Вы находитесь на странице: 1из 14

CAN CELL PHONE TOWER GIVE CHARGING TO

YOUR MOBILE ?
J.JENIFER RIZANI , S. RASMA , A.REMIYA
SECOND YEAR CSE
DMI COLLEGE OF ENGINEERING
NAGERCOIL

Abstract— This paper investigates the optimal resource allo- I. I NTRODUCTION


cation in wireless powered communication network with user
cooperation, where two single-antenna users first harvest energy
from the signals transmitted by a multi-antenna hybrid access
point (H-AP) and then cooperatively send information to the
H-AP using their harvested energy. To explore the system infor-
R ECENTLY, radio-frequency (RF) energy harvesting
(EH) has attracted much interest due to its capabil-
ity of providing reliable and perpetual energy sources for
mation transmission performance limit, an optimization problem energy-constrained wireless networks, including Internet of
is formulated to maximize the weighted sum-rate (WSR) by Things (IoT) and wireless sensor networks (WSNs) [1]–[7].
jointly optimizing energy beamforming vector, time assignment, It was reported that 27μW of wireless power can be harvested
and power allocation. Besides, another optimization problem is
also formulated to minimize the total transmission time for given from RF signals at distances of 10 meters over 915 MHz [8].
amount of data required to be transmitted at the two sources. Since RF signals simultaneously carry energy and
Because both problems are non-convex, we first transform them information, simultaneous wireless information and power
to be convex by using proper variable substitutions and then transfer (SWIPT) was proposed as a promising solution to
apply semi-definite relaxation to solve them. We theoretically energy-constrained networks [9]–[11]. In [9] and [10], ideal
prove that our proposed methods guarantee the global optimum
of both problems. Simulation results show that system WSR and SWIPT receiver was assumed to harvest energy and receive
transmission time can be significantly enhanced by using energy information from the same signals. In order to make SWIPT
beamforming and user cooperation. It is observed that when the workable, several practical receiver architectures were pro-
total amount of information of two users is fixed, with the increase posed in [11]. So far, SWIPT has been widely studied in
of the information amount of the user relatively farther away various wireless systems, see e.g. [12]–[25].
from the H-AP, the transmission time of the user cooperation
scheme decreases while that of the direct transmission increases. Recently, another important application of RF EH, i.e., wire-
Besides, the effects of user position on the system performances less powered communication network (WPCN), was pre-
are also discussed, which provides some useful insights. sented [4], where some wireless nodes harvest energy and
Index Terms— Wireless powered communication net- use the harvested RF energy to transmit/decode information
works, energy beamforming, time assignment, semidefinite to/for other nodes. In WPCNs, RF energy sources can be either
programming. ambient RF sources (e.g., TV towers, WI-FI signals) or dedi-
cated RF energy transmitters (e.g., power stations). Compared
with ambient RF sources, dedicated ones are controllable,
which are able to provide more stable energy supply for
EH communication devices. Besides, more energy can be
harvested at wireless devices, because the dedicated power
station is only responsible for energy transfer in WPCNs.
Up to now, a lot of works on WPCNs with dedi-
cated power station can be found in the literature, see
e.g. [26]–[28], [31]–[37]. In [26]–[28], multi-user systems
were studied and the total throughput of all users were max-
imized. Specifically, in [26], a single-antenna hybrid access
point (H-AP)-assisted WPCN was investigated, where the
H-AP was with steady energy supply and it first transferred
energy to a group of users and then the users transmitted
information to the H-AP by using the harvested energy. In [27],
a WPCN with heterogeneous users was considered, where
some users had to harvest energy from H-AP and the others
were with sufficient energy supply. In [28], the multi-carrier
full-duplex OFDM WPCN was studied. As is known, in WSNs
and IoT systems, it is sometimes not available to equip multi-
ple antennas at small-size nodes, in this case, user cooperation
can be employed to exploit the cooperative diversity to throughput, sum-rate, WSR or rate fairness were optimized,
enhance system information transmission efficiency [29], [30]. in our work, besides the WSR-maximization design, the total
In [31], system throughput was maximized for a three-node transmission time minimization design of the system is also
relaying network, where a source and a relay harvest energy investigated to meet the low-power design requirement of
from a dedicated multi-antenna power beacon and then trans- WSNs and green communication design requirement of future
mit information to a destination via cooperative relaying mode. communication systems [39].
In [32] and [33], outage performance was analyzed for a three- Our contributions can be summarized as follows.
node WPCN, where one source and one relay first harvested Firstly, in order to explore the system information trans-
energy from a H-AP and then the source transmitted informa- mission performance limit, we also formulate an optimization
tion via the relay over a two-hop relaying channel to H-AP. problem to maximize the WSR of the system by jointly
In their work, the relay node itself had no own information to optimizing energy beamforming and time assignment.
transmit. In [34], user cooperation was introduced into H-AP- Secondly, with given amount of data required to be delivered
assisted WPCNs and the system WSR was maximized by at the two users, we formulate an optimization problem to
jointly optimizing the time assignment and power allocation, minimize the total transmission time of the system by jointly
where two users cooperatively transmitted information to a optimizing energy beamforming and time assignment.
common H-AP by using their harvested energy from the H-AP. Thirdly, as the both problems are neither convex nor con-
In [35], the max-min rate fairness was studied for two-user cave, we first transform them to be convex by using variable
WPCNs, where two users were firstly wireless powered by substitution operations and then applying the semidefinite
a power station and then they helped each other to transmit relaxation (SDR) method to solve them efficiently. We also
information to a common destination. However, in these works theoretically prove that with our proposed method, the global
mentioned above, only single antenna was employed at the optimum of the solutions associated with the two problems
H-AP, so no multi-antenna technology was involved. can be guaranteed.
As energy beamforming technique is able to focus the Fourthly, simulation results are provided, which show that
transmitted energy on a specific receiver and also combat the system performances of WSR and the transmission time are
small-scale channel fading, which improves the energy trans- significantly enhanced by using energy beamforming and user
mission efficiency, some works began to investigate energy cooperation. It is also observed that when total amount of
beamforming in WPCNs by equipping multiple antennas on information of two users is fixed, with the increase of the
the energy tower or H-AP. In [36], energy beamforming was information amount of the user relatively farther away from the
optimally designed for a point-to-point WPCN, in which a H-AP, the transmission time of the user cooperation scheme
multi-antenna H-AP emits energy to a receiver to drive the decreases while that of the direct transmission increases.
information transmission of the receiver. In [37], joint energy Besides, the effects of user position on the system performance
beamforming design and time assignment was investigated to are also demonstrated, which provides insights on the WPCN.
maximize the sum rate of multiple users in a WPCN, where Finally, we also extend our scheme to multiple user pair
the users harvest energy from a dedicated power station and scenarios, it shows that our proposed solution method is still
then use the harvested energy to transmit information to their valid and employing our proposed joint energy beamforming
common destination. In [38], a WPCN with group cooperation and user cooperation may bring more system performance gain
was considered, where two communication groups cooperate in multiple user pair scenarios.
with each other via wireless power transfer and time sharing Note that although the three-node WPCN topology seems
to fulfill their expected information delivering. a little bit simple, the joint energy beamforming and time
In this paper, we also focus on the H-AP-assisted WPCN, assignment with user cooperation is not trivial, since the
in which a H-AP broadcasts energy to two distributed users beamforming vector and the time assignment vector are cou-
in the downlink (DL) and the users transmit their indepen- pled together, which results in the difficulty to solve the
dent information using their individually harvested energy problems. Nevertheless, we present efficient solution meth-
to the H-AP in the uplink (UL) through time-division- ods which are capable of guaranteeing the global optimum.
multiple-access (TDMA). Different from existing works, For more H-APs or more user scenarios, the corresponding
e.g., [26]–[28], [36], [37], where no user cooperation was transmission strategies, optimization problems and algorithms
involved, the user cooperation between the two users is con- shall be reconsidered and redesigned, which is beyond the
sidered in our work, where the user relatively near to the H-AP scope of this paper. As alternative, the three-node group
is employed as a relay node to help forward the information decomposition of a multi-user or multi-relay system may be
transmission of the user relatively far away from the H-AP an easy way to get system gains apparently. Actually, our
by consuming a part of its allocated UL time and DL har- considered H-AP-assisted WPCN model can be applied to
vested energy. Although in [31]–[35] the user cooperation was various communication networks, including WSNs and IoT.
considered, only single antenna was equipped at the H-AP, For example, in WSNs, sensor nodes are deployed to acquire
so that no energy beamforming was involved in [31]–[35]. environmental data such as temperature or humidity. Their
Comparably, multiple antennas are equipped at the H-AP and available energy is often very limited due to the embedded
the energy beamforming is jointly designed and optimized small-capacity battery. Thus, to ensure the network availability,
with the time assignment in our work. In addition, different sensor nodes are required to be powered with steady energy.
from all existing works mentioned above, where only system As the sink node in WSNs is generally with fixed sufficient
Fig. 2. Framework of the transmission protocol.

Fig. 1. System model.


to Ui . z i is the additive white Gaussian noise at Ui , in which
each element is with zero mean and variance N0 and ω is an
energy source, which also collects data from the sensors, and N × 1 vector with
thus it may be treated as a H-AP. Moreover, sink node is often
with much bigger size than sensors, so it may be equipped ω2 ≤ 1, (1)
with multiple antennas. Therefore, energy beamforming is able
to be employed to enhance the energy transfer efficiency. denoting the energy beamforming vector, which is calculated
Furthermore, in practical WSNs, sensor nodes are generally and employed at U0 . Therefore, the received energy at the Ui
distributed with different distances from the sink node, so user in the ET phase can be given by
cooperation among sensors can be utilized to enhance the E i = τ0 η|hiH ω|2 P0 , ∀i = 1, 2,
system performance.
The rest of this paper is organized as follows. Section II where η ∈ (0, 1) represents the energy conversion efficiency.
describes the system model. Section III formulates an opti- In the IT phase, U1 and U2 transmit their information to U0
mization problem to maximize the WSR of two users and using only their harvested energy in the ET phase. In general,
provides the solution method. Section IV formulates an opti- U1 and U2 have different distances from U0 . Without loss
mization problem to minimize the total transmission time of generality, we assume that U2 is closer to U0 than U1 .
of two users and present the solution method. Section V In this case, U2 acts as a relay node to help the information
extends our work to multiple user pair scenarios. In Section VI, transmission from U1 to U0 . Also, U2 has its own information
simulation results are provided. Finally, Section VII concludes to be delivered to U0 , which is transferred over the U2 → U0
this paper. direct link. To complete the information transmission from
U1 and U2 to U0 , the IT phase is further divided into three
II. S YSTEM M ODEL durations with time length τ1 , τ2(1) and τ2(2) , respectively.
Consider a wireless network consisting of a hybrid access Specifically, in the first duration of τ1 , U1 broadcasts infor-
point (H-AP), referred to as U0 , and two users denoted by mation to U0 and U2 . The received signal at U0 and U2 from
U1 and U2 , as shown in Figure 1. U0 is equipped with N U1 can be respectively given by
antennas and U1 and U2 are with single antenna. U1 and U2 
y0(1) = P1 h1 s1 + z0 ,
desire to transmit information to U0 who has fixed energy
supply. U1 and U2 are energy-constrained nodes that need and
to harvest energy from the RF signals transmitted by U0 in 
order to support their information transmission. Quasi-static y2(1) = P1 gs1 + z 2 ,
fading channel are assumed, so that at each fading state all where s1 is the information symbol of U1 with unit power
channel coefficients are considered to be constant. Moreover, and g is the complex channel coefficient between U1 and U2 .
the channel reciprocity between any two nodes is assumed. z0 represents the additive white Gaussian noise vector at U0 ,
That is, at each fading state, the channel coefficient from node in which each element is with zero mean and variance N0 , and
u to v is the same with that from v to u. P1 represents the transmit power of U1 . As the total consumed
Since U1 and U2 have no power to initiate the information energy at U1 must be smaller than its harvested energy E 1 ,
transmission, the transmission period is divided into two P1 must satisfy that τ1 P1 ≤ E 1 , i.e.,
phases, i.e., the DL energy transfer (ET) phase and the UL
information transmission (IT) phase, as shown in Figure 2. τ1 P1 ≤ η|h1H ω|2 P0 τ0 . (2)
In the ET phase with time length of τ0 , U0 transfers
energy to both U1 and U2 through RF transmission, as shown Decode-and-forward (DF) relaying operation is employed
in Figure 1 (a). The received signal at Ui (i = 1, 2) from U0 at U2 , so that in the second duration of τ2(1), U2 decodes the
is given by information received from U1 and then forwards the decoded
√ information to U0 .1 Thus, the received signal at U0 from U2
yi,0 = P 0 hiH ω + z i , ∀i = 1, 2,
1 Similar to many existing works see e.g. [32]–[34], we ignore the energy
where P0 denotes the transmit power at U0 and hi denotes the consumption associated with decoding and receiving at U2 , since it is
complex channel vectors of size N × 1 for the links from U0 negligible compared with transmission power.
associated with the data of U1 is into four parts, i.e., T = τ0 + τ1 + τ2(1) + τ2(2) . Let τ =
 (1) (2)
[τ0 , τ1 , τ2 , τ2 ], which can be regarded as the time fraction
(1) (1) (1)
y0,2 = P2 h2 s2 + z0 , assignment vector. Without loss of generality, T is normalized
to 1 in this section. As a result, it satisfies that
where P2(1) denotes the power consumed at U2 to assist the

(1)
information transmission of U1 and s2 denotes the transmit τ0 + τ1 + τ2(1) + τ2(2) ≤ 1,
(1) (2) (7)
signal in this duration. According to [42], the achievable end- τ0 , τ1 , τ2 , τ2 ≥ 0.
to-end information rate (Nats/s/Hz) from U1 to U0 over the
DF relaying channel can be given by Our goal is to maximize the system WSR by jointly
  optimizing energy beamforming design, time assignment and
(12) (10)
R1 = min R1 , R1 , (3) power allocation. Thus, the problem can be mathematically
expressed as
(12)
where R1 is the information rate over the U1 → U2 link,
which is P1 : max α1 R1 + α2 R2
{ω,τ ,P}
 
(12) |g|2 P1 s.t. (1), (2), (6), (7), (8)
R1 = τ1 C
N0 where α1 and α2 respectively denote the given nonnegative
(10) rate weights for U1 and U2 to reflect the users’ fairness, and
and R1 is the sum information rate over the U1 → U2 and
U2 → U 1 links, which is P = [P1 , P2(1) , P2(2) ] represents the power allocation vector.
   From (2), (3), (4) , (5) and (6), it can be seen that ω, τ and
(10) h1 2 P1 (1) h2 2 P2(1) P are multiplicatively coupled with each other, which makes
R 1 = τ1 C + τ2 C , (4)
N0 N0 Problem P1 neither convex nor concave. Thus, Problem P1
cannot be easily solved by using known methods. Therefore,
where C (x)  ln(1 + x). we propose an efficient method to solve it as follows.
In the third duration of τ2(2) , U2 transmits its own informa- Following (3), we have that
tion to U0 . The received signal at U0 from U2 associated with  2 
the data of U2 can be given by |g| P1
R 1 ≤ τ1 C (9)
 N0
(2) (2) (2)
y0,2 = P2 h2 s2 + z0 , and
   (1)

where P2(2)denotes power consumed at U2 to transmit its h1 2 P1 (1) h2 2 P2
(2) R 1 ≤ τ1 C + τ2 C . (10)
own information and s2 denotes its own information symbol. N0 N0
Accordingly, the achievable information rate from U2 to U0
is given by By treating R1 as an optimization variable, Problem P1 can
 be equivalently transformed into
(2) h2 2 P2(2)
R 2 = τ2 C , (5) P1-A : max α1 R1 + α2 R2
N0 {R1 ,ω,τ ,P}
s.t. (1), (2), (6), (7), (9), (10) (11)
Moreover, since the total consumed energy at U2 must be
smaller than its harvested energy E 2 , we have that Problem P1-A is also non-convex, as ω, τ and P are
(1) (1) (2) (2) still multiplicatively couples with each other. To solve it,
τ2 P2 + τ2 P2 ≤ η|h2H ω|2 P0 τ0 . (6) (1) (2)
we introduce an auxiliary vector t = [t1 , t2 , t2 ], where
(1) (1) (1) (2) (2) (2)
For the transmission protocol with user cooperation t1 = τ1 P1 , t2 = τ2 P2 and t2 = τ2 P2 . Then, we have
described above, we shall explore its maximum WSR and that
minimum total transmission time in Section III and Section IV, ⎧ t1

⎪ P1 = ,
respectively. ⎪
⎪ τ

⎪ 1

⎨ (1) t (1)
III. W EIGHTED S UM -R ATE M AXIMIZATION D ESIGN P2 = 2(1) , (12)

⎪ τ2
Based on the system model described in section II, in this ⎪
⎪ (2)

⎪ (2) t2
section, we shall discuss how to jointly optimize the energy ⎪
⎩ P2 = (2) .
beamforming and time assignment to maximize the WSR of τ2
the system. We first formulate a WSR maximization problem. (1) (2) t1 t2
(1)

Then, we transform the problem into some equivalent ones by By substituting P1 , P2 and P2 with τ1 , (1) and
τ2
using proper variable substitutions. After that, we use the SDR (2)
t2
method to solve it efficiently. Finally, we theoretically prove (2) , respectively, in terms of (12), we can respectively
τ2
that by using our proposed method, the global optimum of the transform (5) into
obtained solution can be guaranteed.  (2)

(2) h2 2 t2
In terms of the transmission process described in Section II, R2 = τ2 C (2)
, (13)
one can see that the total transmission period T is divided N0 τ2
(9) into Fortunately, we also found that the rank-one optimal solu-
  tion V ∗ of Problem P1-C always exists, which means that by
|g|2 t
R 1 ≤ τ1 C N0 τ1
1
, (14) using our solution method mentioned above, the global optimal
solution of P1 can always be obtained. To prove this, we first
and (10) into
  introduce the following lemma [41].
  (1)
Lemma 1: ([42, Th. 3.2]) The following optimization
h1 2 t1 (1) h2 2 t2
R 1 ≤ τ1 C N0 τ1 + τ2 C (1) . (15)
N0 τ2 problem
L
Moreover, by defining V  τ0 ωω H , constraint (1) can be min Tr(C l X l )
X 1 ,...,X L l=1
equivalently rewritten to be
⎧ s.t. X l 0, l = 1, . . . , L,
⎪ L
⎨Tr(V) ≤ τ0 , Tr( Aml X l ) ≥ bm , m = 1, . . . , M.
l=1
V 0, (16)

⎩ always has an optimal solution (X 1∗ , . . . , X ∗L ) such that
rank(V) = 1,
L
constraints (2) and (6) can be respectively transformed into rank2 (X l∗ ) ≤ M,
l=1

t1 ≤ η P0 Tr(h1 h1H V), (17) where Aml , X l , C l are K × K Hermitian matrices and bm is
a real number.
and The detail information associated with how to generate a
(1) (2) rank-constrained solution can be referenced to [41]. Based on
t2 + t2 ≤ η P0 Tr(h2 h2H V). (18)
Lemma 1, we present the Theorem 2 as follows.
By doing so, Problem P1-A is equivalently transformed into Theorem 2: The rank-one optimal solution V ∗ of Prob-
lem P1-C always exists.
P1-B : max α1 R1 + α2 R2 Proof: Consider the following optimization problem
{R1 ,V,t,τ }
s.t. (7), (14), (15), (16), (17), (18). (19) P2 : min Tr(V)
V
Due to the rank-one constraint in (16), Problem P1-B must s.t. t1∗ ≤ η P0 Tr(h1 h1H V),
be non-convex, so we adopt the semidefinite relaxation (SDR) (1)∗ (2)∗
t2 + t2 ≤ η P0 Tr(h2 h2H V), V 0.
technique to deal with it. By dropping the rank-one constraint,
we obtain the relaxed problem of Problem P1-B as Suppose the optimal solution of Problem P2 is V . We first
prove that V is feasible for Problem P1-C as follows.
P1-C : max α1 R1 + α2 R2 (20) Since the number of constraints of Problem P1-C is more
{R1 ,V,t,τ }
s.t. Tr(V) ≤ τ0 , V 0, than Problem P2, i.e., the constraints of Problem P1-C is
tighter than those of Problem P2, a feasible solution of Prob-
(7), (14), (15), (17), (18). (21) lem P1-C must be feasible for Problem P2. Thus, Tr(V ) ≤
Fortunately, we find that Problem P1-C is a convex problem, Tr(V∗ ) ≤ τ0∗ , which indicates that V is feasible for
which is emphasized and proved in the following theorem. Problem P1-C. Moreover, since the objective function of
Theorem 1: Problem P1-C is a convex problem. Problem P1-C is only related to t, τ and R1 , it can be
Proof: The constraints (7), (17), (18) and (21) are linear. inferred that (V , t∗ , τ ∗ , R1∗ ) is also the optimal solution of
The right part of the constraint (14)is the perspective function Problem P1-C. According to Lemma 1, there exists anoptimal
 2
of concave functions ln 1 + |g|N0t1 and thus (14) is convex.
2
solution V for the Problem P2 such that rank(V ) ≤ 2.
 
Therefore, the optimal V satisfies V = 0 and thus
The constraint (15) has similar form as (14), and thus it can
rank(V ) = 1. Hence, Theorem 2 is proved. 
also be proved to be convex. From (13), the objective function
Corollary 1: The global optimal solution to Problem P1 is
is the 
non-negative  weighted sum of R1 and concave function
(2)
h2 2 t2
guaranteed by using our proposed solution method.
τ22 ln 1 + N0 τ22 , so the objective function is concave. Proof: Problem P1, Problem P1-A and Problem P1-B
Thus, Theorem 1 is proved.  are equivalent to each other. It is known that once the optimal
Following Theorem, it is known that the global optimal solution of Problem P1-C satisfies the rank-one constraint,
solution (V∗ , t∗ , τ ∗ , R1∗ ) of Problem P1-C can be obtained by it is equivalent to Problem P1-B. Theorem 2 proves that
using CVX tools [40]. With the obtained t∗ , τ ∗ , the optimal P∗ Problem P1-C always has a rank-one optimal solution and
can be derived according to (12). Theorem 1 indicates that the global optimal solution of
Note that our goal is to find the optimal beamforming Problem P1-C can be found. Therefore, the optimal solution
vector ω∗ rather than V∗ . Only when the optimal V∗ of for Problem P1 can always be found by using our proposed
Problem P1-C follows rank-one constraint, the ω∗ calculated solution method. 
from V∗ is the optimal solution of Problem P1-B. Particularly,
in this case, the optimal beamforming vector ω∗ for the IV. T RANSMISSION T IME M INIMIZATION D ESIGN
primary problem P1 can be extracted by eigenvalue decom- In Section III, we explore the potential information
position of V∗ /τ ∗ . transmission capacity of the system by considering the
WSR-maximization problem. In practical applications, it is Similar to Problem P1-A, by dropping the rank-one con-
often of great to deliver all data for the users with as little time straint in (16), Problem P3-A is transformed into
as possible, especially for sensing applications. Actually, small
P3-B : min τ0 + τ1 + τ2(1) + τ2(2)
transmission time is also very important for user experience {V,t,τ }
and real-time applications. Therefore, in this section, we shall s.t. Tr(V) ≤ τ0 , V 0
discuss the transmission time minimization problem by jointly (17), (18), (24), (25), (26). (27)
optimizing the energy beamforming and time assignment.
Denote the amount of information required to be delivered Theorem 3: Problem P3-B is a convex problem.
from Ui to U0 to be Bi nats, where i = 1, 2. Without loss Proof: Constraints (17), (18), (24), (27) and the objec-
of generality, the system bandwidth is normalized to be 1Hz. tive function are linear. Moreover, the left side of the first
To guarantee the successful delivery of all information of U1 term of (25) is a perspective function of the concave func-
tion C |g|N2t1 , so the first term of constraint (25) must be
2
to U0 over the U1 − U2 − U0 three node DF cooperative relay
link, we need to ensure that R1 ≥ B1 . Combining it with (3), convex. Similarly, the second term of constraint (25) and
we have that
constraint (26) can also be proved to be convex. Therefore,
⎧  2 
⎪ |g| P1 Theorem 3 is proved. 

⎨ τ1 C ≥ B1 , As a result, the global optimal solution (V∗ , t∗ , τ ∗ ) of P3-B
 N0 2    (22)

⎪ h1  P1 (1)
(1)
h2 2 P2 can be obtained by interior point solver, e.g., CVX [40].
⎩ τ1 C + τ2 C N0 ≥ B 1 . In terms of the definition of t, τ , with the obtained t∗ , τ ∗ ,
N0
the optimal P∗ can be derived in terms of (12). Similar
Meanwhile, to guarantee the successful delivery of all infor- to Problem P1, to guarantee the global optimal solution of
mation from U2 to the U0 over the U2 → U0 direct link, it is Problem P3, the remaining question is to determine that
required that R2 ≥ B2 , i.e., whether the optimal V∗ of Problem P3-B is rank-one. If yes,
 the global optimal solution of Problem P3 can be guaranteed.
(2) h2 2 P2(2) Fortunately, we found that the rank-one optimal solution V∗ of
τ2 C ≥ B2 . (23)
N0 Problem P3-B also always exists and thus the global optimum
is guaranteed by using our proposed solution method. For
Then the transmission time minimization problem by jointly
emphasizing, we prove it as follows.
optimizing the energy beamforming and time assignment can
Theorem 4: The rank-one optimal solution V∗ of Prob-
be mathematically expressed by
lem P3-B always exists.
P3 : min τ0 + τ1 + τ2(1) + τ2(2) Proof: Compared with Problem P2, since the number
{ω,τ ,P} of constraints of P3-B is more than P2, a feasible solu-
s.t. (1), (2), (6), (22), (23) tion of P3-B is also feasible for P2. Therefore, Tr(V ) ≤
τ0 , τ1 , τ2(1) , τ2(2) ≥ 0. (24) Tr(V∗ ) ≤ τ0∗ , which indicates that V is feasible for
P3-B. As the objective function of P3-B is only related to
Although the objective function of Problem P3 is linear, τ , the feasible solution (V , t ∗ , τ ∗ ) also arrives at the optimal
the constraints (22) and (23) are non-convex sets due to the value of P3-B. Moreover, according to Lemma 1, the optimal
multiplicative coupled variables. Thus, Problem P3 also cannot  2
solution V of P2 exists and it satisfies that rank(V ) ≤ 2.
be directly solved. Following similar approach in previous So, one can infer that V satisfies V = 0, i.e., rank(V ) = 1.
section, the problem can be solved as follows. As a result, Theorem 4 is proved. 
(1) (2)
Firstly, by using the definition t = [t1 , t2 , t2 ], where Corollary 2: The global optimal solution to Problem P3 is
t1 = τ1 P1 , t2(1) = τ2(1) P2(1) and t2(2) = τ2(2) P2(2) , constraint (22) guaranteed by using our proposed solution method.
can be re-written to be Proof: The proof is similar to that of Corollary 1, which
⎧  2  is omitted here. 

⎪ |g| t1
⎨ τ1 C ≥ B1 ,
 N0 τ12    (25) V. E XTENSION TO M ULTIPLE U SER PAIR S CENARIOS

⎪ h1  t1 (1)
h2 2 t2
⎩ τ1 C + τ2(1) C (1) ≥ B 1 , In this section, we apply our proposed solution method to a
N0 τ1 N0 τ2
more complex network scenario, where multiple users desire
and constraint (23) can be re-written to be to harvest energy from a common H-AP and then use the
 harvested energy to transmit information to the H-AP. It is
(2) h2 2 t2(2)
τ2 C ≥ B2 . (26) assumed that every two users are pre-grouped into a user pair
N0 τ2(2) and the number of user pairs is denoted as K . How to pair
the users is beyond the scope of the paper.
Further, also by introducing V = τ0 ωω H , Problem P3 can In the considered K user pair system, Ui,1 and Ui,2 are
be equivalently transformed into the two users of the i -th user pair, i = 1, 2, . . . , K . Suppose
P3-A : min τ0 + τ1 + τ2(1) + τ2(2) Ui,1 is closer to H-AP, which is able to help the farther user
{V,t,τ } Ui,2 to transmit information. The transmission procedure can
s.t. (16), (17), (18), (24), (25), (26). also be divided into the DL ET phase and the UL IT phase.
In the ET phase with time length of τ0 , H-AP broadcasts The problem is non-convex, in order to solve it, by using our
energy to all users with the power P0 . Then, in the IT phase, solution method proposed in Section IV, we introduce some
(1) (1) (1)
all users transmit information to the H-AP, respectively. The auxiliary variables, i.e., ti,1 = τi,1 Pi,1 , ti,2 = τi,2 Pi,2 , and
IT phase is further divided into K sub-phases, and in the (2) (2) (2)
ti,2 = τi,2 Pi,2 . By defining V = τ0 ωω H , (33) is transformed
i -th sub-phase, the i -th user pair transmits its information to into
the H-AP. Note that, the i -th sub-phase is also divided into  (2)
three durations in order to fulfill the cooperative transmission. (2) hi,2 2 ti,2
Ri,2 = τi,2 C (2)
, i = 1, 2, . . . , K (35)
That is, Ui,1 firstly broadcasts its own information with the N0 τi,2
power Pi,1 in the duration of τi,1 , and then Ui,2 decodes the
(31) is transformed into
information of Ui,1 and forwards the decoded information to  2 
(1) (1) |g | t
H-AP with the power Pi,2 in the duration of τi,2 , and at last Ri,1 ≤ τi,1 C Ni0 τi,1i,1 , i = 1, 2, . . . , K , (36)
(2)
Ui,2 transmits own information to H-AP with the power Pi,2
(2) and (32) is transformed into
in the left duration τi,2 . 
  (1)
As the total consumed energy at Ui,1 and Ui,2 (i = hi,1 2 ti,1 (1) hi,2 2 ti,2
1, 2, . . . , K ) must be less than their harvested energy, it sat- Ri,1 ≤ τi,1 C + τi,2 C ,
N0 τi,1 N0 τ (1) i,2
isfies the following constraints, i.e.,
i = 1, 2, . . . , K . (37)
τi,1 Pi,1 ≤ η|hi,1
H
ω|2 P0 τ0 , i = 1, 2, . . . , K , (28)
Besides, the constraints (28) and (29) can be respectively
and transformed into
(1) (1) (2) (2)
τi,2 Pi,2 + τi,2 Pi,2 ≤ η|hi,2
H
ω|2 P0 τ0 , i = 1, 2, . . . , K . ti,1 ≤ η P0 Tr(hi,1 hi,1
H V), i = 1, 2, . . . , K , (38)
(29) and
Additionally, the transmission time should satisfy that (1) (2)
ti,2 + ti,2 ≤ η P0 Tr(hi,2 hi,2
H V), i = 1, 2, . . . , K . (39)

K (1) (2)
τ0 + i=1 (τi,1 + τi,2 + τi,2 ) ≤ 1,
(1) (2) (30)
τ0 , τi,1 , τi,2 , τi,2 ≥ 0, i = 1, 2, . . . , K . As a result, Problem P4 can be equivalently expressed by
K
P4-A : max (αi,1 Ri,1 + αi,2 Ri,2 )
TheK
system WSR of the K user pairs can be expressed by {R,V,t,τ } i=1
i=1 i,1 Ri,1 + αi,2 Ri,2 ), where Ri,1 and Ri,2 , respectively,

s.t. (16), (30), (36), (37), (38), (39), (40)
represent the achievable rate of the user Ui,1 and Ui,2 in the
(1) (1)
i -th user pair, which satisfy that where t denotes the set {ti,1 , ti,2 , ti,2 |i
= 1, 2, . . . , K }.
  Due to the rank-one constraint in (16), Problem P4-A is
|gi |2 Pi,1
Ri,1 ≤ τi,1 C , (31) still non-convex, so we adopt SDR technique to deal with it.
N0 By dropping the rank-one constraint, we obtain the relaxed
   (1)
hi,1 2 Pi,1 (1) hi,2 2 Pi,2 problem of Problem P4-A as
Ri,1 ≤ τi,1 C + τi,2 C , (32)
N0 N0 
K
P4-B : max (αi,1 Ri,1 + αi,2 Ri,2 ) (41)
and {V,t,τ }
 (2) i=1
(2) hi,2 2 Pi,2 s.t. Tr(V) ≤ τ0 , V 0,
Ri,2 = τi,2 C . (33)
N0 (30), (36), (37), (38), (39). (42)
αi,1 and αi,2 (i = 1, 2, . . . , K ) denote the nonnegative rate
weights of Ui,1 and Ui,2 in the i -th user pair, respectively. Similar to Section IV, it can be easily proved that Prob-
hi, j (i = 1, 2, . . . , K , j = 1, 2) denotes the complex channel lem P4-B is a convex problem, and thus it can be solved
vectors of size N × 1 for the links from U0 to Ui, j , and by interior point solver, e.g., CVX [40]. Finally, it is worth
gi (i = 1, 2, . . . , K ) is the complex channel coefficient noting that the obtained optimal solution V∗ is not nec-
between Ui,1 and Ui,2 . essarily rank-one. If so, a suboptimal solution ω can be
The optimization problem to maximize the system WSR by obtained from V∗ by adopting the Gaussian randomization
jointly optimizing energy beamforming, the time assignment approach [43].
and the power allocation can be mathematically given by
K VI. S IMULATION R ESULTS AND D ISCUSSION
P4 : max (αi,1 Ri,1 + αi,2 Ri,2 )
{R,ω,τ ,P} i=1 In this section, some simulation results are provided to
s.t. (1), (31), (32), (28), (29), (30), (34) discuss the performance of the proposed system. The simulated
network topology is shown in Figure 3, where all nodes
where R is the set {Ri,1 |i = 1, 2, . . . , K }, τ is the are placed on a 2-D plane and the position of each node
(1) (1)
set {τ0 , τi,1 , τi,2 , τi,2 |i = 1, 2, . . . , K } and P is the set is determined by its coordinate (x, y). Specifically, H-AP is
(1) (1)
{Pi,1 , Pi,2 , Pi,2 |i = 1, 2, . . . , K }. positioned at the point with coordinate (0,0), U1 is positioned
Fig. 3. Illustration of the simulation network topology on a 2-D plane.

Fig. 5. WSR comparison of our proposed user cooperation scheme and the
direct transmission scheme versus the available transmit power at H-AP for
N = 5 and N = 10.

energy beamforming, the system performance of the WPCN


can be greatly enhanced.
Secondly, we compare our proposed transmission method
(i.e., the user cooperation transmission scheme) with the direct
transmission method. In the direct transmission scheme, each
user transmit their information to the H-AP directly over the
Ui → U0 link, where i = 1, 2 and no user cooperation
between the two users was involved. For the direct transmis-
sion scheme, its ET phase is same with that of our proposed
transmission scheme, but its IT phase is divided into two
Fig. 4. WSR comparison of our proposed scheme and the benchmark scheme
versus the available transmit power at H-AP for N = 5 and N = 10. durations with time intervals τ1 and τ2 rather than three ones.
Specifically, in the first and the second durations, U1 and U2
transmit their information to U0 , respectively. To get the
at (10,0) and U2 can be located at any point in the green optimal WSR of the direct transmission, we considered the
area. All channels were generated by Rayleigh distribution optimization problem
with CN (0, 10−3 duv
−α ), where d
uv is the distance between two P5 : max{ω,τ ,P} α1 R1 + α2 R2 (43)
nodes u, v and the distance between H-AP and U1 is 10m,
and α is the exponential pathloss factor which was set to be 3. s.t. (1), (2),
The noise power N0 was set to be −100dBm and η was set τ0 + τ1 + τ2 ≤ 1, (44)
to be 0.5. τ0 , τ1 , τ2 ≥ 0, (45)
τ2 P2 ≤ η|h2H ω|2 P0 τ0 , (46)
A. Weighted Sum-Rate Comparisons where P2 denotes the power of U2 , P = [P1 , P2 ] represents
In the simulations, the weights for U1 and U2 are set to be the power allocation. τ2 = τ2 /τ and let τ = [τ0 , τ1 , τ2 ] repre-
α1 = 0.8 and α2 = 0.2, respectively. The distance between sent the time fraction assignment and satisfy
 (44) and (45).
R1 = τ1 C h1N0 P1 and R2 = τ2 C h2N0 P2 are the
2 2
H-AP and U2 is set to be 9m and the distance between
U1 and U2 is 1m, i.e., U2 is located at the point (9,0). achievable rate of two users, respectively. (46) represents
Firstly, we compare our proposed scheme with the simple the energy constraint of U2 . A similar solution method of
benchmark method in terms of WSR. In the benchmark problem P1 can be adopted and then the maximum WSR of
method, no energy beamforming is carried out at U0 , i.e. the the direct transmission is obtained by solving Problem P5.
power is allocated uniformly among all antennas. Figure 4 Figure 5 compares the WSRs of two transmission modes
plots the WSR versus the power P of the H-AP for dif- (i.e., with or without user cooperation) with respect to the
ferent numbers of antennas, i.e., N = 5 and N = 10. power P of the H-AP for different number of antennas. One
From Figure 4, it is shown that with the increase of P, can observe that the WSR of the user cooperation transmission
the WSRs of the two schemes increase and the WSR of is superior to the direct transmission.
our proposed scheme surpasses the benchmark scheme. For Thirdly, we compare the user cooperation transmission
example, when P = 30dBm, by using energy beamform- scheme with the direct transmission scheme for different user
ing, the system WSR is increased about 60% and 70% for weights in Figure 6. Without loss of generality, it is assumed
N = 10 and N = 5, respectively, which indicates that with that α1 + α2 = 1. The transmit power of the H-AP is set
Fig. 8. WSR of our proposed user cooperation transmission versus the
Fig. 6. WSR comparison of our proposed user cooperation scheme and the position of U2 .
direct transmission scheme versus the weight of U1 for N = 5 and N = 10.

Fig. 9. WSR Comparison of our proposed user cooperation transmission


and the direct transmission versus the position of U2 , where blue and green
Fig. 7. WSR of the direct transmission versus the position of U2 .
surfaces represent the WSR of direct transmission and user cooperation
transmission, respectively.

to be 10dBm. From Figure 6, it can be observed that when


the weight of U1 is larger than that of U2 , user cooperation show very different behaviors. Particularly, for the direct
transmission is superior to direct transmission. So, in this case, transmission, when U2 is placed close to the H-AP, the WSR
user cooperation transmission scheme is a better choice. When achieves maximum and with the increase of the distance
the weight of U1 is less than that of U2 , the performances between U2 and the H-AP, the WSR gradually decreases.
of direct transmission and user cooperation transmission are While, for the user cooperation transmission, the situation is
very similar to each other. In this case direct transmission more complicated. Although when U2 is placed close to the
scheme may be a better choice in order to achieve a relatively H-AP, the maximum WSR is also achieved, a local maximal
simple system configuration. Moreover, in Figure 6, it shows WSR exists when U2 is located on the straight line between
that when α1 ≥ 0.5, user cooperation may bring performance the H-AP and U1 and closer to U1 .
gains. The reason is that when the farther user U1 has a larger To get more insights, Figure 9 plots the results of Figure 7
weight than U2 , it is with more importance than U2 . Thus, and Figure 8 in the same coordinate system. It can be seen
its information rate must be enhanced at first. In this case, that when U2 is located close to the local optimal point,
in order to enhance the farther users information rate, user the WSR of the user cooperation transmission surpasses the
cooperation is a better choice. direct transmission while when U2 is located at the rest
Fourthly, the effect of the position of U2 on the system WSR locations, the WSR of the user cooperation transmission is
for the direct transmission and user cooperation transmission less than the direct transmission. This observation means that
is discussed via simulations. The simulated network topology 1) by involving cooperative relaying between the two users
is shown in Figure 3. The transmit power of the H-AP is set cannot always bring performance gain to the system, which is
to be 10dBm and the number of antennas is set to be 10. consistent with that of the cooperative relaying in traditional
Figure 7 and Figure 8 plot the effect of the position of U2 non-wireless powered networks; 2) when U2 is positioned
for the direct transmission and user cooperation transmission, relatively close to U1 , the user cooperation transmission is
respectively. It can be seen that their WSR performances a better choice to achieve higher WSR, because when U2
Fig. 10. Transmission mode selection regions.
Fig. 12. Transmission time comparison of our proposed user cooperation
scheme and the benchmark scheme versus the available transmit power at
H-AP for N = 5 and N = 10.

It can be observed that with the increase of the number of


the antennas, the performance can be significantly improved,
but it cannot be improved infinitely as the growing rate of the
curves decreases with the increase of N.

B. Transmission Time Comparisons


In the simulations associated with this subsection, the total
amount of information required to be transmitted by U1 and U2
over unit bandwidth are set to be B1 = B2 = 1nats. U2 is
located at the point (9,0).
Firstly, we compare the proposed scheme with the afore-
Fig. 11. WSR of our proposed user cooperation transmission versus the mentioned benchmark scheme to demonstrate its superiority in
number of antennas. terms of the total transmission time. Figure 12 compares the
transmission time of the two schemes versus transmit power P
is positioned relatively close to U1 , the channel between of H-AP. From Figure 12, it is seen that with the increase
U1 and U2 is with relatively better channel condition. In this of P, the transmission time of both our proposed scheme and
case, U2 has much higher possibility to successfully decode the the benchmark scheme decrease, and the transmission time
information of U1 and then help U1 to forward the information, of our proposed scheme is less than that of the benchmark
which utilizes the cooperative diversity between the two users scheme. For example, when P = 0dBm, the transmission
and may greatly enhance the system achievable information time is decreased about 80% and 90% by adopting energy
rate performance. beamforming for N = 5 and N = 10, respectively.
Following Figure 9, we plot the transmission mode selection Secondly, we compare the proposed scheme with the afore-
regions in terms of the location of U2 in Figure 10, where the mentioned direct transmission scheme in terms of the total
red curve is a circular arc with the distance of 10m to the transmission time. As described previously, in the direct trans-
H-AP and U2 should be located in the area below the red arc mission scheme, the IT phase is divided into two durations
to act as a helping relay for U1 as in this area U2 is closer to with time intervals τ1 and τ2 . In the first and the second dura-
H-AP than U1 . According to our simulation results, the area tions U1 and U2 transmits its information to U0 , respectively.
below the red arc is divided into two regions, i.e., region I and To find the minimum total transmission time, we considered
region II. The part above the red arc is labeled as region III. the following the following optimization problem
In the region I, user cooperation transmission mode is a better
choice. In region II, the direct transmission mode is a better P6 : min{ω,τ ,P} τ0 + τ1 + τ2
choice. In the region III, U2 has a larger distance to H-AP s.t. (1), (2),
than U1 , so U1 should act as the helping node for U2 when τ2 P2 ≤ η|h2H ω|2 P0 τ0 , (47)
user cooperation gain is desired to be achieved. Even though,  
hi 2 Pi
in region III, the choice of the direct transmission or the user τi C ≥ Bi , ∀i = 1, 2 (48)
N0
cooperation transmission also shall be determined by their
τ0 , τ1 , τ2 ≥ 0, (49)
relative locations.
Finally, Figure 11 plots the WSR versus the number of where τ = [τ0 , τ1 , τ2 ] represents the time assignment and sat-
the antennas for our proposed user cooperation transmission. isfies the constraint (24). (47) represents the energy constraint
Fig. 15. Transmission time of the direct transmission versus the
position of U2 .
Fig. 13. Transmission time comparison of our proposed user cooperation
scheme and the direct transmission scheme versus the available transmit power
at H-AP for N = 5 and N = 10.

Fig. 16. Transmission time of our proposed user cooperation transmission


versus the position of U2 .

Fig. 14. Transmission time comparison of our proposed user cooperation


the transmission time of our scheme decreases while that of
scheme and the direct transmission scheme versus the ratio of the amount of the direct transmission scheme increase, which indicates that
information of U1 to U2 . when U1 has more information than U2 , the user cooperation
should be employed to achieve better system performance.
of U2 , and (48) describes the constraints for guaranteeing the Fourthly, the effect of the position of U2 on the transmis-
successfully information delivering of U1 and U2 . sion time for the direct transmission and user cooperation
Figure 13 compares the minimal transmission time of transmission is demonstrated. The network topology is shown
our proposed scheme and the direct transmission versus the in Figure 3, where U2 is located at any point in the green
transmit power P of H-AP. It is observed that with the area. The transmit power of the H-AP is set to be 10dBm
increase of P, the transmission time of our proposed user and the number of antennas is set to be 10. The total amount
cooperation transmission scheme and the direct transmission of information required to be transmitted by U1 and U2 are
scheme decreases, and the user cooperation transmission out- set to be B1 = B2 = 1nats. Figure 15 and Figure 16 plot
performs the direct transmission, which indicates that with the effect of the position of U2 for the direct transmission
user cooperation, the system performance of the WPCN can and user cooperation transmission, respectively. It can be seen
be greatly enhanced. Besides, it is also observed that more that their transmission time performances show very different
transmit antennas may yield lower transmission time for the features. For the direct transmission, when U2 is placed close
system. to the H-AP, the transmission time achieves minimum. Very
Thirdly, Figure 14 shows the system performance with differently, for the user cooperation transmission, when U2 is
respect to the ratio of the amount of information of U1 to that placed on the straight line from the H-AP to U1 and closer
of U2 , where the total information
 amount of the two users are to U1 , the transmission time achieves minimum.
fixed to be 2nats, i.e., 2i=1 Bi = 2nats. The transmit power Figure 17 compares the transmission time for the direct
of the H-AP was set to be 10dBm. It is seen that our proposed transmission and user cooperation transmission with respect
user cooperation transmission scheme and the direct trans- to the position of U2 . It is shown that when U2 is
mission scheme show very different performance behaviors. placed around U1 , the performance of the user cooperation
That is, with the increase of the amount of information of U1 , transmission is superior to that of the direct transmission,
simulations, the distance between H-AP and the Ui,1 is
randomly selected within [9.5, 10.5]m, the distance between
H-AP and the Ui,2 is randomly selected within [8.5, 9.5]m, and
all the distances between two users in each user pair is 1m. The
weight of the farther user and that of the closer user is set to be
identical respectively. That is, the weight of the farther user is
set to be 0.8 and the weight of the closer user is set to be 0.2.
The number of antennas are 5. It shows that the system WSR
increases with the increment of K . This is because that more
user pairs, more flexibility to optimally assignment the system
resources, which may increase the system sum rate. But the
system WSR cannot be increased infinitely when K grows
due to the limit number of antennas and the total available
power of the system. It is also seen that in our simulations,
the performances of user cooperation transmission is always
Fig. 17. Transmission time comparison of our proposed user cooperation superior to that of the direct transmission, which is consistent
transmission versus the position of U2 , where blue and green surfaces
represent the transmission time of direct transmission and user cooperation with our theoretical discussion in this paper.
transmission, respectively.
VII. C ONCLUSION
This paper studied the joint energy beamforming design,
time assignment and power allocation for WPCN. We for-
mulated two optimization problems to maximize the system
WSR and minimize the total transmission time, respectively.
The problems were solved by using variable substitution
operations and SDR method. We theoretically proved that the
global optimum can be guaranteed. Simulation results show
that system performances of WSR and transmission time can
be significantly enhanced by using energy beamforming and
user cooperation. It is also shown that when total amount of
information of the two users is fixed, with the increase of
the information amount of the user relatively farther away
from the H-AP, the transmission time of the user coopera-
tion scheme decreases while that of the direct transmission
increases. Besides, the effects of user position on the system
Fig. 18. WSR comparison of our proposed user cooperation scheme and the
direct transmission scheme versus the number of user pairs.
performance are also demonstrated, which provides insights on
the WPCN. That is, to achieve the maximum WSR, U2 should
be placed close to the H-AP while to achieve the minimum
which means that when U2 is positioned relatively close transmission time, U2 be placed in the straight line from
to U1 , the user cooperation transmission is a better choice the H-AP to U1 and closer to U1 . We also extended our
to achieve lower total transmission time, because when U2 scheme to multiple user pair scenarios and showed its system
is positioned relatively close to U1 , as the system achievable performance gain by simulations.
information rate is improved, a lower transmission time also
can be achieved. R EFERENCES
Further, combining the results in Figure 9 and 17, one [1] X. Lu, P. Wang, D. Niyato, D. I. Kim, and Z. Han, “Wireless networks
can observe that for the direct transmission, when U2 is with RF energy harvesting: A contemporary survey,” IEEE Commun.
Surveys Tuts., vol. 17, no. 2, pp. 757–789, 2nd Quart., 2015.
placed close to the H-AP, both WSR and the total trans- [2] H. J. Visser and R. J. M. Vullers, “RF energy harvesting and transport
mission time achieve their optimum. However, for the user for wireless sensor network applications: Principles and requirements,”
cooperation transmission, the two performance indexes cannot Proc. IEEE, vol. 101, no. 6, pp. 1410–1423, Jun. 2013.
[3] S. Bi, C. K. Ho, and R. Zhang, “Wireless powered communication:
simultaneously achieve the optimum. Specifically, to achieve Opportunities and challenges,” IEEE Commun. Mag., vol. 53, no. 4,
the maximum WSR, U2 should be placed close to the H-AP pp. 117–125, Apr. 2015.
while to achieve the minimum transmission time, U2 be placed [4] S. Bi, Y. Zeng, and R. Zhang, “Wireless powered communication
networks: An overview,” IEEE Wireless Commun., vol. 23, no. 2,
in the straight line from the H-AP to U1 and closer to U1 . pp. 10–18, Apr. 2016.
[5] M.-L. Ku, W. Li, Y. Chen, and K. J. R. Liu, “Advances in
energy harvesting communications: Past, present, and future chal-
C. The Performances of Multiple User Pair Networks lenges,” IEEE Commun. Surveys Tuts., vol. 18, no. 2, pp. 1384–1412,
In this subsection, we present the simulation results for 2nd Quart., 2016.
[6] S. Ulukus et al., “Energy harvesting wireless communications: A review
multiple user pair systems, where the system WSR versus of recent advances,” IEEE J. Sel. Areas Commun., vol. 33, no. 3,
the number of user pairs K is plotted in Figure 18. In the pp. 360–381, Mar. 2015.
[7] T. Li, P. Fan, Z. Chen, and K. B. Letaief, “Optimum transmission policies [30] Y. Li, W. Wang, J. Kong, W. Hong, X. Zhang, and M. Peng, “Power
for energy harvesting sensor networks powered by a mobile control allocation and subcarrier pairing in OFDM-based relaying networks,” in
center,” IEEE Trans. Wireless Commun., vol. 15, no. 9, pp. 6132–6145, Proc. IEEE ICC, May 2008, pp. 2602–2606.
Sep. 2016. [31] C. Zhong, G. Zheng, Z. Zhang, and G. K. Karagiannidis, “Optimum
[8] T. Le, K. Mayaram, and T. Fiez, “Efficient far-field radio frequency wirelessly powered relaying,” IEEE Signal Process. Lett., vol. 22, no. 10,
energy harvesting for passively powered sensor networks,” IEEE J. pp. 1728–1732, Oct. 2015.
Solid-State Circuits, vol. 43, no. 5, pp. 1287–1302, May 2008. [32] Y. Gu, H. Chen, Y. Li, and B. Vucetic, “An adaptive transmission
[9] L. R. Varshney, “Transporting information and energy simultaneously,” protocol for wireless-powered cooperative communications,” in Proc.
in Proc. IEEE ISIT, Jul. 2008, pp. 1612–1616. IEEE ICC, Jun. 2015, pp. 4223–4228.
[10] P. Grover and A. Sahai, “Shannon meets Tesla: Wireless information [33] H. Chen, Y. Li, J. L. Rebelatto, B. F. Uchôa-Filho, and B. Vucetic,
and power transfer,” in Proc. IEEE ISIT, 2010, pp. 2363–2367. “Harvest-then-cooperate: Wireless-powered cooperative communica-
[11] R. Zhang and C. K. Ho, “MIMO broadcasting for simultaneous wire- tions,” IEEE Trans. Signal Process., vol. 63, no. 7, pp. 1700–1711,
less information and power transfer,” in Proc. IEEE GLOBECOM, Apr. 2015.
Dec. 2011, pp. 1–5. [34] H. Ju and R. Zhang, “User cooperation in wireless powered communica-
[12] L. Liu, R. Zhang, and K.-C. Chua, “Wireless information transfer tion networks,” in Proc. IEEE GLOBECOM, Dec. 2014, pp. 1430–1435.
with opportunistic energy harvesting,” IEEE Trans. Wireless Commun., [35] M. Zhong, S. Bi, and X.-H. Lin, “User cooperation for enhanced
vol. 12, no. 1, pp. 288–300, Jan. 2013. throughput fairness in wireless powered communication networks,” in
[13] K. Huang and E. Larsson, “Simultaneous information and power transfer Proc. ICT, 2016, pp. 1–6.
for broadband wireless systems,” IEEE Trans. Signal Process., vol. 61, [36] X. Chen, C. Yuen, and Z. Zhang, “Wireless energy and information
no. 23, pp. 5972–5986, Dec. 2013. transfer tradeoff for limited-feedback multiantenna systems with energy
[14] D. W. K. Ng, E. S. Lo, and R. Schober, “Wireless information and power beamforming,” IEEE Trans. Veh. Technol., vol. 63, no. 1, pp. 407–412,
transfer: Energy efficiency optimization in OFDMA systems,” IEEE Jan. 2014.
Trans. Wireless Commun., vol. 12, no. 12, pp. 6352–6370, Dec. 2013. [37] Q. Sun, G. Zhu, C. Shen, X. Li, and Z. Zhong, “Joint beamform-
[15] X. Zhou, R. Zhang, and C. K. Ho, “Wireless information and power ing design and time allocation for wireless powered communication
transfer in multiuser OFDM systems,” IEEE Trans. Wireless Commun., networks,” IEEE Commun. Lett., vol. 18, no. 10, pp. 1783–1786,
vol. 13, no. 4, pp. 2282–2294, Apr. 2014. Oct. 2014.
[16] Q. Zhang, X. Huang, Q. Li, and J. Qin, “Cooperative jamming aided [38] K. Xiong, C. Chen, G. Qu, P. Fan, and K. B. Letaief, “Group coop-
robust secure transmission for wireless information and power transfer eration with optimal resource allocation in wireless powered commu-
in MISO channels,” IEEE Trans. Commun., vol. 63, no. 3, pp. 906–915, nication networks,” IEEE Trans. Wireless Commun., vol. 16, no. 6,
Mar. 2015. pp. 3840–3853, Jun. 2017.
[17] A. A. Nasir, X. Zhou, S. Durrani, and R. A. Kennedy, “Relaying [39] K. Xiong, P. Fan, Y. Lu, and K. B. Letaief, “Energy efficiency with
protocols for wireless energy harvesting and information process- proportional rate fairness in multirelay OFDM networks,” IEEE J. Sel.
ing,” IEEE Trans. Wireless Commun., vol. 12, no. 7, pp. 3622–3636, Areas Commun., vol. 34, no. 5, pp. 1431–1447, May 2016.
Jul. 2013. [40] M. Grant and S. Boyd. (Sep. 2013). CVX: MATLAB Software for
[18] C. Zhong, H. A. Suraweera, G. Zheng, I. Krikidis, and Z. Zhang, Disciplined Convex Programming, Version 2.0 Beta. [Online]. Available:
“Wireless information and power transfer with full duplex relay- http://cvxr.com/cvx
ing,” IEEE Trans. Commun., vol. 62, no. 10, pp. 3447–3461, [41] Y. Huang and D. P. Palomar, “Rank-constrained separable semidefinite
Oct. 2014. programming with applications to optimal beamforming,” IEEE Trans.
[19] Z. Ding, S. M. Perlaza, I. Esnaola, and H. V. Poor, “Power allocation Signal Process., vol. 58, no. 2, pp. 664–678, Feb. 2010.
strategies in energy harvesting wireless cooperative networks,” IEEE [42] Y. Liang and V. V. Veeravalli, “Gaussian orthogonal relay channels:
Trans. Wireless Commun., vol. 13, no. 2, pp. 846–860, Feb. 2014. Optimal resource allocation and capacity,” IEEE Trans. Inf. Theory,
[20] K. Xiong, P. Fan, C. Zhang, and K. B. Letaief, “Wireless information vol. 51, no. 9, pp. 3284–3289, Sep. 2005.
and energy transfer for two-hop non-regenerative MIMO-OFDM relay [43] Z.-Q. Luo, W.-K. Ma, A. M.-C. So, Y. Ye, and S. Zhang, “Semidefinite
networks,” IEEE J. Sel. Areas Commun., vol. 33, no. 8, pp. 1595–1611, relaxation of quadratic optimization problems,” IEEE Signal Process.
Aug. 2015. Mag., vol. 27, no. 3, pp. 20–34, May 2010.
[21] X. Di, K. Xiong, P. Fan, and H.-C. Yang, “Simultaneous wire-
less information and power transfer in cooperative relay networks
with rateless codes,” IEEE Trans. Veh. Technol., vol. 66, no. 4,
pp. 2981–2996, Apr. 2017.
[22] K. Ishibashi, “Dynamic harvest-and-forward: New cooperative diversity
with RF energy harvesting,” in Proc. WCSP, 2014, pp. 1–5.
[23] I. Krikidis, “Relay selection in wireless powered cooperative networks
with energy storage,” IEEE J. Sel. Areas Commun., vol. 33, no. 12,
pp. 2596–2610, Dec. 2015.
[24] I. Krikidis, S. Timotheou, S. Nikolaou, G. Zheng, D. W. K. Ng, and
R. Schober, “Simultaneous wireless information and power transfer in
modern communication systems,” IEEE Commun. Mag., vol. 52, no. 11,
pp. 104–110, Nov. 2014.
[25] K. Xiong, B. Wang, and K. J. R. Liu, “Rate-energy region of SWIPT
for MIMO broadcasting under nonlinear energy harvesting model,” IEEE
Trans. Wireless Commun., vol. 16, no. 8, pp. 5147–5161, Aug. 2017.
[26] H. Ju and R. Zhang, “Throughput maximization in wireless powered
communication networks,” IEEE Trans. Wireless Commun., vol. 13,
no. 1, pp. 418–428, Jan. 2014.
[27] M. A. Abd-Elmagid, T. ElBatt, and K. G. Seddik, “Optimization of
wireless powered communication networks with heterogeneous nodes,”
in Proc. IEEE GLOBECOM, Dec. 2015, pp. 1–7.
[28] H. Kim, H. Lee, M. Ahn, H.-B. Kong, and I. Lee, “Joint subcarrier
and power allocation methods in full duplex wireless powered commu-
nication networks for OFDM systems,” IEEE Trans. Wireless Commun.,
vol. 15, no. 7, pp. 4745–4753, Jul. 2016.
[29] M. Shaqfeh and H. Alnuweiri, “Joint power and resource allocation for
block-fading relay-assisted broadcast channels,” IEEE Trans. Wireless
Commun., vol. 10, no. 6, pp. 1904–1913, Jun. 2011.

Вам также может понравиться