the
tance, reflection, scattering, and fading (which is normally
uplink, while in WPT, no information exchange the base station is a power higher in the omni-directional transmission case) motivates
charger only. the use of multi-antenna techniques like beamforming in EH
enabled communication networks, for its appealing feature
the information, while the remaining are dedicated for EH. of increasing both wireless power and information transfer
Figure 2, illustrates the different receivers’ architectures. efficiency [18], [19].
A new opportunistic receiver architecture scheme for Beamforming has been widely considered in conventional
multiple-input single-output (MISO) SWIPT EH-enabled wireless networks (with no EH) to exploit the spatial domain
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1331

and gain extra degrees of freedom. Generating pencil beams • In WPT scheme it is proved by [14] that a special
to focus the signal in a narrow direction yielding many scheme of beamforming named “energy beamforming”
advantages such as: is the optimal transmission strategy that leads to the
• Improve the efficiency of wireless transmission: beam- achievability of the maximum harvested power.
forming increases the signal to interference and noise • Beamforming is the optimal solution strategy for the
ratio (SINR) by focusing the signal toward the intended problem of maximizing the rate subject to EH and
receiver, and hence, achieves high data rate by reducing maximum transmit power constraints for Multiple-Inputs
the multipath attention and the interference. Multiple-Outputs (MIMO) and MISO SWIPT technique
• Increased Coverage area: due to the targeted nature of in the case of separated receivers [14].
beamforming, frequencies can be reused more efficiently • In [20] a hybrid form of beamforming is proved optimal
allowing an increased coverage area and higher number for MISO SWIPT with PS-type receiver in two and three
of users. users’ case.
• Reduced interference: interference is usually caused by Furthermore, in physical layer security, beamforming is proved
omni-directional antenna transmissions which radiate in with high significance to this aim. Wyner [21] proved that
all directions, in beamforming scheme, interference is secrecy rate is always positive when the source-destination
less likely to occur due to the directionality in transmis- channel is better compared to the source-eavesdropper chan-
sion. This enhances both the SINR and the efficiency of nel. Particularly, by exploiting the multi-antennas’ extra
frequencies reuse. degrees of freedom, to enhance the SINR at the intended
• Provide value added services: beamforming techniques destination, or to inject an artificial noise (AN) signal
provide some new powerful functionalities and applica- in the communication channel. The AN technique guaran-
tions like, higher quality of service (QoS), higher data tees the information secrecy even when the channel condition
rate, and user localization. of the destination is worse than that of the eavesdropper [22].
• Enhanced security: beamforming provides naturally Beamforming is more likely the optimal transmit scheme if
increased security, as the signals are not transmitted in perfect and global CSI is available, as CSI plays a vital role
all directions like in the conventional omni-directional in the system performance, as channel conditions acquirement
transmission. This creates an extra condition on the eaves- helps the transmitter to construct the transmit beams, select
droppers, meaning that they need to be at the same the optimal transmit scheme, and to change it adaptively as
location or between the two communicating devices in the channel conditions change. In frequency division duplex
order to intercept the transmission. communication systems, a quantization codebook is used, so
However, beamforming technique poses many challenges and the transmitter acquires the CSI, of course increasing the size
further requirements to be implemented like: of this codebook will enhance the CSI estimation, but yet it
• SCI requirements: in order to implement beamforming will increase the overhead of the feedback process. While in
transmission scheme, the CSI of users should be acquired time division duplex communication systems, the transmitter
at the transmitter. SCI acquisition process requires SCI estimates the CSI by taking advantage of channel reciprocity
acquiring technique and feedback mechanism to pro- concept. The CSI at the transmitter is either perfect, where
vide the transmitter with the channel status of the users, the transmitter has full knowledge of the channel conditions,
yielding higher overhead and complexity. or imperfect CSI which can be either stochastic or deter-
• Higher complexity: beamforming technique is far more ministic (worst-case). In imperfect stochastic CSI model, the
complicated than a conventional transmission scheme, CSI uncertainty is modelled as random variables usually with
as it requires signal processing technology to be imple- Gaussian distribution, while it is bounded by a certain thresh-
mented. old in the imperfect deterministic CSI model. However, perfect
• Higher cost: due to the need to antenna arrays, in addition and imperfect CSI acquisition isn’t trivial in EH-Enabled wire-
to signal processing blocks, beamforming is more expen- less systems, as it isn’t necessary that the EH receiver has an
sive to be deployed in terms of money and consuming ID circuit to feedback its own CSI to the energy transmit-
power. ter. Furthermore, the CSI of the ID receiver in EH systems is
• Larger size: beamforming utilizes antenna arrays, mak- highly impaired because of the interference of both informa-
ing the system’ size larger than traditional systems. This tion and power signals; hence its CSI might be imperfect, and
means that large base stations are needed to make this robust implementation is needed. In addition, in physical layer
method efficient. security application, the CSI of the eavesdroppers is very dif-
The appealing feature of beamforming high received SINR, ficult to acquire as these malicious nodes are usually hidden
makes it the potential candidate transmission scheme in EH- and not active. Moreover, the formulation of optimal transmit
enabled wireless networks, as EH receivers require higher or receive beamforming weights requires higher computational
signal level than that of the ID receiver. In a response to the complexity than that of the conventional networks, as another
question about the optimal transmission scheme in energy EH energy constraints are added in beamforming design process,
wireless communication systems, the problem of increasing resulting in degradation in the network performance.
both energy and information transfer efficiency for the differ- Table I, represents the advantages and disadvantages of
ent EH broadcasting strategies is formulated and the solutions deploying beamforming in both conventional and EH-enabled
yield the following results. wireless networks. Many criteria have been considered in
1332 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

TABLE I
C OMPARISON OF B EAMFORMING D EPLOYING IN C ONVENTIONAL AND EH-E NABLED W IRELESS N ETWORKS

beamforming designing in EH wireless networks, such as max- objective. A system model on beamforming EH commu-
imizing system throughput, minimizing total transmit power, nication network is provided in Section II. Section III
maximizing ergodic capacity, minimizing the outage probabil- presents beamforming in EH-enabled broadcast channel (BC).
ity, and information and energy trade-off. These optimizing Section IV introduces the usage of beamforming in EH-
objectives are usually accompanied by different practical con- enabled relay channel (RC). Section V analyses the imple-
straints such as the total transmit power should be under a mentation of beamforming in EH-enabled Cognitive Radio
certain value referred to as maximum transmit power, or the Networks (CRN). Section VI investigates the CSI requirements
power received at the EH receiver should more than a thresh- and techniques for deploying beamforming in EH-enabled
old in order to initiate the EH process, this condition is referred communication systems. Section VII provides the research
to as EH constraint. The SINR at the ID receiver should be challenges and future work concerning the application of
above a quality of service threshold so the ID receiver is beamforming EH technique in the envisioned systems, and
able to decode the information message correctly, this con- the open research areas that need further investigation.
straint normally named SINR constraint or QoS constraint. In
most of the cases the optimization problems with the joint
II. S YSTEM M ODEL
constraints are non-convex problems, that can be solved by
means of Semi-Definite Programming (SDP) or Semi-Definite We consider a multi-user downlink EH-enabled communi-
Relaxation (SDR) [34], Second Order Cone Programming cation system, where a base station is equipped with N >
(SOCP) [35], sub-optimal algorithms, or transferring the non- 1 antennas serves K single-antenna users. We assume that
convex problem into a convex one by several techniques beamforming is employed by the base station to transmit infor-
like, Successive Convex Approximation (SCA) [36]. Table II mation and power signals via SWIPT technique to ID and EH
summarizes the list of acronyms used in the paper. users. Hence, the complex baseband transmitted signal at the
base station is given by
B. Comparison With Related Survey Articles 
K
In this paper, we present most of the contemporary state- x= wk sk , (1)
of-the-art on exploiting beamforming technique in EH-enabled k=1
wireless communication and its physical layer security appli- where sk ∼ CN (0, 1) denotes the transmitted symbol and wk
cation. Due to the importance of this topic a lot of literature is the transmit beamforming vector for user k.
has appeared after [37], that we want to cover in order to Let hk denote the conjugate channel row vector between the
demonstrate the recent progress in this field. Zhao et al. [38] base station and user k multiplied by the path loss. Then, the
focus on investigating the interference impact in EH systems, received signal at user k before any energy harvesting process
while [19] introduces a full performance survey of WPCNs. is expressed as
Physical layer security in SWIPT communication systems
is investigated by [39]. Unlike most of the surveys on EH yk = hH
k x + nk , k = 1, 2, . . . , K, (2)
communication systems [37], [40], we focus on wireless multi-
antenna EH system and its applications in physical layer where nk ∼ CN (0, σk2 ) is the additive white Gaussian noise
security, as from the discussion above beamforming is proved (AWGN) introduced by the receiver antenna at user k.
with high significance on both information and energy transfer The received signal will be treated according to the corre-
efficiency in EH wireless networks. sponding receiver architecture. We will consider the different
receiver types: PS, TS, AS and separated-receivers architec-
C. Organization of the Paper tures.
The rest of this paper is organized by taking into consid-
eration the different EH-enabled network topologies where A. Power-Splitting Receiver
beamforming has been deployed, while each network topol- In PS architecture, the received signal power at user k is
ogy section is classified according to the beamforming design divided between ID and EH processes with a splitting factor
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1333

TABLE II
TABLE OF ACRONYMS

0 < ρk < 1. Then, the signal for ID at receiver k is modelled as Accordingly, the SINR for ID at the user k is expressed as
⎛ ⎞
k wk |
| hH 2
√  SINRk =
, ∀k, (9)
yID ⎝ H
k = ρk hk wk sk +

k wi si + nk + zk , ∀k,
hH | hH
k wi | +σk
(3) 2 2
i=k
i=k
and the average throughput of user k
where zk ∼ CN (0, δk2 )
is the additional AWGN introduced by Rk = αk log2 (1 + SINRk ), ∀k. (10)
RF to baseband conversion at user k.
The signal for EH is given by The harvested energy by the user k is given by
 K
 K 
  Ek = ηk (1 − αk )T | hk wk | +σk , ∀k,
H 2 2
(11)
yk = 1 − ρk
EH
hk wk sk + nk , ∀k.
H
(4)
k=1
k=1
and the average harvested power during the given duration
Accordingly, the SINR for ID at the user k is expressed as time of transmission T
 K
ρk | hH
k wk |
2 
SINRk =
, ∀k, (5) Pk = ηk (1 − αk ) | hk wk | +σk , ∀k.
H 2 2
(12)
ρk i=k | hH
k wi | +ρk σk + δk
2 2 2
k=1
and the throughput of user k
C. Antenna-Switching Receiver
Rk = log2 (1 + SINRk ), ∀k. (6) In AS scheme, we assume that each user is equipped with
M > 1 antennas. AS techniques splits the receiving antennas
The harvested power by the EH of user K is given by into two subsets. First subset is connected to the ID circuits,
 K while the second subset is connected to the EH circuits.

Pk = ηk (1 − ρk ) | hH The total signal received at user k in AS approach is
k wk | +σk , ∀k,
2 2
(7)
k=1 given by

where ηk ∈ (01] is the energy conversion efficiency factor of yk = Hk x + nk , ∀k, (13)


the EH circuits at user k. where Hk ∈ CM×N denotes the channel response between the
base station and the user k, nk ∼ CN (0, I) is a circularly
B. Time-Switching Receiver symmetric AWGN vector introduced by the receiver antenna
In TS scheme, for total transmission duration presented by a at user k, with the mean vector 0 and the covariance matrix I.
block time of T, during the first phase of duration αk T, where The signal for ID at receiver k in AS approach can be
0 < αk < 1 is the time switching parameter, the receiver k expressed as
decodes the own intended information message. In the remain-
k = Qk Hk x + Qk nk , ∀k,
yID (14)
ing time of duration (1 − αk )T, the receiver k harvests energy
from the received signal during the given time. where Qk ∈ RM×M is the diagonal AS matrix, where the
The signal for ID at receiver k is expressed as diagonal entries are either 1 or 0. If the diagonal entry
 j, j ∈ {1, 2, . . . M} of the matrix Qk is 1, the antenna j is
k = hk wk sk +
yID k wi si + nk , ∀k.
H
hH (8) selected for ID process, otherwise this antenna is selected
i=k for EH.
1334 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

The throughput of user k is given by E. Receiver Architecture Schemes Performance


⎛ ⎛ ⎞−1 Comparisons among the different receiver’s architectures are
⎜  carried out with respect to rate-energy region in MIMO chan-
Rk = log2 det⎝I + ⎝I + Qk Hk wi wH H H⎠
i Hk Qk nel [14]. The comparisons demonstrate that the PS architecture
i=k
attains the upper bound of the rate-energy region (provided

by the ideal receiver scheme) and thus optimal, if the process
H H⎟ of RF to baseband conversion is noise free. In practice, the
k Hk Qk ⎠. ∀k,
Qk Hk wk wH (15)
noise-free conversion condition is hard to be perfectly met.
Therefore the upper bound of the rate-energy region cannot
and harvested power by the EH of user K in the AS approach be achieved in practical implementations [14]. A special case
is given by [14] of PS-type receiver named uniform PS (all antennas have the
 same power splitting ratio) is compared with different vari-
Pk = ηk Tr Q̄k Hk PT HH
k Q̄k , ∀k,
H
(16) ations of the TS schemes. It proves better than fixed power
constraint TS receiver, but performs worse than the TS scheme
where Q̄k ∈ RM×M is the complementary matrix of Qk , i.e., with flexible power constraint. However, no theoretical study
the diagonal entry j, j ∈ {1, 2, . . . M} in the matrix Q̄k is 1 yet compares the best TS scheme (under flexible power con-
if the j-th

diagonal2 entry in the matrix Qk is 0 and vice versa. straint) and the general arbitrary PS scheme (random values
PT = K k=1 | wk | is the transmit power. of PS ratios). Furthermore, AS scheme is proved to have
the smallest rate-energy region and hence with the poorest
D. Separated Receivers performance [14].
In separated receivers scenario, the user is either energy In [41], comparisons between PS scheme and different
harvester or information decoder. Thus, two subsets of users deviations of TS schemes like time-division mode switching
are found, subset of ID user KI , and EH users subset Kε . The (TDMS), time-division multiple access (TDMA) are per-
transmitted signal can be expressed as formed. The comparisons are carried out with respect to
  maximum throughput in MISO interference EH communica-
x= wi si + vj ej , (17) tion system. The results illustrate that PS performs best when
i∈KI j∈Kε the interference is well controlled in the system by the extra
degrees of freedom (the base station’s number of antennas are
where si ∼ CN (0, 1) and ej ∼ CN (0, 1) denote the larger than the number of users), otherwise TS and its devia-
information-bearing signal intended to the i-th ID user and tion schemes provide better performance than that of the PS
the energy-carrying signal targeted for the j-th EH user respec- scheme. Furthermore, it is shown that there is no dominant
tively, wi and vj represent the beamforming vector for the i-th scheme for all considerations and setups, i.e., each channel
ID user and the j-th EH user respectively. type has its own optimal receiver design. For example, it is
If a user k ∈ KI , i.e., an ID user, then the received signal shown that TS is dominant in interference channel, while PS
is expressed as is ideal in interference controlled scenarios.
 
k =
yID k wi si +
hH k vj ej + nk , ∀k ∈ KI .
hH (18)
i∈KI j∈Kε F. Beamforming Design Objectives
Accordingly, the SINR for ID at the user k is expressed as Beamforming transmit vector w can be optimized in order
to achieve several

design goals such as minimizing total trans-
k wk |
| hH 2 mit power ( K k=1 | wk | ) under different constraints like the
2
SINRk =

, ∀k ∈ KI ,
| ID QoS constraint (SINRk > thID k , ∀k), or and the minimum
k wi |
hH + j∈Kε | hH
k vj | +σk
2 2 2
i =k,i∈KI
(19) harvested power constraint (Pk > thEHk , ∀k). This optimization
problem can be formulated as
and the throughput of user k is given by

k
Rk = log2 (1 + SINRk ), ∀k ∈ KI . (20) minimizewk PT = |wk |2
k=1
If user k ∈ Kε , i.e., energy harvester, then the received signal subject to SINRk > thID
k , ∀k,
at EH user is given by
Pk > thk , ∀k
EH
(23)
k = hk x + nk , ∀k ∈ Kε ,
yEH H
(21)
The aforementioned optimization problem has been tack-
and the harvested power is expressed as led by several approaches in different network topologies
⎛ ⎞ such as broadcast channel (Section III-A) and relay channel
  (Section IV-B).
Pm = ηk ⎝ | hH
k wi | +
2
| hH 2⎠
k vj | +σk , ∀k ∈ Kε .
2
The weights of the beamformer are optimized in other
liter-
i∈KI j∈Kε ature to maximize the sum rate of the system RT = K k=1 Rk
(22) subject to different constraints like total transmit power and
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1335


K
k=1 | wk | < Pmax and harvested power con-
minimum 2 than energy threshold needed to initiate the EH process. This
straints. Hence, the optimization problem is expressed as goal is achieved in [20], by first using fixed beamforming
schemes like, normal or regularized zero forcing, and max-

K
maximizewk RT = Rk imum ratio transmission. In zero forcing case, a closed-form
k=1
always feasible solution has been derived, while for the maxi-
mum ratio transmission and regularized zero forcing schemes,

K
subject to | wk |2 < Pmax , the problem is transformed into a SOCP formulation, that
k=1 leads to an optimal solution. In addition, a hybrid beamform-
Pk > thEH ing scheme is introduced (maximum ratio transmission-zero
k , ∀k (24)
forcing), and proved to be superior to all other fixed weights
The total sum rate maximization problem is solved by a variety beamforming schemes. In addition, in order to find the optimal
of techniques and algorithms in the different networks types beamforming scheme, approximate SDP formulation, that pro-
(Sections III-C, IV-A, and V-A). vides optimal solutions for two and three users cases has been
Furthermore, beamforming optimization for guaranteeing derived.
information security in wireless communication systems has The same problem is solved in [42], by joint design of
gained a lot of attention in the literature. In the secrecy sce- beamforming vector weights and the receiver’s power splitting
nario, the most common beamforming design objective is ratio (JBPS). After deriving the sufficient and the neces-
maximizing the secrecy rate in the system in the presence sary conditions in order to have a feasible solution, the
of eavesdroppers. This design objective has been tackled by SDR technique is applied to achieve the optimal solution.
a considerable amount of work in the different channel types: Furthermore, based on beamforming and PS ratio separa-
broadcast (Section III-D), relay (Section IV-E) and cognitive tion technique, two suboptimal, but with low computational
radio (Section V-D). Maximizing secrecy rate is carried out complexity are proposed. The suboptimal algorithms start by
with guaranteeing different constraints such as the total trans- fixing the beamforming schemes (zero forcing and optimal
mit power constraint and the minimum harvested power. This SINR respectively), then searching for the optimal PS ratio.
optimization problem can be formulated as Comparisons between the optimal algorithm and the two sub-
optimal algorithms, showed that the performance gap between
maximizewk Rk − Re
them increases as the transmission power increases. JBPS

K
is also used in [43] to solve the total power minimization
subject to | wk |2 < Pmax ,
problem by providing a relaxation algorithm with its sufficient
k=1
optimality condition. This is in addition to a low-complexity
Pk > thEH
k , ∀k (25) suboptimal algorithm that can be applied in the case when
where Re is the rate of the eavesdropper. the optimality condition is not satisfied. Unlike [42]–[44]
Some approaches optimize jointly the beamforming vector considers using SOCP technique instead of the conventional
w and the PS ratio ρ or the TS factor α in what referred SDR in order to tackle the JBPS problem. The new proposed
to JBPS or JBTS respectively to achieve the desired design SOCP-based relaxation solves the JBPS problem by decom-
objective. posing it into a master problem (solved via sub-gradient
method), and a set of sub-problems (solved by using coordi-
III. B EAMFORMING IN E NERGY H ARVESTING nate descent method). Comparisons between the new proposed
B ROADCAST C HANNEL SOCP-based method and the SDR technique show that feasible
solution to the joint optimization problem is always guaran-
Using beamforming in EH-enabled Broadcast Channel (BC) teed in the SOCP-based method, with lower complexity, and
has been widely investigated, as this channel model is the very close performance to the minimum transmit power.
most common model, and interference, multipath, and large Previous work considers perfect CSI at the base sta-
distances in such channels necessitate multi-antenna tech- tion when designing the beamforming vector, which is very
nique deployment, to improve information and energy transfer difficult to realize in practice. SCI uncertainty makes the
efficiency. Work on Beamforming in EH-enabled broadcast JBPS optimization problem a homogeneous quadratically con-
channel can be mainly divided according to the beamforming strained quadratic problem, which is difficult to be solved.
design objectives. Beamforming design under both perfect and imperfect deter-
ministic CSI scenarios at the base station is being addressed
A. Beamforming Design for Total Transmit Power in [45]. Tight robust and non-robust SDR-based solutions in
Minimization perfect and imperfect CSI scenarios respectively, are obtained,
A lot of literature investigates solving the non-convex by solving the optimization problem. The uncertainties of
problem of designing the beamforming vector in order to CSI are considered deterministic and modelled by Euclidean
minimize the total transmit power in a MISO SWIPT BC, ball-shaped uncertainty set in [46] and [47]. In [46], two
under two constraints. firstly the SINR should be greater than robust solutions based on loose approximation (SDR-based
a certain threshold so the information receiver will be able technique) and tight approximation (Lagrangian multiplier
to decode the information message. The second constraint method) have been proposed respectively. While a robust solu-
implies the energy received at the energy receiver being greater tion using a penalty function algorithm is provided by [47],
1336 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

based on reverse convex non-smooth optimization. The com- are provided. Furthermore, a practical two iterative algorithms
parison between SDR based and penalty function based tech- that provide local optimal solutions and a feasible pre-coding
niques yields that the later achieves a more reliable and better vector for the optimal solution have also been provided. The
solution. previous work is extended to different imperfect CSI models
In [48] and [49], the CSI uncertainties are modelled by in [55]. In ellipsoidal-bounded error CSI case, the constraints
a deterministic norm-bounded uncertainty CSI model. The are transformed into quadratic matrix inequality forms, that
optimal algorithm that solves the JBPS problem provided can be relaxed as an SDR problem. For local CSI and limited
in [48] is achieved via a two-step technique. First, by fixing information exchange cases, a decentralized optimal algorithm
the PS ratio, the problem is transformed into an SDR- has been proposed based on the alternating direction method
based problem that can be solved by standard interior point of multipliers.
algorithms. Second, the optimal PS ratio is found by one/multi- Choi and Kim [56] propose a model for deploying beam-
dimensional search over the minimum target of the transmit forming in WPCN BC, where each receiver has a data queue
power. While in [49], the original problem is first relaxed using and an energy queue. The problem of minimizing the total
the SDR technique, then the interior point algorithm is applied transmit power subject to stabilized data queue at each node
to obtain the optimal solution. The CSI uncertainty is mod- is solved by an online Lyapunov optimization-based algorithm.
elled by stochastic Gaussian channel uncertainties in [50]. In This online algorithm steers the beamforming vector toward
order to tackle this non-convex problem, the authors exploit the the receivers with low power queue, and schedules large data
Bernstein-type inequality and Gaussian error function to trans- queues receivers for data transmission. The proposed algo-
fer the optimization problem for Gaussian statistical channel rithm is proved optimal in both random and deterministic
uncertainties into deterministic one. Moreover, a more general channel scenarios.
case of channel uncertainties, i.e., channel uncertainties based
on random distribution is investigated, and a robust solution is
proposed, by exploiting the conditional value at risk functional
and the SDR technique. B. Beamforming Design for Energy Efficiency Maximization
JBPS problem in a MIMO SWIPT EH-enabled BC is first Another important design objective for beamforming in BC
considered in [51] under both perfect and imperfect CSI mod- is maximizing the system energy efficiency in bits per Joule,
els. The total transmit power is minimized by an SDR-based or energy-rate trade-off to control the data and energy flow in
tight algorithm in the perfect CSI case, and a worst-case the system. Subject to both SINR and EH constraints with PS-
design-based robust algorithm in the erroneous CSI case. The type receivers, Shi et al. [57] deploy zero-forcing beamforming
model in [51] is extended to a massive MIMO channel in [52] scheme to solve the non-convex energy efficiency problem
under both perfect and imperfect (stochastic with Gaussian dis- by means of Lagrangian relaxation and Dinkelbach methods.
tribution) CSI models. The optimization problem in this model The solution provided by the proposed algorithm is closed-
is solved by using the asymptotic orthogonality of the channel form, and proved always feasible under some conditions. The
vectors technique. The new solution proved to be with less proposed algorithm can be implemented under zero-forcing
computational complexity than the conventional SDR-based beamforming scheme only, hence, other techniques should be
solution, and independent of the number of transmit antennas. addressed to handle the power efficiency problem in more
A more realistic constraint in the total transmit power general scenarios. JBPS is suggested by [58] to solve the
minimization problem with separate-type receiver is consid- energy efficiency maximization problem locally. Subject to
ered in [53], where the sum power constraint is replaced both maximum total transmit power and EH constraints, the
with a per-receiver power constraints, as each receiver has its non-convex optimizing problem solved by an iterative-based
own antenna and amplifying block. Under this new individual algorithm that guarantees the achievability of a local stationary
power constraints, two optimization problems are investigated. solution. Comparisons with other suboptimal known solutions
Firstly, minimum power transfer for the downlink channels such as equal PS, and zero-forcing equal PS, demonstrate that
with a single antenna at each receiver (MISO). Secondly, the proposed algorithm has superior performance in terms of
capacity-achieving transmitter design for downlink channels convergence behaviour and the number of transmit antennas.
with multiple antennas at each receiver (MIMO). Using the In [59] the problem of information and energy harvest-
downlink-uplink duality, the above problems can be trans- ing transmission efficiency for information and energy users
ferred into an equivalent minimax optimization problem. The respectively is investigated. The problem is solved for two
downlink-uplink duality is very significant as many difficult fixed beamforming schemes (zero forcing and maximum
downlink optimization problems can be transferred into a beamforming), by proposing an iterative power allocation
simple uplink dual problem which can be solved in more algorithm to achieve optimal power strategy under both beam-
effective ways. The same problem is investigated in [54], by formers. Comparisons between the two different beamforming
taking advantage of the constructive interference and exploit- schemes reveal that maximum beamforming can provide bet-
ing the data-aided beamforming technique. After deriving the ter information and energy transfer efficiency. Chen et al. [60]
necessary and sufficient feasibility conditions for the design generalize the model of [59] to long distance energy and
problem, upper bounds to the optimal solution are obtained via information transfer in large-scale MIMO EH-enabled BC.
SOCP and SDP techniques. In addition, based on exponential The problem of maximizing the energy efficiency of infor-
complexity mixed-integer SOCP, tight upper and lower bounds mation transmission, subject to delay constraints, is solved by
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1337

jointly optimizing the transmit power and the transfer duration a global solution of the original problem. Similar to [64], the
yielding an efficient energy resource allocation strategy. objective in [65] is maximizing the worst energy receiver har-
The achievable rate-energy regions for TS and PS schemes vested power, but subject to both minimum SINR constraints
in MIMO channel are presented in [14] for the co-located at the ID receivers and the total transmit power constraints.
receiver case, where the ID and EH receivers have an iden- Again, the original optimization problem is reformulated as
tical MIMO channel. For the case where the ID and the SDP problem, which is solved optimally via standard inte-
EH are separated, i.e., have different MIMO channels, the rior point methods. Furthermore, based on bisection method,
optimal beamforming scheme that achieves different maximal an iterative algorithm that provides near upper bound robust
rate-energy transmission trade-offs is presented, and hence, beamforming has been also provided.
the rate-energy region is defined for this case. While the Feng et al. [66] address the distributed beamforming scheme
performance of the energy transmitter in WPCN is enhanced in MISO SWIPT EH-enabled BC. The problem of maximizing
in [61], via sending the CSI at the receiver using a quantiza- the harvested energy subject to the individual power con-
tion codebook. Furthermore, a lower bound of the average straints at each transmitter, and the SINR constraints at ID
information rate is derived for a given codebook size. By receivers, in the corresponding system is solved by means of
exploiting the relationship between the EH time, the trans- iterative singular value decomposition method.
mission time ratio, the feedback size, and transmission power,
the optimal energy and information transfer trade-off has been
achieved. Furthermore, the authors provide the optimal rate- C. Beamforming Design for Throughput Maximization
energy trade-off scheme for both perfect and imperfect CSI The capacity region of a MISO SWIPT EH-enabled BC
cases. is studied in [67], by finding a solution to the problem of
The receiver’s ability of cancelling interference signals in maximizing the weighted sum rate under both total transmit
MISO SWIPT EH-enabled BC is proposed in [62] and [63], power constraints and the EH constraints at all EH receivers.
where there are two types of ID receivers, able or unable of In order to tackle this problem, a new optimal iterative algo-
cancelling the interference signals coming from the energy rithm is proposed by extending the general BC multiple access
signals. In [62], the beamforming scheme of joint informa- channel duality with the help of the ellipsoid method. The
tion and energy transmit is designed, in order to maximize capacity region provided by the proposed algorithms repre-
the weighted sum power delivered to all EH receivers sub- sents a performance upper bound to all possible beamforming
ject to a given minimum SINR constraints at both types ID and coding schemes that can be applied in MISO SWIPT EH-
receivers. The optimization problem is a non-convex quadrat- enabled BC. The problem of capacity maximization subject to
ically constrained quadratic program (QCQP), that can be EH constraints in MIMO channel under imperfect CSI model
solved optimally via SDP and uplink-downlink duality. The is discussed in [68]. The beamforming covariance matrix is
solution shows that for the receiver without interference can- optimized at transmitter nodes with respect to both minimum
celling capability, the optimal scheme is achieved without mean-square-error and mutual information metrics to achieve
dedicated energy beam, while for the second receiver type the optimal performance.
case, employing only one energy beam is optimal. The objec- Huang et al. [69] calculate the average throughput and
tive in [63] is to minimize the average total transmit power provide it in closed-form expressions in high SNR regimes,
under EH constraints and SNR and SINR constraints for in both delay-limited and delay-tolerant transmission cases.
ID receivers with and without cancelling interference ability Furthermore, closed-form expressions for throughput-optimal
respectively under worst-case CSI model. For ID receivers EH time at high SNR regimes have also been provided. The
without cancelling capability, upper and lower bounds are analysis demonstrates that the transmit power and the anten-
provided by a robust Lagrangian multiplier method-based nas number are the main parameters in the system, and that
algorithms, while an optimal robust SDR-based algorithm is the throughput is in proportional relation to them, while the
introduced in the Id receivers with cancelling interference value of the optimal energy harvesting time is inversely pro-
capability case. portional to them. The work in [69] is extended to TS-type
The maximization of the worst-case harvested energy or receiver under Nakagami-m fading in [70], and the average
SINR ratio for EH receivers or ID receivers respectively, is throughput is derived in analytic expressions for both modes
another optimization goal that has been taken into consid- (delay tolerant and delay intolerant). Additionally a very
eration to guarantee that all EH receivers and ID receivers accurate closed-form solution of the optimal time switching
are able of doing EH and ID processes to increase the technique, which maximizes the average throughput has been
system fairness. Subject to the minimum rate constraints also proposed. The study that is also conducted for co-channel
for all ID receivers, and under imperfect CSI assumption, interference impact on the average throughput shows that
Xiang and Tao [64] investigate the problem of maximiz- this impact (either positive or negative) is highly dependent
ing the worst-case harvested energy, which is a semi-infinite on the system and propagation parameters like, transmit and
non-convex quadratically constrained programming (QCQP) interference power, path loss, fading, and network geometry.
problem. As the non-convex QCQP problem is NP-hard, it In [71], convex restriction approaches are used in order
is reformulated into a relaxed SDR problem by the means to solve the problem of maximizing the outage constrained
of certain transformation techniques. Moreover, it has been rate subject to minimum average energy transfer to the
proven that this relaxation method is tight and always yields EH receivers constraints under imperfect CSI assumption.
1338 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

Furthermore, the authors exploit the temporal degrees of free- problem is solved by finding the optimal beamforming scheme
dom of the stochastic beamforming approach, that give the first using the SDR technique, then via one/multi-dimensional
beamforming the freedom to vary randomly in time follow- search algorithms, the optimum PS factor design is found.
ing a certain disruption for maximizing the outage-constrained Realizing SWIPT in OFDMA EH-enabled broadband com-
rate. Comparisons demonstrate that the stochastic beamform- munication system is addressed in [77]. The problem of
ing approach has a better rate-energy region than those of other maximizing the system throughput subject to the receiver and
conventional schemes. transmitter power constraints is solved by proposing power
In MISO WPCN BC a TS-type receiver is adopted, where control strategies. Different scenarios have been considered:
an access point transmits energy signals in the downlink to single and multi-user systems, uplink/downlink information
multiple nodes in the first time slot, and in the second time transfer, and fixed and variable data rate. For fixed information
slot the nodes are sending information back to the access point rate, the optimal power control gives the receivers the needed
in the uplink. Time switching parameter is the most crucial fac- power for decoding information in ascending order until the
tor to be tuned in order to optimize the system performance, entire amount of power is over. While in the variable infor-
in addition, similarly to JBPS, TS period can be jointly opti- mation rate, the power control policy is a water filling-based
mized with the beamforming vector (JBTS) to achieve certain algorithm.
system design goals. Liu et al. [72] investigated maximizing Wu et al. [78] propose a new scheme to solve the
the minimum throughput for all the users by optimizing the optimization problems in SWIPT MISO BC called rank-
time allocation scheme in both downlink and uplink. This non- 2 transmit beamformed Alamouti space-time code scheme.
convex optimization problem is solved optimally via two step Instead of SDR-based technique which can be considered as
algorithm. First, the downlink/uplink time allocation is fixed, 1-rank, the new scheme combines transmit beamforming and
and by using the techniques of alternating optimization and space-time coding techniques, leading to rank-2 SDR, which is
the non-negative matrix theory the optimal transmit energy dealt with by using rank-2 SDR technique, followed by a rank
beamforming scheme, downlink power allocation, and receive reduction procedure. The analysis of the worst-case approxi-
beamforming scheme that maximize the minimum SINR of mation accuracy revealed that the accuracy degrades with the
all users are obtained (this problem is non-convex but can rate of root square of the number of users served, which is
be converted to a spectral radius minimization problem). a considerable improvement over the rank-1 approximation
Then, the optimum global solution is found by perform- where the accuracy degrades with the rate of the number of
ing one-dimensional search over the optimal downlink/uplink the users.
time allocation factors. The high complexity of the optimal Table III represents a summary of literature regarding
proposed linear MMSE-based algorithm incites the authors to exploiting beamforming in EH-enabled broadcasting channels.
propose two other less complex suboptimal algorithms based
on zero-forcing beamforming scheme. Sun et al. [73] gener-
alise the previous work by considering perfect and imperfect D. Beamforming Design for Implementing Physical Layer
CSI scenarios at the energy transmitter, by solving the problem Security
of maximizing the system sum-throughput in this system. In The open nature of wireless communications necessitates
the case of perfect CSI, a time complexity reducing algorithm taking measurements in order to guarantee information secu-
that yields a semi-closed form solution has been proposed, by rity, such as Code Division Multiple Access (CDMA) tech-
exploiting the strict concavity of the JBTS problem. In the case nique, frequency hopping, and data encryption. But fully
of stochastic Gaussian CSI errors, the problem of maximiz- secure communications can be only achieved by exploiting
ing the system throughput under both SNR outage probability physical layer security [21], [79]. Physical layer security is a
constraints, has been solved by transferring the non-convex paradigm where secrecy is achieved by exploiting the physical
problem into a convex one using Bernstein-type probability layer signal format and physical characteristics of the commu-
inequality. The resulting convex problem can be efficiently nication system, such as noise, fading, and the interference,
solved by the interior point method. where signal processing techniques like beamforming are
Max-min SINR subject to both total transmit power and deployed for turning the communication channel’s imperfec-
EH constraints in a dense small cell with PS-type receivers tions into a source of secrecy. Physical layer security is realised
problem is investigated in [74] and [75]. The optimization in [79] by using the idea of symmetric key encryption systems,
problem in this scenario turns to be highly non-convex and and it is proved that perfect secrecy can be achieved if the
even conventional approaches like SDR-based algorithms will entropy of the encryption key (a measure of random variable
not be able to solve it. In [74], the concept of spectral uncertainty) is as large as that of the confidential message.
optimization in order to change the rank-one constraints into Wyner introduced the concept of wiretap channel in which
a single reverse convex non-smooth constraint is used, and the communication channel is secured without the usage of
an iterative SDR-based algorithm that leads to optimal solu- restoring secret keys.
tion have been provided. While in [75], an iterative quadratic Physical layer security in EH-enabled wireless
programming-based algorithm that solves the beamforming network [80], [81] has drawn further attention, as increasing
vectors directly is provided. The problem of maximizing the the signal level to fulfil the EH receivers needs of power,
average SINR subject to the individual EH constraint under results in more unintended receiver to get the transmission
imperfect CSI assumption is addressed in [76]. The JBPS and apply decoding on it. In addition, the EH receiver
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1339

TABLE III
B EAMFORMING IN EH-E NABLED B ROADCAST C HANNELS

14

itself can be able of decoding information and thus could confidential messages, or merging the original signal with
be a potential eavesdropper. Many techniques have been AN signal known a priori at the intended receiver in order
considered to guarantee a secure communication such as to maximize the secrecy rate, which can be defined as the
1340 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

difference between the legitimate receivers’ information not satisfied, a suboptimal secure multicast algorithm based
rate and that of the eavesdroppers. In this subsection we on rank-two Gaussian randomization is proposed. The same
summarize research that addresses deploying beamforming to optimization problem is solved in MISO SWIPT video BC
guarantee the information secrecy in EH-enabled BC. in [89], by replacing the non-convex constraints by determin-
Maximizing the secrecy rate of the ID receiver problem istic ones, which leads to a smaller feasible solution set, then
under the EH constraints at each EH receiver in MISO BC SDR technique is applied to obtain an upper bound power
is solved in [82] by a two-step procedure. The first step is allocation solution to the new formulated problem. In addi-
fixing the target SINR ratio for the EH and ID receivers tion to the sufficient conditions so this solution is the global
respectively, and deriving the optimal beamforming scheme optimal solution. The previous work is extended in [90],
and the power allocation solution respectively by means of under different statistical channel uncertainty models. The
SDR techniques, with no loss of optimality. In the second step, non-convex optimization is transferred into deterministic forms
one-dimensional search over the optimal SINR target is per- using Bernstein-type inequalities, and a rank-one robust secure
formed for all EH receivers and the ID receiver. AN technique beamforming scheme is obtained. Furthermore, a suboptimal
is used in [83] to solve the same problem subject to both maxi- low-complexity iterative algorithm that uses successive con-
mum total transmit power and minimum EH constraints. Under vex approximation method has been also proposed, in order
both CSI uncertainties deterministic and statistical assump- to transform the main problem into a SOCP problem.
tions, a robust algorithm that jointly designs the beamforming Zhang et al. [91] address the JBPS design for maximiz-
scheme and the AN covariance matrix is provided. The optimi- ing the secrecy energy efficiency subject to information delay
sation problem is divided into a series of SDR-based problems sensitivity constraints at the ID receiver and the EH con-
referred to as inner loop. Using the one-dimensional search, straints at the EH receiver. The JBPS design problem is
the optimal solution of the outer loop is provided in the non- solved by first fixing the SINR value, and then using the
bounded CSI uncertainty scenario. In the statistical channel SDR technique the optimal solution for the reformulated
uncertainty model, a suboptimal robust algorithm is proposed problem is obtained. After this the global optimal solu-
using Bernstein-type inequality. The same problem proposed tion for the original problem is found by performing a full
in [83] is solved in [84] via a two level optimization-based search over all possible combinations of joint PS ratios and
algorithm which leads to rank-one solution in the perfect CSI beamforming schemes. Furthermore, a suboptimal, but lower
case, while the successive convex approximation is used to complexity algorithm was also introduced based on the genetic
provide a robust scheme in the imperfect CSI case. algorithm.
Shi et al. [85] consider beamforming design for guaran- The problem of total transmit power minimization in secure
teeing physical layer security a MIMO BC, where different EH-enabled BC is tackled in [92], under the following con-
data streams scenarios are recognised. For single-stream and straints: SINR constraints at the legitimate receiver, SINR and
specific full-stream data cases, global optimal solutions based outage probability less than maximum level at the potential
on SDR technique are suggested. In the general case, inex- eavesdroppers, the power transferred to the idle legitimate
act block coordinate descent-base algorithm that provides an receivers more than certain value, with imperfect CSI of the
optimal solution to arbitrary number of streams is proposed. potential eavesdroppers, and no CSI of the passive eavesdrop-
In addition, the authors extended the inexact block coordi- pers at the transmitter. To handle this non-convex optimization
nate descent algorithm to the multiple energy receivers’ case problem, first, the non-convex requirements are replaced with
by linearising the EH constraints for all the EH receivers convex ones, then SDR technique is used to achieve the
with respect to beamforming scheme. The optimal solution optimal global solution of the new problem. A suboptimal,
is obtained, using any efficient interior-point algorithms. A but low complexity resource allocation algorithm has been
beamforming design for enhancing information security in a also proposed, to provide secure connections and easy wireless
multi-cell EH BC is considered in [86]. Where the infor- power transfer. The same design objective is addressed [93],
mation security of the far users is threatened by an already by jointly optimizing the beamforming scheme, the PS ratios,
existed eavesdropper or by any near user. JBPS is adopted to and the covariance matrix of the AN. A joint design of trans-
maximize the minimum secrecy rate under EH constrains. The mit and energy beamforming is adopted in [94] to maximize
non-convex optimization problem is tackled by novel iterative the harvested energy of the EH receivers while maintaining the
path-following algorithms, where in each iteration one simple SINR of the ID receiver high enough. The SDR-based joint
convex quadratic program is involved. design is done under perfect and imperfect CSI assumptions,
Another objective has been taken into consideration by sev- and proved tight.
eral literature which is minimising the total transmit power An optimal energy-efficient power allocation scheme for
while guaranteeing both secrecy rate and harvested energy secure BC is proposed in [95], by solving the problem of
above certain threshold which is a non-convex problem. In maximizing the secrecy energy efficiency subject to informa-
order to tackle this problem, the SDR technique is exploited tion delay sensitivity constraints at the ID receivers and the
in [87] to obtain a rank-one beamforming optimal design. EH constraints at the EH receivers. The problem is solved
While, the optimal solution is obtained in [88] with the aid of in [96], by using convex restriction approaches for keeping
rank-two beamformed Alamouti coding and SDR technique, the outage probability of both the secrecy rate and the har-
in addition to the sufficient conditions so this solution will vested energy less than a certain threshold under imperfect CSI
be optimal. Furthermore, when the sufficient conditions are assumptions. It is proved that using the safe approximation
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1341

TABLE IV
B EAMFORMING IN S ECURE EH-E NABLED B ROADCAST C HANNEL

approaches-based beamforming schemes always yield an


obtainable rank-one optimal solution.
Table IV represents a summary of the research concerning
using beamforming in secure EH-enabled broadcast channel.

E. Summary and Insights


In this section, beamforming implementation in EH-enabled
broadcast channel is addressed. The different approaches are
classified according to beamforming design objectives. Subject
Fig. 3. Illustration of Beamforming Approaches in Broadcast Channel.
to different constraints such as QoS at the ID receiver and min-
imum energy harvested at the energy harvester, total transmit
power minimization objective has been considered in a lot the secrecy harvested energy, or minimize total transmit power
of literature. This design problem turns to be non-convex in with guaranteeing that the secrecy rate is above its target
general, and methods like SDR, SOCP, or successive con- threshold. Techniques like successive convex approximations,
vex approximations are adopted to tackle this optimization Bernstein-type inequalities, rank-two beamformed Alamouti
objective. coding, and path-following algorithms are utilized to tackle
Beamforming design for maximizing the energy efficiency the beamforming security design approach.
in broadcast channel is another objective that has been inves- In PS, or TS type receivers, the different design approaches
tigated in the literature. Under different constraints, energy discussed before have been tackled by JBPS or JBTS. Where
efficiency maximization design approach problem, is solved the beamforming weights are jointly designed with the PS
by means of Dinkelbach method, iterative algorithms, or by ratios or the TS factor in order to optimize the system
proposing a power allocation policy. performance. This joint optimization problem is usually tack-
Throughput maximization in EH-enabled broadcast channel led by fixing the PS or TS ratio, and optimizing the beam-
is also studied in the literature. Subject to minimum transmit forming weights. Then by means of one-dimensional search,
power and minimum harvested energy, this non-convex design the optimal value of the PS or TS is found.
problem is tackled via SDR, singular value decompositions, It is worth mentioning that in general it is difficult to com-
Ellipsoid method, or beamforming covariance optimization pare these different beamforming approaches. This difficulty
techniques. arises as these approaches are considered for different design
Exploiting physical layer security in EH-enabled by means objectives, under different constraints and assumptions, with
of beamforming has been examined in a considerable pro- several receiver’ types, in several system models. Figure 3
portion of the literature. Beamforming vector weights are represents the different beamforming design objectives in
optimized in order to maximize the secrecy rate, to maximize broadcast channel.
1342 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

IV. B EAMFORMING IN E NERGY H ARVESTING the high complexity of the proposed scheme, another low-
R ELAY C HANNEL complexity, but suboptimal PS based algorithm that results in
Relaying technique is a suitable strategy to be consid- closed-form solution has been also introduced.
ered in wireless EH technique, as it overcomes attenuation Zhou et al. [112] consider MISO half-duplex PS coopera-
and fading problems, and thus improves information and tive EH-enabled relaying system. The problem of throughput
energy transmission efficiency [97]–[101]. In addition, pow- maximization by optimizing PS ratio, battery operation, and
ering relay nodes with wireless power encourages these nodes beamforming scheme is investigated. In order to solve this
to relay other nodes’ messages as this cooperation won’t problem, a layered optimization method that decomposes the
drain their batteries. Existing literature on EH beamform- joint PS and battery operation schemes into two layers is
ing relay networks involves different types of cooperative proposed, and thus the original problem is reformulated as
relaying policies like Amplify-and-Forward (AF), and Decode- a dynamic programming problem. The proposed method is
and-Forward (DF) [102], [103], working in Full-Duplex (FD) first applied in non-casual CSI case, leading to the theoretical
and Half-Duplex (HD), in addition to revisiting the best relay bounds. Then it is extended to the general causal CSI by using
selection criteria in EH-enabled relay networks [104]. Also the the CSI statistical properties. In order to balance complexity
physical layer security in these networks is being investigated and performance, a greedy method that jointly optimizes PS
from an informational theory point of view [79], [105], [106]. ratios and battery operation schemes in the casual CSI case has
Work on beamforming EH-enabled relay channel is catego- been also presented. High data rate SWIPT AF EH-enabled
rized according to beamforming design objective. relay channels are addressed in [113], by proposing a new
beamforming scheme that can handle many source-destination
nodes simultaneously in addition to the EH receiver. The
proposed scheme provides a noticeable enhancement over the
A. Beamforming Design for Throughput Maximization traditional techniques that deal with one source-destination
Optimizing the beamforming vector in order to maximize only. The optimal corresponding beamformer is obtained by
the system throughput in EH-enabled relay networks is the solving the problem of maximizing information sum rate
most common optimization objective in the literature that has subject to the EH constraints. This problem is non-convex
been tackled in different relaying scenarios. Gong et al. [107] NP-hard problem; thus, it is transferred into a series of con-
consider a system of multiple EH relays helping the com- vex problems by means of first-order Taylor expansion-based
munication between the source and its destination node out linear constraints. The new reformulated problem is solved
in a HD manner under several variations of channel condi- optimally via an iterative-based algorithm.
tions. The overall throughput is maximized by joint design The achievable sum rate maximization in a MIMO SWIPT
of the EH allocation strategy and the beamforming scheme, AF EH-enabled relay system, is solved by exploiting the
so the relay node will be able to optimally harvest energy full diversity gain in orthogonal space time block code-based
and transmit information. In addition, an optimal centralized in [114]. The signals received at relay node are aligned via
scheme is proposed by exploiting the monotonicity of the singular value decomposition method, hence, ratio combin-
system structure. Furthermore, a distributed sub-optimal game ing process takes place at the relay node (which is proved
theory-based scheme that leads to a near optimal solution has optimal in terms of full diversity gain). The optimization
been also discussed. Optimal relay beamforming vector that problem is first approached by alternating optimization and
maximizes the achievable rate in a AF MISO EH-enabled relay line search-based algorithm resulting in local optimal solu-
channel is investigated in [108]. The non-convex problem is tion, then a near optimal algorithm that replaces the main
transferred into a convex form by means of rank-one SDR and problem with its approximation at a high SNR regime is
Charnes-Cooper transformation techniques. In addition, a low proposed. Mohammadi et al. [115] maximize the instanta-
computational complexity sub-optimal closed-form solution neous throughput by means of JBTS in MIMO full-duplex
has also been proposed. DF EH-enabled TS-type relay network. An optimal rank-one
A new distributed energy beamforming and information SDR-based algorithm in addition to many less-complex sub-
transfer technique for maximizing the achievable sum-rate in optimal algorithms have been proposed. In some sub-optimal
SWIPT multiway AF EH-enabled relay channels with TS-type scenarios (like zero forcing, maximum ratio combining, and
receiver is suggested in [109]. In addition, a suboptimal low- maximum ratio transmission beamforming) a closed-form
complexity TS-based technique is also provided. The TS ratio expression of the outage probability is provided to character-
and beamforming vector at the relay node are jointly optimized ize the delay-constrained throughput. Sum-rate enhancement
to maximize the information rate and the harvested energy in in multi-user MIMO DF relay broadcasting channel with EH
AF TS-type relay channel in [110]. Numerical comparisons relays is studied in [116]. The NP-hard problem is simplified
demonstrate that the proposed algorithm outperforms the con- using a block diagonalization-based method in order to can-
ventional intuitive benchmark beamforming algorithm, and the cel the interference signals. This is in addition to proposing
middle source-destination distance is not the optimal position a TDMA-based solution that splits the time for information
for the relay node any more in EH-enabled relay channels. transmission between source to the relays and from the relays
Another distributed energy beamforming scheme is proposed to the destinations.
in [111], in order to maximize the achievable sum-rate in The investigations in [117] focus on SWIPT non-
the two-way EH-enabled relay channel. Furthermore, due to regenerative two-way EH-enabled relay channel. In order
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1343

to solve the non-convex problem of sum-rate maximization B. Beamforming Design for Total Transmit Power
subject to both transmit power constraints at the relay Minimization
and the EH constraints at the EH receiver constraints, a The problem of total transmit power minimization
global optimal iterative SDR-based algorithm is proposed. in SWIPT EH-enabled relay channel, is tackled
The proposed optimal algorithm leads to a global optimal by [122] and [123]. In [122], a PS-type multi-antenna
solution, but with high computational complexity. Thus a EH relay is helping the communication out between the
constrained concave convex procedure-based iterative local source and the destination node. The total transmit power
algorithm that leads to local optimal solution has also been is minimized by joint design of the beamforming matrix (at
provided. In addition a low-complexity sub-optimal general- the source and the relay) and the PS ratios, using a rank-one
ized eigenvectors method-based algorithm has been presented. SDR-based algorithm. While in [123], a multi-antenna AF
Furthermore, iterative alternating optimization-based algo- relay helps a multi-antenna base station to communicate with
rithms have been introduced for multi-antenna sources single antenna PS-type receivers. The problem of total trans-
case. mit power minimization under both SINR and EH constraints,
In [118], a relay channel, consisting of two EH sources com- is first solved via alternating optimization-based algorithm,
municating via multiple relay nodes is studied. The objective that optimizes the transceiver coefficients in an iterative
is to maximize the sum-rate in this network subject to trans- manner. Then the previous algorithm is extended using SDR
mit power and EH constraints, under the assumption that the and S-procedure techniques to the CSI norm bounded errors
sources are able to decode information and harvest energy model. In addition, a similar performance but with low over-
simultaneously. This non-convex problem is solved via local head and computational complexity switched relaying-based
optimal SDR and iterative successive convex approximation- algorithm, with its permutation matrix codebook have been
based algorithm. Due to hardware limitations, the ability to also presented. Furthermore, the design of the permutation
decode information and harvest energy at the same time is matrix codebook is investigated, and two efficient approaches
not practical, and thus an alternative scheme, where the node based on concave-convex procedure and sub-gradient-type
is either decoding information or harvesting energy has been iterative algorithms have been also proposed for both perfect
proposed by extending the iterative algorithm to this more and imperfect CSI scenarios respectively.
practical case. The work in [118] is extended in [119] to the
imperfect CSI scenario, and a robust beamforming scheme
that maximizes the weighted sum rate has been proposed. In
the imperfect CSI case the optimization problem is more dif- C. Beamforming Design for Minimum Mean Square Error
ficult and solved by a two-phase algorithm, first the SINR Criterion Minimization
ratio is fixed, then based on SDR, S-procedure, and succes- The study in [124] focuses on SWIPT MIMO full-duplex
sive convex approximation techniques an iterative algorithm is AF system, consisting of a multi-antenna source communicat-
presented. ing with a multi-antenna source via a multi-antenna AF relay
A new self-energy recycling relaying strategy in the AF node. The problem of total minimum mean square error crite-
SWIPT EH-enabled relay channel is suggested in [120], where rion, subject to total transmit power and the EH constraints, is
the relay nodes harvest the self-interference signal resulting solved by joint beamforming scheme design at both source
from the DF mode to transmit information. The problem of and relay nodes, using alternating optimization-based algo-
maximizing the achievable data rate under the available trans- rithm. Furthermore, a low-complexity channel parallelization-
mitted power constraints at the relay nodes in this network, is based algorithm has been also proposed. Joint source-relay
solved by an SDR-based technique. Furthermore, the same beamforming scheme optimization is readdressed in [125],
optimization problems in TS and PS relaying systems are in order to minimize the total mean square error sub-
solved using SDR and full dimensional search. Comparisons ject to total transmit power constraints and EH constraints
illustrate that the new proposed relaying scheme can achieve at the relay in MIMO PS-type EH-enabled relay channel.
higher data rate compared to the conventional TS and PS This non-convex optimization problem is reformulated using
techniques. Zeng and Zhang [121] revisit the self-energy Lagrangian dual function and successive convex approxi-
relay technique in two phase DF EH-enabled relay system. mations, then an iterative-base algorithm that results in a
Here the source is sending information to the relay node in local optimum solution is proposed. The high complexity
the first time slot, while in the second time slot it sends of the proposed algorithms motivates the author to propose
energy to the relay node while the relay is resending the a sub-optimal water filling-based algorithm using singular
information to the destination node. This new two-phase value decomposition. In order to tackle the same problem a
strategy has significant throughput gains in SWIPT full- new SDR-based technique that uses both iterative algorithm
duplex EH-enabled relay channel. The new proposed scheme and optimization theory is proposed in [126]. A general-
has been proved better than other schemes as it does not ized singular value decomposition-based algorithm is proposed
involve either TS or PS techniques and thus no information in [127], that designs the beamforming vector of the source
transmission interruption. Furthermore, in MISO case (multi- and the relay jointly. In addition to designing an iterative
antenna relay node), the optimal power allocation strategy algorithm based on successive convex approximations for
and beamforming transmission scheme at the relay node are further enhancements on the singular value decomposition
obtained. method.
1344 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

D. Beamforming Design for User Fairness optimal performance in the proposed relay system. While in
In [128] multi-user MISO two way EH-enabled relay system the relaying scheme introduced by Chen et al. [133] both relay
is considered, where a multi-antenna PS-type relay node is and source nodes can harvest energy. The performance of the
securing the communication between multiple single-antenna proposed system is thoroughly investigated for both PS and TS
users. The max-min SINR problem is studied to enhance the techniques. The new scheme demonstrates a considerable gain
user fairness in the proposed relay system. The JBPS problem for fixed location assumption, however, the random locations
is solved by an optimal Dinkelbach-type SDP-based algorithm. case has not been considered.
The proposed algorithm is proved to be fast convergence, A lower and upper bound to the outage probability of a
especially in the case where all antennas at the relay are two-hop MIMO AF EH-enabled relay channel are introduced
employing the same PS ratios scheme. Vahidnia et al. [129] in [134]. In addition, the ergodic capacity, in both transmission
provide the optimal relay beamforming design, that maxi- schemes (delay limited and delay tolerant) is provided. The
mizes the minimum SINR in the system of two multiple outage probability of AF EH-enabled TS-type relay channel
antenna transceivers communicating with each other, via two- is calculated in [135] and provided in closed-form expressions.
way EH relay nodes. Furthermore, the closed-form solutions in The problem of optimal relay placement in multi-hope relay
two different diversity techniques, equal-gain combining and channel with energy transfer is investigated in [136]. The high
maximal-ratio combining are also obtained. non-convex optimization problem is solved by an alpha-branch
and bound algorithm-based model that efficiently determines
the relay location on a Euclidean x-y plane. In the special
E. Beamforming Design for Realizing SWIPT in Relay case of two-hop relay system, the optimization problem is
Channel a pseudo-concave problem that can be solved by distributed
Another critical issue to be considered is how to realize beamforming scheme.
SWIPT in EH-enabled relay channel. Krikidis et al. [130] Table V represents a summary of literature concerning
propose a new low-complexity technique in order to realize exploiting beamforming in EH-enabled relay channel commu-
SWIPT technique in a MIMO DF EH-enabled relay channel, nications.
consisting of a single antenna source communicating with a
single antenna destination via a multi-antenna relay node. The
new generalized selection combiner-based technique, allocates F. Beamforming Design for Implementing Physical Layer
the strongest channels for information decoding process, while Security
the remaining channels are allocated for the EH process (and In this subsection, we present the literature that addressee
vice versa). Furthermore, the number of the strongest channels the physical layer security implementation in EH relay chan-
and the allocation scheme are optimized in order to minimize nel, in order to guarantee the information secrecy in this
the outage probability, which is derived in close-form expres- network.
sions for different scenarios. The proposed technique has been Physical layer security in a two-hop FD WPCN EH-enabled
also extended to the multi-user case, with zero-forcing beam- relay system is addressed in [137] and [138], where a source
former at the relay node, so the external interference signals is communicating with a destination node via a wireless pow-
coming from information signal will be cancelled. The corre- ered multi-antenna relay, with the presence of an eavesdropper.
sponding outage probability closed-form expressions are also The beamforming vector at the relay is optimized in order to
obtained. Another new SWIPT communication system but in maximize the achievable secrecy rate subject to the EH con-
a MIMO AF EH-enabled relay is proposed in [131]. The relay straints at the relay node. Confidential messages technique at
in the proposed system has the ability of storing the remain- EH relay channel under both AF and DF relaying strategies
ing power of the relaying process. The relation between the is investigated in [137]. In the AF case, an algorithm that
rate and the stored energy is thoroughly studied under three optimizes the beamforming scheme is proposed, while the
different receiver’s types, ideal, PS, and TS. Furthermore, the beamforming scheme is jointly optimized with TS ratio by
impact of imperfect on the rate-energy region has been also means of majorization minimization programming and rank
examined. relaxation in the DF case. While Li et al. [138] consider
A three nodes system consisting of a multi-antenna access Alamouti-based rank-two beamforming, with AN deployed
point, a single-antenna relay, and a single-antenna user is at the AF relay. The optimization problem in this scenario
investigated in [132]. In the proposed system, the relay node is solved by means of SDR and majorization minimization
harvests power in the downlink to help the user in forwarding techniques.
his message in the uplink using the harvested power. In order Physical layer security in EH relay channel is guaranteed
to optimize the performance of the system the beamform- in [139] by joint optimization of beamforming vector at the
ing vector is jointly designed with time split factor between source, the PS ratios, and the AF coefficient ratios at the relay
harvesting energy and relaying. Optimal and sub-optimal but nodes. The non-convex joint problem is solved by dividing it
lesser computationally complex solutions in addition to closed- into three sub-problems, finding the PS ratios, beamforming
form expressions are proposed. The authors concluded that vector, and the AF ratios. Then the global problem is solved
exploiting beamforming technique at the access point has sig- efficiently. In order to tackle this objective, Feng et al. [140]
nificant permanence enactments, as is it sufficient to only exploit the duality between the original problem and the
apply the sub-optimal beamforming scheme to achieve the near secrecy rate constrained power minimization problem, in
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1345

TABLE V
B EAMFORMING IN E NERGY H ARVESTING -E NABLED R ELAY C HANNELS

addition to the SDR and the one-dimensional search, to obtain optimum. Secondly a low complexity sub-optimal zero forcing
an always exists rank-one efficient solution. Furthermore, the beamforming-based algorithm. Thirdly a low complexity sub-
suggested robust algorithm performance is compared with AN- optimal SDR and closed-form-based algorithms. Comparisons
based algorithms, and it is shown that it has better worst-case among the proposed algorithms reveal that the constrained
secrecy rate performance. concave convex procedure-based algorithm can achieve the
The JBPS problem for maximizing the secrecy rate in EH best achievable secrecy rate, with lower computational com-
relay channel is tackled in [141], by presenting a local optimal plexity than that of the SDR-based algorithm, but higher than
block-wise penalty function-based algorithm that jointly opti- the closed-form algorithm.
mizes the beamforming vector and the PS ratios for all active Gao et al. [143] consider a multi-user MISO EH-enabled
relays. In [142] three secure beamforming schemes for solv- relay network, consisting of a multi-antenna information and
ing this problem are proposed. Firstly a constrained concave energy source, a multi-antenna self-sustaining relay node,
convex procedure-based iterative algorithm that leads to a local multiple single-antenna users, in addition to the existence
1346 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

TABLE VI
B EAMFORMING IN S ECURE E NERGY H ARVESTING -E NABLED R ELAY C HANNELS

of multiple eavesdroppers. The objective is to minimize the addition a low computational complexity non-iterative algo-
source transmit power under SINR threshold constraint at rithm has been discussed. This is solved in [148] by iterative
both users and relay nodes, EH constraint the relay, and the alternatively optimizing-based algorithm that jointly optimizes
outage constraints of the eavesdroppers. In order to solve the transmit beamforming weights at the AF relay and the
this non-convex problem, it is first transferred into convex covariance matrix of the AN signals.
form by successive convex approximations, then by means Table VI represents a summary of the literature concerning
of first-order Taylor expansion, an optimal iterative algo- exploiting beamforming in secure EH-enabled relay channel
rithm is proposed to solve the reformulated problem. The communications.
same design problem is tackled in [144] using AN-aided
beamforming under imperfect CSI scenario. The problem is
effectively solved by S-procedure and SDR techniques for both G. Summary and Insights
PS and TS receiver’s type. Furthermore, it is proved that the Literature that investigates exploiting beamforming in EH-
robust secure beamforming design in PS scheme has a bet- enabled relay network are discussed in this section. In relay
ter performance than that of TS scheme in the corresponding network architectures beamforming can be implemented at
channel architecture. the source or relay node, or at both nodes to achieve the
A novel secrecy technique is proposed in [145] and [146], desired objective. In addition, in EH-enabled relay networks
called ‘Harvest and Jam’, to provide secure communications it might be the relay or the source node or both nodes
in EH-enabled relay networks. In this two-phase strategy relay that are capable of performing the EH process. Again the
nodes are used as jammers for impairing the eavesdroppers different beamforming approaches in EH relay channel com-
channel. In the first phase, the relay nodes split the power munications are organized according to beamforming design
received from the sources into two power streams, one used objectives under the different relaying mode AF or DF in FD
to generate AN signals for jamming the eavesdroppers, and the or HD manner. Beamforming for maximization the through-
other energy stream is used to transmit the information in the put in relay networks is the approach most addressed in
second phase. In order to maximize the achievable secrecy rate the literature. This optimization problem has been tackled
under individual power constraints of the relay nodes under by different methods such as exploiting the monotonicity of
both global and local CSI scenarios, Xing et al. [145] propose the relay system, distributed beamforming algorithms, layered
an SDR-based algorithm and a fully distributed robust algo- optimization method, SDR and Charnes-Cooper transforma-
rithm for global and local CSI scenarios respectively. In [146] tion. Alternating optimization technique is adopted in order to
an SDR-based algorithm that jointly optimizes the AN covari- minimize total transmit power in EH-enabled relay network.
ance matrix and the beamforming vector is proposed in the Minimizing the total minimum square error objective and user
perfect CSI case, and the technique of worst-case is employed fairness are other beamforming design objectives that have
to maximize the minimum achievable secrecy rate in the been approached by singular value decomposition-based or
imperfect CSI case. This same problem is solved in [147], by SDR-based techniques respectively.
jointly designing the source signal, the AN, and the relay for- Several protocols have been suggested for realizing SWIPT
warding beamforming scheme. This non-convex optimization technique in beamforming EH-enabled relay networks. Some
problem is first solved by alternatively optimizing beamform- protocols allocate the strongest channels for ID process while
ing matrices at the source and the relay separately. Then a joint the rest are allocated for EH process in MIMO DF EH-enabled
optimization-based iterative algorithm has been introduced. In relay network. Other proposed protocols that depend on the
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1347

mobility, to decide when to free the currently used spectrum


band to be returned to primary system users.
Secondary users can operate in three main paradigms
depending on white space access: interweave, underlay, and
overlay [153]. In interweave paradigm, secondary users are
allowed to use the licences spectrum when it is free by means
of spectrum sensing and mobility techniques [154], [155].
Fig. 4. Illustration of Beamforming Approaches in Relay Channel. Spectrum utilization efficiency in interweave paradigm is
increased by secondary users’ opportunistic white space reuse.
Interweave paradigm requires secondary users full knowledge
relay ability of storing the remaining power to be used in the of the primary system activity in order to periodically sense
relaying process. In addition to introducing a protocol that the radio spectrum, detect white space and then opportunisti-
allows the relay nodes to harvest energy in the downlink to cally communicates with minimal interference to the primary
help users in the uplink. active users.
In addition, different beamforming approaches have been In overlay paradigm, secondary users can transmit with pri-
considered in order to guarantee the information security in mary users on the same licensed frequency bands as long as
EH-enabled relay networks. Beamforming is exploited at the this concurrent transmission does cause interference with the
relay node to generate a jamming signal to impair eaves- primary system. Interference avoidance in overlay paradigm
droppers channel, or to realize Harvest and Jam technique, is carried out by detecting the presence of primary users and
where the relay nodes used the harvested power to jam modifying the secondary users transmitted signal accordingly
the malicious nodes. Methods like majorization-minimization, to avoid interference with primary users. Overlay communica-
S-Procedure, problem dividing or block-wise penalty function- tion can be classified into two main approaches, cooperative
based algorithm are adopted in order to solve the non-convex and non-cooperative [153]. In the cooperative approach [156],
beamforming optimization design of maximizing the secrecy a sort of inter-system collaboration scheme is established
rate in the relay network. Furthermore, the problem of optimal to avoid secondary users interference to primary receivers.
relay location in EH-enabled relay networks has been read- One way to avoid interference in cooperative approach is by
dressed, where the criteria which are adopted in conventional exploiting the code domain techniques such as spread spec-
relay networks are proved to be not valid. The optimal relay trum techniques (frequency hopping or direct sequence), where
allocation in Euclidean plane is obtained by means of Alpha- primary and secondary users use nearly orthogonal codes to
branch and bound algorithms. The different beamforming transmit simultaneously. In non-cooperative approach, sec-
design approaches in relay networks are illustrated in Figure 4. ondary users alone determine the availability of white space
that doesn’t cause any interference when accessed. Overlay
paradigm requires secondary users’ full knowledge of the
V. B EAMFORMING IN E NERGY H ARVESTING primary users’ codebooks and messages [155].
C OGNITIVE R ADIO N ETWORKS Underlay paradigm allows simultaneous primary and sec-
The excessive demand on the spectrum resource caused ondary transmissions as long as the interference level at
by the ever increasing number of wireless communication the primary system doesn’t exceed a predefined tolera-
systems and functionalities has become extremely high, mak- ble interference threshold called the “interference tempera-
ing the available bandwidth spectrum very limited. Out of this ture” [157]. Underlay paradigm has drawn a lot of attention,
problem of spectrum scarcity, the idea of Cognitive Radio due to this simultaneous transmission property that yields
Network (CRN) is born. The principle of the CRN is to let higher spectrum efficiency. Underlay paradigm doesn’t involve
secondary users (also referred to as cognitive users) to coex- any spectrum sensing process [158] or opportunistic exploita-
ists in the licensed bands allocated to the primary users (also tion of white space; instead, it only requires knowledge of
known as non-cognitive or licensed users) as long as this the interference caused by the secondary transmission to the
coexistence doesn’t degrade the performance of the primary primary receiver. Underlay is the most suitable paradigm for
system [149], [150]. In other words, there exists a temporary low-power short range communications such as spread spec-
spectrum void, referred to as white space, defined as “band trum, ultra-wide band communication, cognitive femtocell,
of frequencies that are not being used by the primary user machine-to-machine communications [159].
of that band at a particular time in a particular geographic Interference provoked by secondary system at primary
area” [151]. This white space can be utilized by secondary system is controlled by two major approaches [160]. In the
users opportunistically. In order to implement the concept of systems where not all primary and secondary users can be
CR, secondary users should be capable of performing four supported with their SINR threshold (infeasible systems), a
basic functions [152]: Spectrum sensing, to detect the avail- joint power and admission control algorithm is used to protect
able white spaces to be utilized for their own transmissions. all primary users along with maximum number of secondary
Spectrum decision, at this stage secondary users decide which users. In the schemes where all secondary users can reach their
white space to be used depending on the results of spectrum target SINR (feasible systems), it is generally desired to assign
sensing process. Spectrum sharing, to decide how to share the the secondary users with the highest possible level of SINR
available white spaces with other secondary users. Spectrum to maximize the secondary system’s throughput subject to the
1348 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

interference level constraints of the primary system. Normally and space are needed, to this end spectrum occupancy mea-
these two interference management techniques start by charac- surement is introduced. Spectrum occupancy measurement is
terising the feasible cognitive interference region, that defines an approach to describe the utilization rate of the channel or
the allowed levels of interference provoked by the secondary band by measuring the portion times where the spectrum is
system at any primary receiver while guaranteeing that all occupied, to provide the efficiency of the spectrum alloca-
primary users are protected. tion policies. Höyhtyä et al. [164] focus on the importance
Another important key feature of CRNs is primary system of the spatial dimension in spectrum occupancy measure-
activity modelling, as the secondary users knowledge of spec- ments, by acquisition data at separate locations in different
trum utilization will greatly enhance the system performance areas, by exploiting interference maps tool to analyse spectrum
of CRNs [161]. Activity modelling helps secondary users to occupancy over a geographic area of interest.
predict the status of a spectrum band depending on the past From the above discussion, it’s obvious that CRNs involve a
activity of the primary users, and hence they can choose considerable amount of signal processing, hence extra power is
the best available white space for their future transmission. needed for both signal processing and transmission processes.
Spectrum activity modelling approaches can be categorized Therefore, EH-enabled CRN presents a promising solution for
as follows [162]: Probability density function and cumula- increasing both energy and frequency efficiency. As in addi-
tive density function: this model characterizes the dynamic tion to overcoming the frequency scarceness problem [165],
range of the primary system signal values, and the how often it prolongs the network lifetime and encourages inter-system
these values occur. Secondary users’ knowledge of the statis- cooperation [166], [167]. Exploiting the SWIPT technique to
tical information of the primary systems signals helps them to transmit information to the primary receivers and energy to the
choose the transmission time and the allowed transmit power secondary nodes creates more cooperation incentives between
level. Markov Chain: as the spectrum occupancy changes primary and secondary systems [168] in power-constrained
between utilized and not-utilized status only, Markov chain CRNs, where the cooperation between primary and secondary
is a natural choice to characterize the transition probability system is threatened by the energy shortage in the system.
among possible states. Different variations of this model can In addition, a lot of research demonstrates that powering
be found such as continues-time Markov chain which is the the secondary system provides higher data rate than that of
most common model, continuous-time semi-Markov chain, the conventional CRNs. EH-enabled CRN considerable atten-
and two-state and three-state Markov chain. Linear regres- tion of late and substantial amounts of research has been
sion model: Markov chain models only characterises the white done in this domain, either by suggesting information and
space in time dimension, and doesn’t cover frequency and geo- energy cooperation schemes, cooperative relay selection, or
graphical location dimensions of the white space. Hence, linear the physical layer security issue.
regression model is used to provide a full representation of
the white space occupancy in the three dimensions. Bayesian
model: it is normally used to describe joint conditional prob-
ability distribution among different related random variables. A. Beamforming Design for Throughput Maximization
In CRNs, Bayesian model is used to describe the correlation Zhu et al. [169] address rate maximization problem in a
in time, space, and frequency of the white space occupancy SWIPT EH-enabled CRN, where the secondary user shares
activity. the same transmit spectrum with the ID and EH primary users.
All the models presented above describe the statistical char- In order to solve this problem, a rank-one SDR technique is
acteristic of the spectrum occupancy in some applications the proposed. Furthermore, to reduce the computational complex-
actual values of spectrum usage are needed. Furthermore, the ity, two sub-optimal algorithms that yield an upper and lower
four basic cognitive radio functions (spectrum sensing, spec- rate bounds have been also presented. Cooperation between
trum decision, spectrum sharing, spectrum mobility) introduce primary and secondary system is motivated in [170] by adopt-
time delays that reduce the spectrum utilization efficiency. ing a dedicated energy beamforming in an overlay EH-enabled
Thus, spectrum occupancy prediction techniques are very CRN system. This cooperation carried out via energy access
important for SRNs designs, in order to reduce the delay points in two transmission phases. In the first phase, the
of finding the white space, minimize the energy consump- full-duplex energy access points charge the secondary trans-
tion, reduce the interference with primary system signals mitter while receiving the primary transmitter’s signal at
and increasing the system throughput [163]. Several spectrum the same time. In the second phase, forward the primary
occupancy prediction techniques have been introduced in the transmitter’s message. The optimal receiving beamforming
literature, that can be classified according to their application. scheme that maximizes the weighted sum-rate maximization
For example, hidden Markov model-based technique is used in the proposed system, under the transmit power and the
for spectrum decision and spectrum mobility functions, while total cost constraints, is tackled by means of successive con-
multilayer prediction neural network-based technique is used vex approximation techniques. Furthermore, a sub-optimal
for spectrum sensing process, Bayesian inference and moving zero-forcing-based scheme has been also introduced.
average techniques are adopted to predict spectrum sensing, Rate maximization in EH-enabled overlay CRN with non-
spectrum decision and spectrum mobility functions. However, regenerative two-way relay is addressed in [171] and [172].
in order to accurately describe the utilization rate of the spec- In [171], the system is consisting of a multi-antenna sec-
trum, meaningful data about spectrum use in time, frequency ondary user, that harvests energy from the primary users’
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1349

signals using the PS technique. Subject to minimum achiev- communicate with a secondary base station. The idea of the
able rate requirements of the primary users, and maximum suggested two-hops framework is to power the secondary users
total transmit power at the secondary user transmitter con- with the excess energy, so it can do the dynamic spectrum
straints, and under both perfect and imperfect CSI at the access. In the first phase, the primary transmitter communi-
secondary user transmitter assumptions, the achievable rate cates with the relay while powering the secondary users, in
for the secondary users is maximized for both CSI scenar- the second phase, the secondary users utilize the energy har-
ios. In the perfect CSI case, the JBPS optimization problem vested in the previous phase to do dynamic spectrum access
is solved optimally by Charnes-Cooper transformation and to communicate with its base station, while the relay is com-
one-dimensional search, and sub-optimally by zero forcing municating with the primary receiver. In addition, the outage
beamforming-based scheme. In the imperfect CSI case, sub- probability of the secondary system is derived and provided
optimal zero forcing scheme and SDR-based algorithm are in closed-form expressions.
presented, and the corresponding rank-one solution is pro- In [176], three joint information and energy cooperation-
vided by Gaussian randomization method. A Bisection Search based protocols are proposed, employing ideal backhaul,
method-based algorithm is proposed in [172], to solve the practical PS and TS schemes. In addition, for each protocol,
problem of rate maximization under the constraint that the the optimal and the sub-optimal zero forcing-based solutions
interference in the secondary relaying network is less than a are derived. Comparisons show that PS scheme provides larger
certain threshold. The proposed sub-optimal algorithm jointly rate-energy region than that of the TS scheme specially in
optimizes the relay selection and the power allocation scheme. high efficiency power transfer regime. Cooperative self-energy
recycling CRN is proposed in [177], where the secondary user
harvests energy from primary signals, and in return it relays
B. Beamforming Design for Realizing SWIPT in CRN the primary signals in AF mode. The objective is to provide
In [173], several cooperative SWIPT schemes in the EH- the optimized beamforming scheme which guarantees that the
enabled overlay CRN are proposed. The principle of the secondary user has the sufficient energy level to help the pri-
suggested energy cooperative schemes is to let all the sec- mary system communication. In order to reach the optimal
ondary nodes communicate in the first time slot, whilst the solution, an SDR-based algorithm is presented. Furthermore,
primary transmitter is harvesting energy from these secondary the same optimization problem is solved for both TS and PS
signals. Then in the next time slot, the primary transmit- protocol using SDR and one-dimensional search, in order to
ter sends its own data to the primary and secondary nodes, illustrate the improvement on the achievable rate provided by
and in the third time slot, the secondary nodes amplify the self-energy technique over TS and PS schemes.
and forward the received primary data. Depending on sec-
ondary access and powering primary transmitter strategies, C. Beamforming Design for Energy Efficiency Maximization
five different cooperative schemes are provided, two of them
The focus in [178] is to design a robust transceiver for a
used distributed energy beamforming schemes. Furthermore,
SWIPT MIMO EH-enabled CRN under the worst-case CSI
for all schemes, the primary and secondary rates are cal-
assumption. The new design maximizes the sum harvested
culated and provided in closed-form formulas. In addition
energy at the EH receivers subject to minimum mean square
to the cooperative schemes, a relay-selection strategy that
error at the secondary ID receiver and the SINR at the pri-
minimizes the minimum mean square error of the received
mary receiver’s constraints. The main problem is alternatively
primary data at the primary receiver has also been presented.
optimized by optimizing the transmit covariance matrix at the
Comparisons among the different schemes demonstrate that
secondary user transmitter, and the pre-processing matrix at
the distributed beamforming-based schemes provide a higher
the secondary user ID receivers.
achievable primary rate.
Table VII represents a summary of the literature regarding
The suggested energy cooperative scheme proposed in [174]
exploiting beamforming in energy harvesting cognitive radio
is for orthogonal EH-enabled CRN, where in each frequency
networks.
channel there is a limited energy EH primary user and a sec-
ondary user. In the proposed scheme, the only secondary user
found in the band is aided by the nearby secondary coopera- D. Beamforming Design for Implementing Physical
tive beamforming relays with certain probability, where these Layer Security
relays keep the received secondary data packets in a relaying EH-enabled CRN and CRN system in general entails new
queue when the direct link is not available, to transmit when security challenges due to its characteristics and nature of
the secondary user direct link is available. Closed-form expres- openness. Most of these challenges arise from the coexistence
sions for the outage probabilities and the secondary throughput of both primary and secondary users in the network and the
in this new protocol are derived. In addition, the impact of mechanisms deployed by CR systems. For instance, the sec-
the proposed system parameters like the energy and relay- ondary users should be able to recognize primary users from
ing queue capacity, and number of relays are investigated. malicious ones. Furthermore, the sensing mechanism should
Another two-hops strategy to realize the SWIPT scheme is be fully secured so no malicious nodes can attack and threaten
suggested in [175], where a multi-antenna primary transmit- the correctness of the sensing information. However, not much
ter communicates with its primary receiver via a DF relay, work has been conducted to derive fundamental and theoretical
while in secondary system there are multiple EH transmitters secrecy limits for CRN [179].
1350 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

TABLE VII
B EAMFORMING IN E NERGY H ARVESTING -E NABLED C OGNITIVE R ADIO N ETWORKS

Physical layer security in EH-enabled CRN PS-type resource allocation scheme aims to fulfil three objectives:
receivers is visited in [180], where the primary system is Maximizing the EH efficiency, maximizing total transmit
consisting of a transmitter and a primary receiver, while in power and minimizing interference to transmit power ratio,
the secondary system, there are a secondary transmitter, a subject to secure communication in secondary system con-
trusted secondary receiver, and multiple secondary users that straints, CSI imperfection of the idle secondary receivers and
can be considered as potential eavesdroppers. The objective of the primary receiver channels at the secondary transmit-
is to maximize the secondary system rate under both secrecy ter. This multi-objective, non-convex optimization problem,
QoS requirements of the primary system and secondary trans- can be solved optimally using SDR techniques. Furthermore,
mitter power constraints. This non-convex joint PS ratio and for the case when the solution is not available, two sub-
secure beamforming optimization problem is solved by two optimal algorithms have been proposed. However, the objec-
stage techniques. Firstly, fixing the PS ratio, and finding the tive of maximizing the interference power provoked by the
optimal secure beamforming scheme. Secondly, performing secondary system to the primary system is not reason-
one-dimensional search over all possible PS ratios. able, as this interference signals may exceed the tolerable
Wu and Chen [181] revisit the security problem in EH- threshold, causing performance degradation in the primary
enabled CRN for a different system model. The proposed system.
system consists of a multi-antenna secondary transmitter com- The problem of EH fairness in a secure MISO CRN
municating with a secondary receiver, and helping a primary system with help of AN-aided Beamforming is examined
transmitter to communicate with a primary receiver. Both in [183]. Under imperfect CSI model (bounded and proba-
secondary and primary receivers are assumed to decode infor- bilistic), the problem of Max-Min harvested energy is solved
mation and harvest energy using the PS technique. By taking by jointly designing the AN beamforming vector and the PS
into consideration the existence of potential eavesdroppers and values. The non-convex problem is solved optimally by an
imperfect CSI at the secondary transmitter, the objective is to s-procedure-based algorithm in the bounded CSI error case,
minimize the transmit power subject to minimum information while a Bernstein type inequalities-based sub-optimal solution
rate and harvested energy constraints at both receivers. This is proposed for the probabilistic case. In addition, a trade-off
is in addition to guaranteeing minimum rate at the poten- is established between the secrecy rate of the secondary user
tial eavesdropper. The non-convex joint optimization of the and the harvested energy of the EH receivers under a max-min
PS ratio and the secure beamforming scheme problem is fairness criterion.
first reformulated using the S-Procedure technique, then the Lei et al. [184] investigate the secrecy outage probabil-
SDR technique is used to resolve the new relaxed problem ity of underlay MIMO CRN with EH and AS receiver type.
to obtain the optimal joint robust beamforming and PS Depending on the SCI availability, two antenna selection
scheme. algorithms are proposed. The antenna selection-based scheme
A resource allocation scheme in secure EH-enabled CRN shows a considerable gain when comparing to the conventional
is presented in [182], where the primary system is harvest- time or space transmission schemes.
ing energy in the idle mode, while the SWIPT secondary Table VIII represents a summary of the literature regarding
system provides secure communication to all secondary exploiting beamforming in secure energy harvesting cognitive
users. The proposed weighted Tchebycheff approach-based radio networks.
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1351

TABLE VIII
B EAMFORMING IN S ECURE E NERGY H ARVESTING -E NABLED C OGNITIVE R ADIO N ETWORKS

E. Summary and Insights


In this section, we survey the state-of-art on beamforming in
EH-enabled CRNs. In cognitive radio networks, beamforming
can be deployed at the primary or the secondary transmitter
in order to realize different objectives under different spec-
trum sharing paradigms in cooperative and non-cooperative
scenarios (see Figure 5).
Beamforming design for throughput maximization in EH-
CRNs under different system models and constraints, has
drawn a lot of attention in the literature. This NP-hard Fig. 5. Illustration of Beamforming Approaches in Cognitive Radio
optimization problem has been tackled by optimal and sup- Networks.
optimal algorithms. Techniques like successive convex approx-
imation, Charnes-Cooper transformation, and Bisection search
method are widely used to solve this problem. Realizing and training strategies yielding an overhead. The trade-off
SWIPT in beamforming EH-enabled CRNs has been inves- between energy transmitting and training process should be
tigated in the literature by proposing new protocols. The made carefully, so that the CSI is acquired with sufficient
proposed schemes exploit the EH capability to facilitate accuracy without consuming too much energy. In this section,
primary and secondary systems cooperation. Where the sec- literature addressing acquisition, training schemes, feedback
ondary users harvest energy from primary users signals, in mechanisms and overhead-energy trade-off is presented.
return the secondary users used the harvested energy or part Balancing the time resource allocation between channel
of it to relay the primary users signals. While SDR-based algo- estimation and WPT processes in order to maximize the har-
rithm is adopted in order to maximize the energy efficiency in vested energy in MISO EH-enabled communication system,
EH-enabled CRNs. is addressed in [185]. The allocation of energy resource used
Implementing physical layer security in EH-enabled CRNs for wireless energy transfer has been also investigated, by
is another important beamforming approach that has been performing dynamic optimization of the preamble feedback
examined in the literature. Where beamforming weights are information length. The optimal offline (fixed) preamble length
optimized by means of 1-dimensional search, S-procedure, that reduces the feedback complexity has also been obtained.
Bernstein-type inequalities and SDR methods in order to Moreover, the optimal power allocation schemes for the WPT
maximize the secrecy rate, minimizing total transmit power in dynamic-length preamble and fixed-length preamble cases
and enhancing the secrecy energy efficiency respectively. are derived. The proposed transmit power allocation scheme
Beamforming EH implementation in CRNs is of huge assigns power depending on the estimated channel status and
significance regarding system performance. However, exist- the optimal preamble information length. In [186], the optimal
ing literature doesn’t cover this topic sufficiently, and more training strategy is derived by solving the problem of maxi-
research is needed especially on exploiting multi-antenna EH mizing the net average energy harvested at the EH receiver,
in the underlay paradigm, FD cooperative CRN schemes and after taking into consideration the energy consumed for chan-
simultaneous spectrum sensing and data transmitting at the nel training. The forward link channel is efficiently estimated
secondary system. at the energy transmitter via a dedicated feedback channel
from the EH receiver to the energy transmitter. A CSI train-
ing technique for maximizing the transferred energy in MISO
VI. C HANNEL S TATE I NFORMATION F EEDBACK frequency-selective EH-enabled system is proposed in [187].
M ECHANISMS , OVERHEAD , AND T RAINING T ECHNIQUES The proposed training technique can be divided into two
CSI perfect knowledge at the transmitter necessitates huge stages. In the first stage pilot signals are sent over certain avail-
amount of feedback data from all receivers to the transmitter. able frequency sub-bands from the EH receiver to the energy
Of course, this process involves CSI acquisition mechanisms transmitter. Based on the received power of the pilot signals,
1352 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

the transmitter can decide the strongest frequency sub-bands, VII. R ESEARCH C HALLENGES AND
and send their indices back to the EH receiver. In the second F UTURE A PPLICATIONS
stage, the strongest sub-bands are well estimated via further The new emerging beamforming EH technique has
training by the energy harvester to design the beamforming presented itself as a promising solution for many power-
scheme at the transmitter. Moreover, an asymptotic analysis constrained applications, but yet it creates many challenges
is also derived for large number of antennas or sub-bands that need to be addressed in future research to make EH com-
systems. These results can be extended to several areas like munication systems a universal reality. The challenges and
MIMO channel, correlated channels case, or multi-user cases. limitations in EH wireless systems are different from those in
Gangula et al. [188] maximize the throughput rate via opti- the conventional wireless networks [192]. The randomness and
mizing the feedback strategy. It is first assumed that only irregularity of the harvested energy demands the proposition of
the receiver has the EH capability; under this assumption, new power and information resource allocation strategies, SCI
the feedback policy is optimized. Later in the case, where acquisition policies and robust distributed beamforming algo-
both transmitter and receiver have the same EH profile, it is rithms. In addition, considerable research needs to be carried
found that the feedback optimization problem can be simpli- out regarding finding statistical models, fundamental limits,
fied, and the feedback bit allocation and transmission schemes and rate-energy trade-off on EH wireless networks. However,
that maximize the upper bound on the ergodic rate are obtained beamforming EH technique has the potential to boost the
in this case. In the general case, where the EH profiles are performance of many applications and functionalities such
different, no simple algorithm can be found, but under the as non-orthogonal multiple access, unmanned aircraft vehi-
assumption that one EH profile is fixed, optimizing over the cles and Internet-of-Things. These potential directions leave
other EH profile has either directional or staircase water-filling the door open for further research on future beamforming EH
interpretation nature. Different CSI cooperation scenarios are applications.
proposed in [189], namely partial cooperation and signal sep-
aration. In partial cooperation, the CSI is known to both
transmitters and receivers, but with no knowledge of the sig- A. Fundamental Limits of Wireless Energy
nal waveform. The second scenario is signal separation, where Harvesting Channel
the transmitters and receivers have full knowledge of the CSI A lot of literature has addressed the classical information-
and the signal waveforms. For both scenarios, the problem of theoretic problems of capacity and achievable regions in
maximizing the information rate subject to the EH constraints conventional wireless networks. Most of the theoretical limits
(which is a non-convex problem) is solved by introducing new and bounds have been determined, and only some problems
parameters for decoupling the problem into two sequential are still open. But the question is: what are the funda-
problems. The solution of this problem defines the achiev- mental limits of these wireless networks if EH is enabled?
able rate-energy region’s Pareto boundary, which represents This question is of both theoretical and practical importance,
the optimal rate-energy trade-off. as determining the capacity and rate regions achieved in a
Reducing the feedback overhead in two-user MIMO SWIPT communication network provides a trusted answer of the com-
channel is the goal of [190] and [191]. In order to lessen munication feasibility. The literature is scarce with regards
the feedback overhead in [190], a geodesic energy beam- to investigate the fundamental limits of information trans-
forming scheme with partial CSI at the energy transmit- mission in EH-enabled wireless communication systems. The
ter, which is proven optimal, is presented. It is shown achievable AWGN capacity of a stochastic energy harvest-
that under some constraints on transmit signal covariance ing system, with its energy managing policy are analysed
matrix, the feedback overhead of the information transmit- and proposed in [193]. Harvested energy managing scheme
ter can be reduced by proposing a geodesic information and the theoretical capacity of master and child node connec-
beamforming scheme. In addition, under the same con- tion are analysed in [194]. Fundamental limits in EH-enabled
straints the geodesic information/energy beamforming strategy machine-to-machine communication are evaluated and exam-
is proved optimal. Furthermore, the proposed Geodesic infor- ined in [195]. Here the impacts of the system parameters like
mation/energy beamforming strategy can be extended to the implementation losses and the amount of available energy
k-user case, with multiple information and energy transmitters on the theoretical bounds are evaluated. However, the theo-
case by singular value decomposition based approximations. retical capacity of EH-enabled communication system is not
While in [191] the feedback design is simplified, by consider- yet determined from an information theory point of view. All
ing one-bit feedback channel, where this one-bit refers to the previous channel law derivations are done under endless bat-
status of the harvested energy of each receiver either decreas- tery assumptions in Gaussian channels. Thus, deriving the
ing or increasing. According to the feedback CSI information, fundamental limits of EH communications in different scenar-
the transmitter changes its beamforming weights and acquires ios such as finite battery capacity, side information availability
better CSI at the same time by using a training scheme called and capacity-achieving coding and power managing schemes
analytic centre cutting plane method. The performance analy- are still open and further research is needed [196].
sis of the proposed learning strategy reveals that it can estimate
all the channel simultaneously without any effect on the con- B. Channel State Information
vergence speed. Moreover, it has superior performance over all Exploiting multi-antenna techniques in wireless EH systems
similar strategies specially in the large number of users’ case. is highly dependent on the availability of CSI at the transmitter
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1353

side. However, in EH communication system the process of layer in multi-antenna EH-enabled communication systems
CSI attainment from both ID and EH receivers is not triv- is more challenging than conventional wireless systems for
ial. As EH receiver might be a simple EH circuit without any three reasons. Firstly, the EH nodes in the network might try
baseband implementation to convey its own channel condi- to decode the information message sent to the intended and
tions. Hence, research regarding finding solutions in order to hence, potential eavesdroppers. Secondly, malicious nodes are
acquire simple EH receivers CSI signals are needed. On the usually not active, and acquiring its CSI in order to build the
other hand, in EH communication networks, ID receiver suf- beamforming vector is quite difficult. Thirdly, in EH networks,
fers from both information and energy interference signals that transmitters try to increase the transmit power to facilitate
impair its CSI signals. In addition, accurate channel state is energy transfer, that leads to higher potential leakage due to
difficult to acquire or predict, as it is time-varying because of the open nature of wireless networks [18]. Therefore, many
the dynamic activity nature of the wireless network, and chan- research directions toward guaranteeing secure communication
nel estimation process can’t be carried out frequently due to in beamforming EH communication systems are still open to
energy limitations. This imperfect knowledge of CSI should be explored.
be taken into consideration as it causes high degradation on
full system performance, especially in multi-antenna system E. Unmanned Aircraft Vehicles
case. Thus, robust beamforming designs that can handle this
Unmanned Aircraft Vehicles (UAV) surveillance and recon-
imperfect SCI problem, new CSI acquisition technique and
naissance systems are widely used in many military and
feedback mechanisms in EH-enabled wireless communication
civilian applications. Over the last decades, the evolution of
systems are needed.
UAVs is shifting towards increasingly smaller aircraft vehicles.
The endurance or maximum flight time of small size UAVs is
C. Resources Allocation Policies problematic as they need more fuel or batteries to perform
Energy resources allocation policies in EH-enabled com- longer journeys, but yet they are small and lightweight so
munication systems are with higher complexity than that can’t afford large payloads. EH is an attractive solution for
of conventional systems, as their objective is changed from increasing the UAVs endurance by proving a source of energy
minimizing the power consumption in the conventional without adding any significant payloads or increasing the size
systems to a continuous optimizing process over time in EH of the fuel or battery system. Furthermore, Providing the UAVs
networks [197]. Furthermore, as the devices in EH com- by wireless power can’t be carried out by simple omnidirec-
munications systems are power constrained, these policies tional energy broadcasting, the location of the UAV should be
should be of low computation and signalling complexity, with- acquired, and the power signal must be directed toward it in
out causing any feedback overhead. They are attempts to order to guarantee the deliverance of high power levels to the
develop some resource allocation algorithms for this challeng- UAV. To this end techniques like beamforming and directional
ing type of networks, such as spend-what-you-get, constant communications are needed. UAV is a potential application
rate, energy storage threshold-based and energy storage-linear for beamforming EH techniques that leaves the door open for
policies [198]. However, most of the proposed algorithms are more research to be carried out in this area.
designed for single node or single link scenario, and can’t be
extended easily from these stochastic models to larger sce- F. Non-Orthogonal Multiple Access
narios. Hence, proposing low-computation resource allocation
Non-Orthogonal Multiple Access (NOMA) has been recog-
algorithms that optimize the power consumption over time
nised as the potential multiple access schemes for fifth gener-
in more sophisticated multi-nodes and connections network
ation (5G) communication system, for its attractive features
models is still open.
of spectral efficiency, low latency and massive connectiv-
ity [199], [200]. The key idea of NOMA is to use power
D. Physical Layer Security domain for multiple access, whereas the previous communica-
Beamforming EH techniques can be exploited in physical tion generations have been exploiting the time/frequency/code
layer security by transmitting the energy signals waveform as domain, yielding better system performance. In few litera-
a Gaussian pseudo-random sequence, that is known a priori ture, NOMA is combined with multiple-antenna techniques
to all legitimate receivers. This technique can provide secure like beamforming (NOMA-BF) to exploit both power and spa-
communication, as this sequence can serve as interference to tial domains to enhance SINR [201], or to increase the spatial
illegitimate eavesdroppers, while can be cancelled easily at multiplexing gain [199], [202] by serving more than one user
legitimate users terminals by means of beamforming technol- per each beamforming vector. On the other hand, NOMA and
ogy [92]. In addition, secure communication can be guaranteed energy harvesting techniques are merged in some of the lit-
by letting nearby nodes transmit jamming or AN signals to erature to enhance both energy and spectral efficiency and
impair the potential eavesdroppers channels. The processes of overcome energy and spectrum scarcity in the system [203].
generating and broadcasting these signals drain the terminal Sun et al. [204] and Liu et al. [205] introduce SWIPT to
battery, which can be compensated by wireless charging. NOMA system to encourage strong users to relay weak users’
Furthermore, in order to locate and deliver a sufficient energy information messages as this collaboration won’t drain their
amount to the jammers nodes, multi-antenna techniques like batteries. While WPCN scheme is merged with NOMA uplink
beamforming are important. However, exploiting physical system to increase the individual data rate in [206]. Merging
1354 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

NOMA, beamforming EH techniques is a future work that paradigm the primary users’ activity is sensed. If idle, the sec-
should be taken into consideration, as these techniques are ondary users work in overlay mode and transmit at maximum
essential elements of the coming communication systems, and power for higher throughput. Otherwise, they select underlay
combining them yields better spectral and energy efficiency mode and transmit with lower power. This hybrid approach
with higher sum rates, and stimulates the users to cooperate improves the throughput of the secondary system, without
in more effective way [207]. causing any QoS degradation of the primary system. Again
multi-antenna EH technique represents a promising solution
for this hybrid scheme, as directional transmissions and wire-
G. Internet of Things less charging features provided by beamforming EH yield
It is highly believed the Internet of Things (IoT) will be lower interference in addition to guaranteeing the deliverance
one of the anticipated services provided by the 5G com- of adequate energy levels to compensate the power drains
munication system. IoT is a sophisticated architecture that by the additional signal processing and paradigms switching
allows enormous number of devices to be connected and com- involved by this hybrid paradigm.
municate with each other. As these devices are intended to Another promising area for future research is energy har-
communicate without any direct human interference, features vesting beamforming in FD-CRNs. In conventional CRN or
like self-sustainable and long lifetime communications are HD-CRNs, secondary users can either sense the white space
crucial requirements in the implementation of such system. or transmit. Thus, the secondary user spends a significant
Thus, adopting devices that are capable of harvesting energy portion of the given time in sensing the spectrum to ensure
from external resources and adopting EH transmission tech- that the sensing process is accurate as imperfect sensing
niques are highly advised. Wireless EH-enabled IoT system results in data loss and undesired interference to the primary
has been proven to be a promising scheme for improving the system. As a result, the sensing stage might take a consid-
lifetime and the sustainability of the IoT systems, in addi- erable time when compared to that of the transmit process
tion to its appealing features of simplicity in implementation yielding lower secondary system’s throughput. Furthermore,
and availability [208]. Furthermore, introducing multi-antenna spectrum sensing and transmission in HD manner requires
techniques like “Cooperative Beamforming” to IoT systems is two different channels for data transmission and reception.
proved to achieve higher energy efficiency, in addition to pro- The two dedicated channels transmission can be considered
vide a simple framework to exploit the physical layer security as a waste of the scarce spectrum, in addition to the latency
in the corresponding system [209]. resulting from sensing the white space for the two channels.
Therefore, sensing spectrum in transmitting in FD manner can
improve the system performance remarkably [215]. Another
H. Cognitive Radio Networks approach of FD mode in CRNs, is primary and secondary
Multi-antenna application in CRNs is a desirable solution systems cooperation, where the secondary base station relays
for compensating high energy consumption due to continuous the primary signal, and in return it can transmit its own sec-
CR operations like spectrum sensing, sharing and mobility, in ondary signal [167]. Again, working in HD manner in this
addition to the subsequent signal processing and signals trans- scheme will make the cooperation process less productive
missions [210]. However, beamforming EH in CRNs hasn’t due to the need of two orthogonal white space resources and
been addressed adequately in the literature especially for the urge to work in DF mode to improve the system through-
underlay paradigm. Underlay scheme has gained a lot of atten- put. Exploiting FD mode in CRNs enhances both spectrum
tion, as it has the potential of increasing total system spectrum utilization and overall system throughput. Furthermore, FD
usage efficiency. Beamforming EH technique helps the sec- implementation can be extended to secure CRNs to enhance
ondary users in this paradigm to mitigate the interference the information security via transmitting a jamming signal
to primary system as locating both primary and secondary to impair the eavesdropper channel [216]. Yet, working in
users and exploiting the extra spatial dimension by transmit- FD mode yields unwanted self-interference signal resulting
ting in the opposite direction of the primary users yield lower from the small distance between transmitting and receiv-
interference levels at the primary system [211]. In addition, ing antennas, making the FD system impractical. However,
beamforming EH can be exploited in the hybrid underlay-relay beamforming can be exploited at the receiving and transmit-
paradigm, where secondary users are allowed to transmit even ting antennas to mitigate the self-interference signal. Since
if the interference caused is at higher levels, but in return they each communication link in FD system forms beams, a min-
relay the primary systems’ signals [156], [212]. EH technique imized self-interference level with higher data rate can be
can power the secondary users as an incentive for relaying achieved [217]. Moreover, in beamforming EH approach this
the primary users signals without causing high interference by self-interference signal can be seen as another source of energy
means of the directional transmissions of beamforming in what if harvested at the FD device, or harvest-and-jam approach
referred to as harvest and cooperate scheme [213]. Hybrid can be exploited to use the harvested power in transmitting
overlay-underlay paradigm [214] is another hybrid paradigm jamming signals to improve the information security of CRNs.
that can be a potential application for beamforming EH. The However, beamforming energy harvesting approaches have not
hybrid overlay-underlay paradigm captures the benefits of both yet been investigated in detail with respect to FD-CRNs. More
the overlay and underlay spectrum access techniques, by using studies are needed to take EH-beamforming into considera-
both non-utilized and under-utilized spectra. In this hybrid tion for self-interference harvesting and exploiting physical
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1355

layer security in FD-CRNs. Self-interference and spectrum [18] Z. Ding et al., “Application of smart antenna technologies in simulta-
sensing approaches that exploit EH-beamforming represent neous wireless information and power transfer,” IEEE Commun. Mag.,
vol. 53, no. 4, pp. 86–93, Apr. 2015.
challenging future research directions in FD-CRNs. [19] S. Bi, Y. Zeng, and R. Zhang, “Wireless powered communication
networks: An overview,” IEEE Wireless Commun., vol. 23, no. 2,
pp. 10–18, Apr. 2016.
VIII. C ONCLUSION [20] S. Timotheou, I. Krikidis, G. Zheng, and B. Ottersten, “Beamforming
for MISO interference channels with QoS and RF energy trans-
A contemporary survey of the existing literature in the fer,” IEEE Trans. Wireless Commun., vol. 13, no. 5, pp. 2646–2658,
domain of beamforming wireless energy harvesting enabled May 2014.
[21] A. D. Wyner, “The wire-tap channel,” Bell Syst. Tech. J., vol. 54, no. 8,
communication has been presented. Due to the advantages pp. 1355–1387, Oct. 1975.
of multiple-antenna techniques in wireless energy harvesting [22] S. Goel and R. Negi, “Guaranteeing secrecy using artificial noise,”
communication systems, many new research is performed in IEEE Trans. Wireless Commun., vol. 7, no. 6, pp. 2180–2189,
Jun. 2008.
this area. Literature has been categorized according to the [23] G. Kravtsova, Y. Karpitski, D. Park, and J. Yi, “Efficiency of transmis-
network topology, broadcast, relay and cognitive radio chan- sion techniques in multiple-input single-output (MISO) communication
nel. The different approaches in the corresponding networks system,” in Proc. IEEE 66th VTC Conf., 2007, pp. 432–436.
[24] J. Mietzner, R. Schober, L. Lampe, W. H. Gerstacker, and P. A. Hoeher,
are organized according to the different beamforming design “Multiple-antenna techniques for wireless communications—A com-
objectives approaches, with paying more attention to physical prehensive literature survey,” IEEE Commun. Surveys Tuts., vol. 11,
layer security implementation. Feedback process and future no. 2, pp. 87–105, 2nd Quart., 2009.
[25] V. Chandrasekhar, M. Kountouris, and J. G. Andrews, “Coverage
challenges and applications have also been discussed. in multi-antenna two-tier networks,” IEEE Trans. Wireless Commun.,
vol. 8, no. 10, pp. 5314–5327, Oct. 2009.
[26] B. D. Van Veen and K. M. Buckley, “Beamforming: A versatile
R EFERENCES approach to spatial filtering,” IEEE ASSP Mag., vol. ASSPM-5, no. 2,
pp. 4–24, Apr. 1988.
[1] E. Hossain, M. Rasti, H. Tabassum, and A. Abdelnasser, “Evolution [27] H. Krim and M. Viberg, “Two decades of array signal process-
toward 5G multi-tier cellular wireless networks: An interference ing research: The parametric approach,” IEEE Signal Process. Mag.,
management perspective,” IEEE Wireless Commun., vol. 21, no. 3, vol. 13, no. 4, pp. 67–94, Jul. 1996.
pp. 118–127, Jun. 2014. [28] Y.-W. P. Hong, P.-C. Lan, and C.-C. J. Kuo, “Enhancing physical-
[2] V. Raghunathan, S. Ganeriwal, and M. Srivastava, “Emerging tech- layer secrecy in multiantenna wireless systems: An overview of signal
niques for long lived wireless sensor networks,” IEEE Commun. Mag., processing approaches,” IEEE Signal Process. Mag., vol. 30, no. 5,
vol. 44, no. 4, pp. 108–114, Apr. 2006. pp. 29–40, Sep. 2013.
[3] K. Pentikousis, “In search of energy-efficient mobile networking,” [29] Y. Shi, J. Zhang, and K. B. Letaief, “CSI overhead reduction with
IEEE Commun. Mag., vol. 48, no. 1, pp. 95–103, Jan. 2010. stochastic beamforming for cloud radio access networks,” in Proc.
[4] T. Chen, Y. Yang, H. Zhang, H. Kim, and K. Horneman, “Network IEEE Int. Conf. Commun. (ICC), 2014, pp. 5154–5159.
energy saving technologies for green wireless access networks,” IEEE [30] C. Xiong, X. Zhang, K. Wu, and D. Yang, “An efficient parallel algo-
Wireless Commun., vol. 18, no. 5, pp. 30–38, Oct. 2011. rithm with partial decision feedback for near-optimal MIMO detection,”
[5] V. Sharma, U. Mukherji, V. Joseph, and S. Gupta, “Optimal energy in Proc. Glob. Telecommun. Conf. (GLOBECOM), 2009, pp. 1–5.
management policies for energy harvesting sensor nodes,” IEEE Trans. [31] T. Kaiser, A. Bourdoux, M. Rupp, and U. Heute, “Implementation
Wireless Commun., vol. 9, no. 4, pp. 1326–1336, Apr. 2010. aspects and testbeds for MIMO systems,” EURASIP J. Adv. Signal
[6] F. Zhang et al., “Wireless energy transfer platform for medical sensors Process., vol. 2006, no. 1, pp. 1–3, 2006.
and implantable devices,” in Proc. IEEE Annu. Int. Conf. Eng. Med. [32] J. Jelitto and G. Fettweis, “Reduced dimension space-time processing
Biol. Soc. (EMBC), 2009, pp. 1045–1048. for multi-antenna wireless systems,” IEEE Wireless Commun., vol. 9,
[7] S. Sudevalayam and P. Kulkarni, “Energy harvesting sensor nodes: no. 6, pp. 18–25, Dec. 2002.
Survey and implications,” IEEE Commun. Surveys Tuts., vol. 13, no. 3, [33] X. Chen, D. W. K. Ng, W. H. Gerstacker, and H.-H. Chen, “A sur-
pp. 443–461, 3rd Quart., 2011. vey on multiple-antenna techniques for physical layer security,” IEEE
[8] K. Huang and V. K. N. Lau, “Enabling wireless power transfer in cel- Commun. Surveys Tuts., vol. 19, no. 2, pp. 1027–1053, 2nd Quart.,
lular networks: Architecture, modeling and deployment,” IEEE Trans. 2017.
Wireless Commun., vol. 13, no. 2, pp. 902–912, Feb. 2014. [34] Z.-Q. Luo, W.-K. Ma, A. M.-C. So, Y. Ye, and S. Zhang, “Semidefinite
[9] P. Grover and A. Sahai, “Shannon meets Tesla: Wireless information relaxation of quadratic optimization problems,” IEEE Signal Process.
and power transfer,” in Proc. IEEE Int. Symp. Inf. Theory (ISIT), 2010, Mag., vol. 27, no. 3, pp. 20–34, May 2010.
pp. 2363–2367. [35] F. Alizadeh and D. Goldfarb, “Second-order cone programming,” Math.
[10] S. Bi, C. K. Ho, and R. Zhang, “Wireless powered communication: Program. B, vol. 95, no. 1, pp. 3–51, 2003.
Opportunities and challenges,” IEEE Commun. Mag., vol. 53, no. 4, [36] B. Marks and G. P. Wright, “A general inner approximation algorithm
pp. 117–125, Apr. 2015. for nonconvex mathematical programs,” Oper. Res., vol. 26, no. 4,
[11] R. Wang and D. R. Brown, “Throughput maximization in wireless pp. 681–683, 1978.
powered communication networks with energy saving,” in Proc. IEEE [37] X. Lu, P. Wang, D. Niyato, D. I. Kim, and Z. Han, “Wireless networks
48th Asilomar Conf. Signals Syst. Comput., 2014, pp. 516–520. with RF energy harvesting: A contemporary survey,” IEEE Commun.
[12] L. R. Varshney, “Transporting information and energy simultaneously,” Surveys Tuts., vol. 17, no. 2, pp. 757–789, 2nd Quart., 2015.
in Proc. IEEE Int. Symp. Inf. Theory (ISIT), 2008, pp. 1612–1616. [38] N. Zhao et al., “Exploiting interference for energy harvesting:
[13] X. Zhou, R. Zhang, and C. K. Ho, “Wireless information and power A survey, research issues, and challenges,” IEEE Access, vol. 5,
transfer: Architecture design and rate-energy tradeoff,” IEEE Trans. pp. 10403–10421, 2017.
Commun., vol. 61, no. 11, pp. 4754–4767, Nov. 2013. [39] X. Chen, D. W. K. Ng, and H.-H. Chen, “Secrecy wireless informa-
[14] R. Zhang and C. K. Ho, “MIMO broadcasting for simultaneous wire- tion and power transfer: Challenges and opportunities,” IEEE Wireless
less information and power transfer,” IEEE Trans. Wireless Commun., Commun., vol. 23, no. 2, pp. 54–61, Apr. 2016.
vol. 12, no. 5, pp. 1989–2001, May 2013. [40] M.-L. Ku, W. Li, Y. Chen, and K. J. R. Liu, “Advances in energy
[15] H. Ju and R. Zhang, “A novel mode switching scheme utilizing ran- harvesting communications: Past, present, and future challenges,” IEEE
dom beamforming for opportunistic energy harvesting,” IEEE Trans. Commun. Surveys Tuts., vol. 18, no. 2, pp. 1384–1412, 2nd Quart.,
Wireless Commun., vol. 13, no. 4, pp. 2150–2162, Apr. 2014. 2016.
[16] W. C. Brown, “The history of power transmission by radio [41] C. Shen, W.-C. Li, and T.-H. Chang, “Wireless information and energy
waves,” IEEE Trans. Microw. Theory Techn., vol. MTT-32, no. 9, transfer in multi-antenna interference channel,” IEEE Trans. Signal
pp. 1230–1242, Sep. 1984. Process., vol. 62, no. 23, pp. 6249–6264, Dec. 2014.
[17] T. Le, K. Mayaram, and T. Fiez, “Efficient far-field radio frequency [42] Q. Shi, L. Liu, W. Xu, and R. Zhang, “Joint transmit beamforming
energy harvesting for passively powered sensor networks,” IEEE J. and receive power splitting for MISO SWIPT systems,” IEEE Trans.
Solid-State Circuits, vol. 43, no. 5, pp. 1287–1302, May 2008. Wireless Commun., vol. 13, no. 6, pp. 3269–3280, Jun. 2014.
1356 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

[43] Q. Shi, C. Peng, W. Xu, and Y. Wang, “Joint transceiver design [65] H. Zhang, K. Song, Y. Huang, and L. Yang, “Energy harvesting bal-
for MISO SWIPT interference channel,” in Proc. IEEE Int. Conf. ancing technique for robust beamforming in multiuser MISO SWIPT
Acoust. Speech Signal Process. (ICASSP), Florence, Italy, 2014, system,” in Proc. Int. Conf. Wireless Commun. Signal Process. (WCSP),
pp. 4753–4757. Hangzhou, China, 2013, pp. 1–5.
[44] Q. Shi, W. Xu, T.-H. Chang, Y. Wang, and E. Song, “Joint beamform- [66] R. Feng, M. Dai, and H. Wang, “Distributed beamforming in
ing and power splitting for MISO interference channel with SWIPT: MISO SWIPT system,” IEEE Trans. Veh. Technol., vol. 66, no. 6,
An SOCP relaxation and decentralized algorithm,” IEEE Trans. Signal pp. 5440–5445, Jun. 2017.
Process., vol. 62, no. 23, pp. 6194–6208, Dec. 2014. [67] S. Luo, J. Xu, T. J. Lim, and R. Zhang, “Capacity region of MISO
[45] M. R. Khandaker and K.-K. Wong, “SWIPT in MISO multicasting broadcast channel for simultaneous wireless information and power
systems,” IEEE Wireless Commun. Lett., vol. 3, no. 3, pp. 277–280, transfer,” IEEE Trans. Commun., vol. 63, no. 10, pp. 3856–3868,
2014. Oct. 2015.
[46] Z. Zhu, Z. Wang, K.-J. Lee, Z. Chu, and I. Lee, “Robust transceiver [68] C. Xing, N. Wang, J. Ni, Z. Fei, and J. Kuang, “MIMO beamforming
designs in multiuser MISO broadcasting with simultaneous wireless designs with partial CSI under energy harvesting constraints,” IEEE
information and power transmission,” J. Commun. Netw., vol. 18, no. 2, Signal Process. Lett., vol. 20, no. 4, pp. 363–366, Apr. 2013.
pp. 173–181, Apr. 2016. [69] W. Huang, H. Chen, Y. Li, and B. Vucetic, “On the performance of
[47] J. Liao, M. R. Khandaker, and K.-K. Wong, “Robust power-splitting multi-antenna wireless-powered communications with energy beam-
SWIPT beamforming for broadcast channels,” IEEE Commun. Lett., forming,” IEEE Trans. Veh. Technol., vol. 65, no. 3, pp. 1801–1808,
vol. 20, no. 1, pp. 181–184, Jan. 2016. Mar. 2016.
[48] Z. Zhu, Z. Wang, X. Gui, and X. Gao, “Robust downlink beamform- [70] C. Zhong, X. Chen, Z. Zhang, and G. K. Karagiannidis, “Wireless-
ing and power splitting design in multiuser MISO SWIPT system,” powered communications: Performance analysis and optimization,”
in Proc. IEEE Int. Conf. Commun. (ICCC), Shanghai, China, 2014, IEEE Trans. Commun., vol. 63, no. 12, pp. 5178–5190, Dec. 2015.
pp. 271–275. [71] S. X. Wu, Q. Li, W.-K. Ma, and A. M.-C. So, “Robust transmit
[49] Z. Zhu, K.-J. Lee, Z. Wang, Z. Chu, and I. Lee, “Robust precoding designs for an energy harvesting multicast system,” in Proc. IEEE Int.
methods for multiuser MISO wireless energy harvesting systems,” in Conf. Acoust. Speech Signal Process. (ICASSP), Florence, Italy, 2014,
Proc. IEEE 82nd Veh. Technol. Conf. (VTC Fall), Boston, MA, USA, pp. 4748–4752.
2015, pp. 1–5. [72] L. Liu, R. Zhang, and K.-C. Chua, “Multi-antenna wireless powered
[50] Z. Chu, Z. Zhu, W. Xiang, and J. Hussein, “Robust beamforming communication with energy beamforming,” IEEE Trans. Commun.,
and power splitting design in MISO SWIPT downlink system,” IET vol. 62, no. 12, pp. 4349–4361, Dec. 2014.
Commun., vol. 10, no. 6, pp. 691–698, Apr. 2016. [73] Q. Sun, G. Zhu, C. Shen, X. Li, and Z. Zhong, “Joint beamform-
[51] M. R. A. Khandaker and K.-K. Wong, “QoS-based multicast beam- ing design and time allocation for wireless powered communication
forming for SWIPT,” in Proc. IEEE 11th Annu. Int. Conf. Sens. networks,” IEEE Commun. Lett., vol. 18, no. 10, pp. 1783–1786,
Commun. Netw. Workshops (SECON), Singapore, 2014, pp. 62–67. Oct. 2014.
[52] Z. Yue, H. Gao, C. Yuen, T. Lv, and J. Zeng, “Low complexity [74] N. A. Arshad, D. T. Ngo, H. D. Tuan, and S. Durrani, “Iterative
joint beamforming and power splitting for massive MIMO multicasting optimization for max-min SINR in dense small-cell multiuser MISO
SWIPT,” in Proc. IEEE/CIC Symp. Wireless Commun. Syst. (ICCC), SWIPT system,” in Proc. IEEE Glob. Conf. Signal Inf. Process.
Shenzhen, China, 2015, pp. 1–6. (GlobalSIP), Orlando, FL, USA, 2015, pp. 1392–1396.
[53] W. Yu and T. Lan, “Transmitter optimization for the multi-antenna [75] A. A. Nasir, D. T. Ngo, H. D. Tuan, S. Durrani, and D. I. Kim, “Secure
downlink with per-antenna power constraints,” IEEE Trans. Signal beamforming for max-min SINR in multi-cell SWIPT systems,”
Process., vol. 55, no. 6, pp. 2646–2660, Jun. 2007. in Proc. Workshop Wireless Powered Commun. Netw. Theory Ind.
[54] S. Timotheou, G. Zheng, C. Masouros, and I. Krikidis, “Exploiting con- Challenges (WPCNets), 2016, pp. 1–6.
structive interference for simultaneous wireless information and power [76] Z. Zhu, K.-J. Lee, Z. Wang, and I. Lee, “Robust beamforming and
transfer in multiuser downlink systems,” IEEE J. Sel. Areas Commun., power splitting design in distributed antenna system with SWIPT under
vol. 34, no. 5, pp. 1772–1784, May 2016. bounded channel uncertainty,” in Proc. IEEE 81st Veh. Technol. Conf.
[55] T. Peng, F. Wang, Y. Huang, and X. Wang, “Robust transceiver (VTC Spring), 2015, pp. 1–5.
optimization for MISO SWIPT interference channel: A decentralized [77] K. Huang and E. Larsson, “Simultaneous information and power trans-
approach,” in Proc. IEEE 83rd Veh. Technol. Conf. (VTC Spring), fer for broadband wireless systems,” IEEE Trans. Signal Process.,
Nanjing, China, 2016, pp. 1–5. vol. 61, no. 23, pp. 5972–5986, Dec. 2013.
[56] K. W. Choi and D. I. Kim, “Stochastic optimal control for wireless [78] S. X. Wu, A. M.-C. So, and W.-K. Ma, “Rank-two transmit beam-
powered communication networks,” IEEE Trans. Wireless Commun., formed ALAMOUTI space-time coding for physical-layer multicast-
vol. 15, no. 1, pp. 686–698, Jan. 2016. ing,” in Proc. IEEE ICASSP, Kyoto, Japan, 2012, pp. 2793–2796.
[57] Q. Shi, C. Peng, W. Xu, M. Hong, and Y. Cai, “Energy efficiency [79] C. E. Shannon, “Communication theory of secrecy systems,” Bell Labs
optimization for MISO SWIPT systems with zero-forcing beamform- Tech. J., vol. 28, no. 4, pp. 656–715, Oct. 1949.
ing,” IEEE Trans. Signal Process., vol. 64, no. 4, pp. 842–854, [80] W.-C. Liao, T.-H. Chang, W.-K. Ma, and C.-Y. Chi, “QoS-based trans-
Feb. 2016. mit beamforming in the presence of eavesdroppers: An optimized
[58] Q.-D. Vu, L.-N. Tran, R. Farrell, and E.-K. Hong, “An efficiency artificial-noise-aided approach,” IEEE Trans. Signal Process., vol. 59,
maximization design for SWIPT,” IEEE Signal Process. Lett., vol. 22, no. 3, pp. 1202–1216, Mar. 2011.
no. 12, pp. 2189–2193, Dec. 2015. [81] J. Kang et al., “Toward secure energy harvesting cooperative networks,”
[59] M. Sheng et al., “Energy efficient beamforming in MISO heteroge- IEEE Commun. Mag., vol. 53, no. 8, pp. 114–121, Aug. 2015.
neous cellular networks with wireless information and power transfer,” [82] L. Liu, R. Zhang, and K.-C. Chua, “Secrecy wireless information and
IEEE J. Sel. Areas Commun., vol. 34, no. 4, pp. 954–968, Apr. 2016. power transfer with MISO beamforming,” IEEE Trans. Signal Process.,
[60] X. Chen, X. Wang, and X. Chen, “Energy-efficient optimization for vol. 62, no. 7, pp. 1850–1863, Apr. 2014.
wireless information and power transfer in large-scale mimo systems [83] Z. Zhu, Z. Chu, Z. Wang, and I. Lee, “Joint optimization of AN-aided
employing energy beamforming,” IEEE Wireless Commun. Lett., vol. 2, beamforming and power splitting designs for MISO secrecy channel
no. 6, pp. 667–670, Dec. 2013. with SWIPT,” in Proc. IEEE Int. Conf. Commun. (ICC), 2016, pp. 1–6.
[61] X. Chen, C. Yuen, and Z. Zhang, “Wireless energy and information [84] Z. Chu, Z. Zhu, M. Johnston, and S. Y. Le Goff, “Simultaneous wireless
transfer tradeoff for limited-feedback multiantenna systems with energy information power transfer for MISO secrecy channel,” IEEE Trans.
beamforming,” IEEE Trans. Veh. Technol., vol. 63, no. 1, pp. 407–412, Veh. Technol., vol. 65, no. 9, pp. 6913–6925, Sep. 2016.
Jan. 2014. [85] Q. Shi, W. Xu, J. Wu, E. Song, and Y. Wang, “Secure beamforming
[62] J. Xu, L. Liu, and R. Zhang, “Multiuser MISO beamforming for simul- for MIMO broadcasting with wireless information and power trans-
taneous wireless information and power transfer,” IEEE Trans. Signal fer,” IEEE Trans. Wireless Commun., vol. 14, no. 5, pp. 2841–2853,
Process., vol. 62, no. 18, pp. 4798–4810, Sep. 2014. May 2015.
[63] Z. Zhu et al., “Robust beamforming for multiuser MISO interference [86] A. A. Nasir, H. D. Tuan, T. Q. Duong, and H. V. Poor, “Secrecy
channel with energy harvesting constraint,” in Proc. IEEE 10th rate beamforming for multicell networks with information and energy
Int. Conf. Commun. Netw. (ChinaCom), Shanghai, China, 2015, harvesting,” IEEE Trans. Signal Process., vol. 65, no. 3, pp. 677–689,
pp. 126–131. Feb. 2017.
[64] Z. Xiang and M. Tao, “Robust beamforming for wireless information [87] T. A. Le, H. X. Nguyen, Q.-T. Vien, and M. Karamanoglu, “Secure
and power transmission,” IEEE Wireless Commun. Lett., vol. 1, no. 4, information transmission and power transfer in cellular networks,”
pp. 372–375, Aug. 2012. IEEE Commun. Lett., vol. 19, no. 9, pp. 1532–1535, Sep. 2015.
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1357

[88] B. Zhu, J. Ge, Y. Huang, Y. Yang, and M. Lin, “Rank-two beamformed [111] Z. Fang, X. Yuan, and X. Wang, “Distributed energy beamforming for
secure multicasting for wireless information and power transfer,” IEEE simultaneous wireless information and power transfer in the two-way
Signal Process. Lett., vol. 21, no. 2, pp. 199–203, Feb. 2014. relay channel,” IEEE Signal Process. Lett., vol. 22, no. 6, pp. 656–660,
[89] D. W. K. Ng, R. Schober, and H. Alnuweiri, “Secure layered transmis- Jun. 2015.
sion in multicast systems with wireless information and power transfer,” [112] Z. Zhou, M. Peng, Z. Zhao, W. Wang, and R. S. Blum, “Wireless-
in Proc. IEEE Int. Conf. Commun. (ICC), Sydney, NSW, Australia, powered cooperative communications: Power-splitting relaying with
2014, pp. 5389–5395. energy accumulation,” IEEE J. Sel. Areas Commun., vol. 34, no. 4,
[90] Z. Zhu, Z. Chu, Z. Wang, and I. Lee, “Outage constrained robust beam- pp. 969–982, Apr. 2016.
forming for secure broadcasting systems with energy harvesting,” IEEE [113] G. Li, P. Ren, G. Lv, and Q. Du, “High-rate relay beamforming for
Trans. Wireless Commun., vol. 15, no. 11, pp. 7610–7620, Nov. 2016. simultaneous wireless information and power transfer,” Electron. Lett.,
[91] H. Zhang, C. Li, Y. Huang, and L. Yang, “Secure beamforming vol. 50, no. 23, pp. 1759–1761, Nov. 2014.
for SWIPT in multiuser MISO broadcast channel with confiden- [114] Q. Zhang, Q. Li, and J. Qin, “Beamforming design for OSTBC-based
tial messages,” IEEE Commun. Lett., vol. 19, no. 8, pp. 1347–1350, AF-MIMO two-way relay networks with simultaneous wireless infor-
Aug. 2015. mation and power transfer,” IEEE Trans. Veh. Technol., vol. 65, no. 9,
[92] D. W. K. Ng, E. S. Lo, and R. Schober, “Robust beamforming pp. 7285–7296, Sep. 2016.
for secure communication in systems with wireless information and [115] M. Mohammadi et al., “Throughput analysis and optimization of
power transfer,” IEEE Trans. Wireless Commun., vol. 13, no. 8, wireless-powered multiple antenna full-duplex relay systems,” IEEE
pp. 4599–4615, Aug. 2014. Trans. Commun., vol. 64, no. 4, pp. 1769–1785, Apr. 2016.
[93] H. Zhang, C. Li, Y. Huang, and L. Yang, “Simultaneous wire- [116] F. Benkhelifa, A. S. Salem, and M.-S. Alouini, “Sum-rate enhancement
less information and power transfer in a MISO broadcast channel in multiuser MIMO decode-and-forward relay broadcasting channel
with confidential messages,” in Proc. IEEE Glob. Commun. Conf. with energy harvesting relays,” IEEE J. Sel. Areas Commun., vol. 34,
(GLOBECOM), San Diego, CA, USA, 2015, pp. 1–6. no. 12, pp. 3675–3684, Dec. 2016.
[94] M. R. A. Khandaker and K.-K. Wong, “Robust secrecy beamforming [117] Q. Li, Q. Zhang, and J. Qin, “Beamforming in non-regenerative
with energy-harvesting eavesdroppers,” IEEE Wireless Commun. Lett., two-way multi-antenna relay networks for simultaneous wireless infor-
vol. 4, no. 1, pp. 10–13, Feb. 2015. mation and power transfer,” IEEE Trans. Wireless Commun., vol. 13,
[95] X. Chen and L. Lei, “Energy-efficient optimization for physical layer no. 10, pp. 5509–5520, Oct. 2014.
security in multi-antenna downlink networks with QoS guarantee,” [118] D. Li, C. Shen, and Z. Qiu, “Two-way relay beamforming for sum-
IEEE Commun. Lett., vol. 17, no. 4, pp. 637–640, Apr. 2013. rate maximization and energy harvesting,” in Proc. IEEE Int. Conf.
[96] M. R. Khandaker, K.-K. Wong, Y. Zhang, and Z. Zheng, Commun. (ICC), 2013, pp. 3115–3120.
“Probabilistically robust SWIPT for secrecy MISOME systems,” IEEE
[119] D. Li, C. Shen, and Z. Qiu, “Sum rate maximization and energy
Trans. Inf. Forensics Security, vol. 12, no. 1, pp. 211–226, Jan. 2017.
harvesting for two-way AF relay systems with imperfect CSI,” in
[97] A. A. Nasir, X. Zhou, S. Durrani, and R. A. Kennedy, “Throughput Proc. IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP), 2013,
and ergodic capacity of wireless energy harvesting based DF relaying pp. 4958–4962.
network,” in Proc. IEEE Int. Conf. Commun. (ICC), Sydney, NSW,
[120] S. Hu, Z. Ding, and Q. Ni, “Beamforming optimisation in energy
Australia, 2014, pp. 4066–4071.
harvesting cooperative full-duplex networks with self-energy recycling
[98] I. Ahmed, A. Ikhlef, R. Schober, and R. K. Mallik, “Power allocation protocol,” IET Commun., vol. 10, no. 7, pp. 848–853, May 2016.
for conventional and buffer-aided link adaptive relaying systems with
energy harvesting nodes,” IEEE Trans. Wireless Commun., vol. 13, [121] Y. Zeng and R. Zhang, “Full-duplex wireless-powered relay with
no. 3, pp. 1182–1195, Mar. 2014. self-energy recycling,” IEEE Wireless Commun. Lett., vol. 4, no. 2,
pp. 201–204, Apr. 2015.
[99] H. Li, N. Jaggi, and B. Sikdar, “Relay scheduling for cooperative com-
munications in sensor networks with energy harvesting,” IEEE Trans. [122] Y. Yuan, Z. Chu, Z. Ding, K. Cumanan, and M. Johnston, “Joint relay
Wireless Commun., vol. 10, no. 9, pp. 2918–2928, Sep. 2011. beamforming and power splitting ratio optimization in a multi-antenna
[100] Y. Luo, J. Zhang, and K. B. Letaief, “Optimal scheduling and power relay network,” in Proc. IEEE 6th Int. Conf. Wireless Commun. Signal
allocation for two-hop energy harvesting communication systems,” Process. (WCSP), Hefei, China, 2014, pp. 1–5.
IEEE Trans. Wireless Commun., vol. 12, no. 9, pp. 4729–4741, [123] Y. Cai, M.-M. Zhao, Q. Shi, B. Champagne, and M.-J. Zhao, “Joint
Sep. 2013. transceiver design algorithms for multiuser MISO relay systems
[101] A. Minasian, S. ShahbazPanahi, and R. S. Adve, “Energy harvesting with energy harvesting,” IEEE Trans. Commun., vol. 64, no. 10,
cooperative communication systems,” IEEE Trans. Wireless Commun., pp. 4147–4164, Oct. 2016.
vol. 13, no. 11, pp. 6118–6131, Nov. 2014. [124] Z. Wen, X. Liu, N. C. Beaulieu, R. Wang, and S. Wang, “Joint source
[102] J. N. Laneman, D. N. Tse, and G. W. Wornell, “Cooperative diversity and relay beamforming design for full-duplex MIMO AF relay SWIPT
in wireless networks: Efficient protocols and outage behavior,” IEEE systems,” IEEE Commun. Lett., vol. 20, no. 2, pp. 320–323, Feb. 2016.
Trans. Inf. Theory, vol. 50, no. 12, pp. 3062–3080, Dec. 2004. [125] Y. Chen, Z. Wen, N. C. Beaulieu, S. Wang, and J. Sun, “Joint
[103] T.-N. Do and B. An, “Cooperative spectrum sensing schemes with the source-relay design in a MIMO two-hop power-splitting-based relay-
interference constraint in cognitive radio networks,” Sensors, vol. 14, ing network,” IEEE Commun. Lett., vol. 19, no. 10, pp. 1746–1749,
no. 5, pp. 8037–8056, 2014. Oct. 2015.
[104] N. T. Do, V. N. Q. Bao, and B. An, “A relay selection protocol for [126] Y. Chen, Z. Wen, S. Wang, J. Sun, and M. Li, “Joint relay beam-
wireless energy harvesting relay networks,” in Proc. IEEE Int. Conf. forming and source receiving in MIMO two-way AF relay network
Adv. Technol. Commun. (ATC), 2015, pp. 243–247. with energy harvesting,” in Proc. IEEE 81st Veh. Technol. Conf. (VTC
[105] S. Leung-Yan-Cheong and M. Hellman, “The Gaussian wire-tap chan- Spring), Glasgow, U.K., 2015, pp. 1–5.
nel,” IEEE Trans. Inf. Theory, vol. 24, no. 4, pp. 451–456, Jul. 1978. [127] Z. Wen, X. Liu, S. Zheng, and W. Guo, “Joint source and relay
[106] N. Yang, H. A. Suraweera, I. B. Collings, and C. Yuen, “Physical design for MIMO two-way relay networks with SWIPT,” IEEE
layer security of TAS/MRC with antenna correlation,” IEEE Trans. Trans. Veh. Technol., vol. 67, no. 1, pp. 822–826, Jan. 2018,
Inf. Forensics Security, vol. 8, no. 1, pp. 254–259, Jan. 2013. doi: 10.1109/TVT.2017.2727061.
[107] S. Gong, L. Duan, and N. Gautam, “Optimal scheduling and beam- [128] F. Wang, X. Wang, Z. Fang, and R. Zhang, “Optimal beamforming
forming in relay networks with energy harvesting constraints,” IEEE design for two-way relaying with simultaneous energy harvesting,” in
Trans. Wireless Commun., vol. 15, no. 2, pp. 1226–1238, Feb. 2016. Proc. IEEE 6th Int. Symp. Commun. Control Signal Process. (ISCCSP),
[108] J. Huang, Q. Li, Q. Zhang, G. Zhang, and J. Qin, “Relay beam- 2014, pp. 331–335.
forming for amplify-and-forward multi-antenna relay networks with [129] R. Vahidnia, A. Anpalagan, and J. Mirzaei, “Diversity combining
energy harvesting constraint,” IEEE Signal Process. Lett., vol. 21, no. 4, in bi-directional relay networks with energy harvesting nodes,” IET
pp. 454–458, Apr. 2014. Commun., vol. 10, no. 2, pp. 207–211, Jan. 2016.
[109] X. Wang, X. Yuan, and Z. Fang, “Distributed energy beamforming [130] I. Krikidis, S. Sasaki, S. Timotheou, and Z. Ding, “A low complexity
and information transfer: A case study for multiway relay channels,” antenna switching for joint wireless information and energy transfer
in Proc. IEEE/CIC Symp. Wireless Commun. Syst. (ICCC), Shanghai, in MIMO relay channels,” IEEE Trans. Commun., vol. 62, no. 5,
China, 2014, pp. 670–675. pp. 1577–1587, May 2014.
[110] C. Zhong, G. Zheng, Z. Zhang, and G. K. Karagiannidis, “Optimum [131] F. Benkhelifa and M.-S. Alouini, “Precoding design of MIMO amplify-
wirelessly powered relaying,” IEEE Signal Process. Lett., vol. 22, and-forward communication system with an energy harvesting relay
no. 10, pp. 1728–1732, Oct. 2015. and possibly imperfect CSI,” IEEE Access, vol. 5, pp. 578–594, 2017.
1358 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

[132] H. Liang, C. Zhong, H. A. Suraweera, G. Zheng, and Z. Zhang, [155] M. Gastpar, “On capacity under receive and spatial spectrum-sharing
“Optimization and analysis of wireless powered multi-antenna coop- constraints,” IEEE Trans. Inf. Theory, vol. 53, no. 2, pp. 471–487,
erative systems,” IEEE Trans. Wireless Commun., vol. 16, no. 5, Feb. 2007.
pp. 3267–3281, May 2017. [156] X. Chen, H.-H. Chen, and W. Meng, “Cooperative communications
[133] Y. Chen, R. Shi, W. Feng, and N. Ge, “AF relaying with energy har- for cognitive radio networks—From theory to applications,” IEEE
vesting source and relay,” IEEE Trans. Veh. Technol., vol. 66, no. 1, Commun. Surveys Tuts., vol. 16, no. 3, pp. 1180–1192, 3rd Quart.,
pp. 874–879, Jan. 2017. 2014.
[134] A. Almradi and K. A. Hamdi, “The performance of wireless powered [157] I. Macaluso, T. K. Forde, L. DaSilva, and L. Doyle, “Impact of cogni-
MIMO relaying systems with energy beamforming,” in Proc. IEEE Int. tive radio: Recognition and informed exploitation of grey spectrum
Conf. Commun. (ICC), 2016, pp. 1–6. opportunities,” IEEE Veh. Technol. Mag., vol. 7, no. 2, pp. 85–90,
[135] S. Modem and S. Prakriya, “Performance of analog network coding Jun. 2012.
based two-way EH relay with beamforming,” IEEE Trans. Commun., [158] A. Furtado et al., “Detection of licensed users’ activity in a random
vol. 65, no. 4, pp. 1518–1535, Apr. 2017. access ultra wideband cognitive system,” in Proc. IEEE Int. Conf.
[136] D. Mishra and S. De, “Optimal relay placement in two-hop RF Ultra-WideBand (ICUWB), Paris, France, 2014, pp. 91–95.
energy transfer,” IEEE Trans. Commun., vol. 63, no. 5, pp. 1635–1647, [159] H. O. Kpojime and G. A. Safdar, “Interference mitigation in cognitive-
May 2015. radio-based femtocells,” IEEE Commun. Surveys Tuts., vol. 17, no. 3,
[137] H. Kim, J. Kang, S. Jeong, K. E. Lee, and J. Kang, “Secure beamform- pp. 1511–1534, 3rd Quart., 2015.
ing and self-energy recycling with full-duplex wireless-powered relay,” [160] M. Monemi, M. Rasti, and E. Hossain, “On characterization of fea-
in Proc. IEEE 13th Annu Consum. Commun. Netw. Conf. (CCNC), sible interference regions in cognitive radio networks,” IEEE Trans.
Las Vegas, NV, USA, 2016, pp. 662–667. Commun., vol. 64, no. 2, pp. 511–524, Feb. 2016.
[138] Q. Li, W.-K. Ma, and D. Han, “Sum secrecy rate maximization for [161] Y. Chen and H.-S. Oh, “A survey of measurement-based spectrum occu-
full-duplex two-way relay networks using Alamouti-based rank-two pancy modeling for cognitive radios,” IEEE Commun. Surveys Tuts.,
beamforming,” IEEE J. Sel. Topics Signal Process., vol. 10, no. 8, vol. 18, no. 1, pp. 848–859, 1st Quart., 2016.
pp. 1359–1374, Dec. 2016. [162] Y. Saleem and M. H. Rehmani, “Primary radio user activity models for
[139] H. Xing, K.-K. Wong, and A. Nallanathan, “Secure wireless energy cognitive radio networks: A survey,” J. Netw. Comput. Appl., vol. 43,
harvesting-enabled AF-relaying SWIPT networks,” in Proc. IEEE Int. pp. 1–16, Aug. 2014.
Conf. Commun. (ICC), London, U.K., 2015, pp. 2307–2312. [163] X. Xing, T. Jing, W. Cheng, Y. Huo, and X. Cheng, “Spectrum
[140] Y. Feng, Z. Yang, W.-P. Zhu, Q. Li, and B. Lv, “Robust coopera- prediction in cognitive radio networks,” IEEE Wireless Commun.,
tive secure beamforming for simultaneous wireless information and vol. 20, no. 2, pp. 90–96, Apr. 2013.
power transfer in amplify-and-forward relay networks,” IEEE Trans.
[164] M. Höyhtyä et al., “Spectrum occupancy measurements: A survey and
Veh. Technol., vol. 66, no. 3, pp. 2354–2366, Mar. 2017.
use of interference maps,” IEEE Commun. Surveys Tuts., vol. 18, no. 4,
[141] M. Zhao et al., “Joint power splitting and secure beamforming design pp. 2386–2414, 4th Quart., 2016.
in the wireless-powered untrusted relay networks,” in Proc. IEEE Glob.
[165] Y.-C. Liang, K.-C. Chen, G. Y. Li, and P. Mahonen, “Cognitive radio
Commun. Conf. (GLOBECOM), San Diego, CA, USA, 2015, pp. 1–6.
networking and communications: An overview,” IEEE Trans. Veh.
[142] Q. Li, Q. Zhang, and J. Qin, “Secure relay beamforming for simulta-
Technol., vol. 60, no. 7, pp. 3386–3407, Sep. 2011.
neous wireless information and power transfer in nonregenerative relay
networks,” IEEE Trans. Veh. Technol., vol. 63, no. 5, pp. 2462–2467, [166] S. Lee, R. Zhang, and K. Huang, “Opportunistic wireless energy har-
Jun. 2014. vesting in cognitive radio networks,” IEEE Trans. Wireless Commun.,
vol. 12, no. 9, pp. 4788–4799, Sep. 2013.
[143] H. Gao, T. Lv, W. Wang, and N. C. Beaulieu, “Energy-efficient
and secure beamforming for self-sustainable relay-aided multicast [167] G. Zheng, I. Krikidis, and B. Orn Ottersten, “Full-duplex coopera-
networks,” IEEE Signal. Process. Lett., vol. 23, no. 11, pp. 1509–1513, tive cognitive radio with transmit imperfections,” IEEE Trans. Wireless
Nov. 2016. Commun., vol. 12, no. 5, pp. 2498–2511, May 2013.
[144] B. Li, Z. Fei, and H. Chen, “Robust artificial noise-aided secure beam- [168] X. Lu, W. Xu, S. Li, J. Lin, and Z. He, “Simultaneous information and
forming in wireless-powered non-regenerative relay networks,” IEEE power transfer for relay-assisted cognitive radio networks,” in Proc.
Access, vol. 4, pp. 7921–7929, 2016. IEEE Int. Conf. Commun. Workshops (ICC), Sydney, NSW, Australia,
[145] H. Xing, K.-K. Wong, A. Nallanathan, and R. Zhang, “Wireless pow- 2014, pp. 331–336.
ered cooperative jamming for secrecy multi-AF relaying networks,” [169] F. Zhu, F. Gao, and M. Yao, “A new cognitive radio strategy for SWIPT
IEEE Trans. Wireless Commun., vol. 15, no. 12, pp. 7971–7984, system,” in Proc. Int. Workshop High Mobility Wireless Commun.
Dec. 2016. (HMWC), Beijing, China, 2014, pp. 73–77.
[146] H. Xing, K.-K. Wong, Z. Chu, and A. Nallanathan, “To harvest and [170] H. Xing, X. Kang, K.-K. Wong, and A. Nallanathan, “Optimizing
jam: A paradigm of self-sustaining friendly jammers for secure AF DF cognitive radio networks with full-duplex-enabled energy access
relaying,” IEEE Trans. Signal Process., vol. 63, no. 24, pp. 6616–6631, points,” IEEE Trans. Wireless Commun., vol. 16, no. 7, pp. 4683–4697,
Dec. 2015. Jul. 2017.
[147] G. Zhang, X. Li, M. Cui, G. Li, and L. Yang, “Signal and artificial [171] Q. Li, Q. Zhang, and J. Qin, “Beamforming for information and energy
noise beamforming for secure simultaneous wireless information and cooperation in cognitive non-regenerative two-way relay networks,”
power transfer multiple-input multiple-output relaying systems,” IET IEEE Trans. Wireless Commun., vol. 15, no. 8, pp. 5302–5313,
Commun., vol. 10, no. 7, pp. 796–804, May 2016. Aug. 2016.
[148] H. Xing, Z. Chu, Z. Ding, and A. Nallanathan, “Harvest-and- [172] X. Lu, W. Xu, S. Li, Z. Liu, and J. Lin, “Simultaneous wire-
jam: Improving security for wireless energy harvesting cooperative less information and power transfer for cognitive two-way relay-
networks,” in Proc. IEEE Glob. Commun. Conf. (GLOBECOM), 2014, ing networks,” in Proc. IEEE 25th Annu. Int. Symp. Pers. Indoor
pp. 3145–3150. Mobile Radio Commun. (PIMRC), Washington, DC, USA, Sep. 2014,
[149] J. Mitola and G. Q. Maguire, “Cognitive radio: Making software pp. 748–752.
radios more personal,” IEEE Pers. Commun., vol. 6, no. 4, pp. 13–18, [173] A. El Shafie, N. Al-Dhahir, and R. Hamila, “Cooperative access
Aug. 1999. schemes for efficient SWIPT transmissions in cognitive radio
[150] A. Goldsmith, S. A. Jafar, I. Maric, and S. Srinivasa, “Breaking networks,” in Proc. IEEE Globecom Workshops (GC Wkshps),
spectrum gridlock with cognitive radios: An information theoretic San Diego, CA, USA, 2015, pp. 1–6.
perspective,” Proc. IEEE, vol. 97, no. 5, pp. 894–914, May 2009. [174] A. El Shafie, N. Al-Dhahir, and R. Hamila, “A sparsity-aware coop-
[151] P. Kolodzy et al., “Next generation communications: Kickoff meeting,” erative protocol for cognitive radio networks with energy-harvesting
in Proc. DARPA, 2001. primary user,” IEEE Trans. Commun., vol. 63, no. 9, pp. 3118–3131,
[152] I. F. Akyildiz, W.-Y. Lee, M. C. Vuran, and S. Mohanty, “A survey on Sep. 2015.
spectrum management in cognitive radio networks,” IEEE Commun. [175] K. Janghel and S. Prakriya, “Outage performance of dynamic spectrum
Mag., vol. 46, no. 4, pp. 40–48, Apr. 2008. access systems with energy harvesting transmitters,” in Proc. IEEE
[153] F. Akhtar, M. H. Rehmani, and M. Reisslein, “White space: Definitional 25th Int. Symp. Pers. Indoor Mobile Radio Commun., Washington, DC,
perspectives and their role in exploiting spectrum opportunities,” USA, 2014, pp. 648–652.
Telecommun. Policy, vol. 40, no. 4, pp. 319–331, 2016. [176] G. Zheng, Z. K. M. Ho, E. A. Jorswieck, and B. E. Ottersten,
[154] S. Haykin, “Cognitive radio: Brain-empowered wireless communica- “Information and energy cooperation in cognitive radio networks,”
tions,” IEEE J. Sel. Areas Commun., vol. 23, no. 2, pp. 201–220, IEEE Trans. Signal Process., vol. 62, no. 9, pp. 2290–2303,
Feb. 2005. May 2014.
ALSABA et al.: BEAMFORMING IN WIRELESS EH COMMUNICATIONS SYSTEMS: SURVEY 1359

[177] S. Hu, Z. Ding, Q. Ni, and Y. Yuan, “Beamforming optimization [200] Z. Ding, F. Adachi, and H. V. Poor, “The application of MIMO to non-
for full-duplex cooperative cognitive radio networks,” in Proc. IEEE orthogonal multiple access,” IEEE Trans. Wireless Commun., vol. 15,
Workshop Signal Process. Adv. Wireless Commun. (SPAWC), vol. 2016. no. 1, pp. 537–552, Jan. 2016.
Edinburgh, U.K., 2016, pp. 1–5. [201] K. Higuchi and A. Benjebbour, “Non-orthogonal multiple access
[178] C. Xu, Q. Zhang, Q. Li, Y. Tan, and J. Qin, “Robust transceiver (NOMA) with successive interference cancellation for future radio
design for wireless information and power transmission in underlay access,” IEICE Trans. Commun., vol. 98, no. 3, pp. 403–414, 2015.
MIMO cognitive radio networks,” IEEE Commun. Lett., vol. 18, no. 9, [202] B. Kim et al., “Non-orthogonal multiple access in a downlink multiuser
pp. 1665–1668, Sep. 2014. beamforming system,” in Proc. IEEE Mil. Commun. Conf. (MILCOM),
[179] Z. Shu, Y. Qian, and S. Ci, “On physical layer security for cognitive San Diego, CA, USA, 2013, pp. 1278–1283.
radio networks,” IEEE Netw., vol. 27, no. 3, pp. 28–33, May/Jun. 2013. [203] N. T. Do, D. B. Da Costa, T. Q. Duong, and B. An, “A BNBF user
[180] L. Jiang, H. Tian, C. Qin, S. Gjessing, and Y. Zhang, “Secure beam- selection scheme for NOMA-based cooperative relaying systems with
forming in wireless-powered cooperative cognitive radio networks,” SWIPT,” IEEE Commun. Lett., vol. 21, no. 3, pp. 664–667, Mar. 2017.
IEEE Commun. Lett., vol. 20, no. 3, pp. 522–525, Mar. 2016. [204] R. Sun, Y. Wang, X. Wang, and Y. Zhang, “Transceiver design for coop-
[181] Y. Wu and X. Chen, “Robust beamforming and power splitting for erative non-orthogonal multiple access systems with wireless energy
secrecy wireless information and power transfer in cognitive relay transfer,” IET Commun., vol. 10, no. 15, pp. 1947–1955, Oct. 2016.
networks,” IEEE Commun. Lett., vol. 20, no. 6, pp. 1152–1155, [205] Y. Liu, Z. Ding, M. Elkashlan, and H. V. Poor, “Cooperative non-
Jun. 2016. orthogonal multiple access with simultaneous wireless information
[182] D. W. K. Ng, E. S. Lo, and R. Schober, “Multiobjective resource and power transfer,” IEEE J. Sel. Areas Commun., vol. 34, no. 4,
allocation for secure communication in cognitive radio networks with pp. 938–953, Apr. 2016.
wireless information and power transfer,” IEEE Trans. Veh. Technol., [206] P. D. Diamantoulakis, K. N. Pappi, Z. Ding, and G. K. Karagiannidis,
vol. 65, no. 5, pp. 3166–3184, May 2016. “Wireless-powered communications with non-orthogonal multiple
access,” IEEE Trans. Wireless Commun., vol. 15, no. 12,
[183] F. Zhou, Z. Li, J. Cheng, Q. Li, and J. Si, “Robust AN-aided
pp. 8422–8436, Dec. 2016.
beamforming and power splitting design for secure MISO cognitive
[207] M. Ashraf, A. Shahid, J. W. Jang, and K.-G. Lee, “Energy harvesting
radio with SWIPT,” IEEE Trans. Wireless Commun., vol. 16, no. 4,
non-orthogonal multiple access system with multi-antenna relay and
pp. 2450–2464, Apr. 2017.
base station,” IEEE Access, vol. 5, pp. 17660–17670, 2017.
[184] H. Lei et al., “On secure underlay MIMO cognitive radio networks [208] P. Kamalinejad et al., “Wireless energy harvesting for the Internet of
with energy harvesting and transmit antenna selection,” IEEE Trans. Things,” IEEE Commun. Mag., vol. 53, no. 6, pp. 102–108, Jun. 2015.
Green Commun. Netw., vol. 1, no. 2, pp. 192–203, Jun. 2017. [209] S. Vassilaras and G. C. Alexandropoulos, “Cooperative beamforming
[185] G. Yang, C. K. Ho, and Y. L. Guan, “Dynamic resource allocation for techniques for energy efficient IoT wireless communication,” in Proc.
multiple-antenna wireless power transfer,” IEEE Trans. Signal Process., IEEE Int. Conf. Commun. (ICC), Paris, France, 2017, pp. 1–6.
vol. 62, no. 14, pp. 3565–3577, 2014. [210] L. Mohjazi, M. Dianati, G. K. Karagiannidis, S. Muhaidat, and
[186] Y. Zeng and R. Zhang, “Optimized training design for wireless M. Al-Qutayri, “RF-powered cognitive radio networks: Technical
energy transfer,” IEEE Trans. Commun., vol. 63, no. 2, pp. 536–550, challenges and limitations,” IEEE Commun. Mag., vol. 53, no. 4,
Feb. 2015. pp. 94–100, Apr. 2015.
[187] Y. Zeng and R. Zhang, “Optimized training for net energy [211] J. H. Winters, “Smart antenna techniques and their application to
maximization in multi-antenna wireless energy transfer over frequency- wireless ad hoc networks,” IEEE Wireless Commun., vol. 13, no. 4,
selective channel,” IEEE Trans. Commun., vol. 63, no. 6, pp. 77–83, Aug. 2006.
pp. 2360–2373, Jun. 2015. [212] S. Kompella, G. D. Nguyen, C. Kam, J. E. Wieselthier, and
[188] R. Gangula, D. Gesbert, and D. Gündüz, “Optimization of energy har- A. Ephremides, “Cooperation in cognitive underlay networks: Stable
vesting MISO communication system with feedback,” IEEE J. Sel. throughput tradeoffs,” IEEE/ACM Trans. Netw., vol. 22, no. 6,
Areas Commun., vol. 33, no. 3, pp. 396–406, Mar. 2015. pp. 1756–1768, Dec. 2014.
[189] H. Lee, S.-R. Lee, K.-J. Lee, H.-B. Kong, and I. Lee, “Optimal beam- [213] H. Chen, Y. Li, J. L. Rebelatto, B. F. Uchôa-Filho, and B. Vucetic,
forming designs for wireless information and power transfer in MISO “Harvest-then-cooperate: Wireless-powered cooperative communica-
interference channels,” IEEE Trans. Wireless Commun., vol. 14, no. 9, tions,” IEEE Trans. Signal Process., vol. 63, no. 7, pp. 1700–1711,
pp. 4810–4821, Sep. 2015. Apr. 2015.
[190] J. Park and B. Clerckx, “Joint wireless information and energy transfer [214] J. Zou, H. Xiong, D. Wang, and C. W. Chen, “Optimal power allo-
with reduced feedback in MIMO interference channels,” IEEE J. Sel. cation for hybrid overlay/underlay spectrum sharing in multiband
Areas Commun., vol. 33, no. 8, pp. 1563–1577, Aug. 2015. cognitive radio networks,” IEEE Trans. Veh. Technol., vol. 62, no. 4,
[191] J. Xu and R. Zhang, “Energy beamforming with one-bit feed- pp. 1827–1837, May 2013.
back,” IEEE Trans. Signal Process., vol. 62, no. 20, pp. 5370–5381, [215] M. Amjad, F. Akhtar, M. H. Rehmani, M. Reisslein, and T. Umer,
Oct. 2014. “Full-duplex communication in cognitive radio networks: A survey,”
[192] T. D. P. Perera, D. N. K. Jayakody, S. K. Sharma, S. Chatzinotas, and IEEE Commun. Surveys Tuts., vol. 19, no. 4, pp. 2158–2191, 4th
J. Li, “Simultaneous wireless information and power transfer (SWIPT): Quart., 2017.
Recent advances and future challenges,” IEEE Commun. Surveys Tuts., [216] T. Zhang et al., “Secure transmission in cognitive wiretap networks,”
to be published. in Proc. IEEE 83rd Veh. Technol. Conf. (VTC Spring), Nanjing, China,
[193] O. Ozel and S. Ulukus, “Achieving AWGN capacity under stochas- 2016, pp. 1–5.
tic energy harvesting,” IEEE Trans. Inf. Theory, vol. 58, no. 10, [217] D. Kim, H. Lee, and D. Hong, “A survey of in-band full-duplex
pp. 6471–6483, Oct. 2012. transmission: From the perspective of PHY and MAC layers,” IEEE
[194] T. Li, Y. Dong, P. Fan, and K. B. Letaief, “Wireless communications Commun. Surveys Tuts., vol. 17, no. 4, pp. 2017–2046, 4th Quart.,
with RF-based energy harvesting: From information theory to green 2015.
systems,” IEEE Access, vol. 5, pp. 27538–27550, 2017.
[195] J. Rinne, J. Keskinen, P. R. Berger, D. Lupo, and M. Valkama,
“Feasibility and fundamental limits of energy-harvesting based M2M
communications,” Int. J. Wireless Inf. Netw., vol. 24, no. 3,
pp. 291–299, 2017.
[196] O. Ozel, K. Tutuncuoglu, S. Ulukus, and A. Yener, “Fundamental limits Yamen Alsaba received the B.Eng. degree in
of energy harvesting communications,” IEEE Commun. Mag., vol. 53, electrical and telecommunication engineering from
no. 4, pp. 126–132, Apr. 2015. the Higher Institute of Applied Science and
[197] M. Gorlatova, A. Wallwater, and G. Zussman, “Networking low-power Technology, Damascus, Syria, in 2005 and the
energy harvesting devices: Measurements and algorithms,” IEEE Trans. M.S. degree from Supelec, Paris, France, in 2010.
Mobile Comput., vol. 12, no. 9, pp. 1853–1865, Sep. 2013. He is currently pursuing the Ph.D. degree with
[198] A. Kansal, J. Hsu, S. Zahedi, and M. B. Srivastava, “Power manage- the Wireless Communication Center, Universiti
ment in energy harvesting sensor networks,” ACM Trans. Embedded Teknologi Malaysia, Johor Bahru, Malaysia. His
Comput. Syst., vol. 6, no. 4, p. 32, 2007. research interests include simultaneous wireless
[199] J. Choi, “Minimum power multicast beamforming with superposition information and power transfer, nonorthogonal
coding for multiresolution broadcast and application to noma systems,” multiple access, beamforming, physical layer secu-
IEEE Trans. Commun., vol. 63, no. 3, pp. 791–800, Mar. 2015. rity, large scale network analysis, and 5G.
1360 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 20, NO. 2, SECOND QUARTER 2018

Sharul Kamal Abdul Rahim received the degree Chee Yen Leow (S’08–M’12) received the B.Eng.
in electrical engineering from the University of degree in computer engineering from Universiti
Tennessee, USA, in 1996, the M.Sc. degree in Teknologi Malaysia (UTM), Johor Bahru, Malaysia,
engineering (communication engineering) from the in 2007 and the Ph.D. degree from Imperial College
Universiti Teknologi Malaysia (UTM) in 2001, London, U.K., in 2011. Since 2007, he has been
and the Ph.D. degree in wireless communication with the Faculty of Electrical Engineering, UTM. He
systems from the University of Birmingham, U.K., is currently a Senior Lecturer with the Faculty and
in 2007. He is currently an Associate Professor a member of the Wireless Communication Center,
with the Wireless Communication Centre, Faculty Higher Institution Centre of Excellence, UTM. His
of Electrical Engineering, UTM Skudai. He has research interests include cooperative communica-
published over 50 journal papers and technical pro- tion, MIMO, UAV communication, physical layer
ceedings on rain attenuations, smart antenna system, microwave design, and security, convex optimization, communications theory, wireless power transfer
reconfigurable antenna in national and international journals and conferences. and communication, nonorthogonal multiple access, and 5G.
His research interest is smart antenna on communication systems. He is also
a member of IEEE Malaysia Section, the Board of Engineer Malaysia, the
Institute of Engineer Malaysia, and Eta Kappa Nu Chapter, International
Electrical Engineering Honour Society, University of Tennessee.

Anda mungkin juga menyukai