SPECIAL SECTION ON WIRELESSLY POWERED NETWORKS ALGORITHMS,

APPLICATIONS AND TECHNOLOGIES

Received October 30, 2018, accepted November 12, 2018, date of publication November 19, 2018,
date of current version December 19, 2018.
Digital Object Identifier 10.1109/ACCESS.2018.2882196

OFDM Based SWIPT for Two-Way

AF Relaying Network
WEIDANG LU 1 , (Member, IEEE), WEILIN ZHAO1 , SU HU 2, (Member, IEEE),
XIN LIU 3 , (Member, IEEE), BO LI 4 , (Member, IEEE),
AND YI GONG 5 , (Senior Member, IEEE)
1 College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023, China
2 National Key Laboratory on Communications, University of Electronic Science and Technology of China, Chengdu 611731, China
3 School of Electronic Information and Electrical Engineering, Dalian University of Technology, Dalian 116024, China
4 School of Information and Electrical Engineering, Harbin Institute of Technology, Weihai 264209, China
5 Department of Electrical and Electronic Engineering, Southern University of Science and Technology, Shenzhen 518055, China

Corresponding author: Su Hu (husu@uestc.edu.cn)

This work was supported in part by the National Natural Science Foundation of China under Grant 61871348, Grant 61601221, and
Grant U1833102, and in part by the Project funded by the China Postdoctoral Science Foundation under Grant 2017M612027.

ABSTRACT We consider a three-node two-way relaying network, where an amplify-and-forward (AF) relay
helps to forward the information of source nodes by using the energy harvested from the radio frequency
signals transmitted from the two source nodes. The source and relay nodes work on half-duplex, which
completes the transmission in two phases. Specifically, in the first transmission phase, two source nodes
transmit their signals to relay. The received signals at relay will be divided into two subcarrier groups,
to perform information decoding and energy harvesting separately. In the second transmission phase, after
subcarrier pairing, relay amplifies the received signals then forwards them to the two source nodes with the
harvested energy. An joint resource optimization problem, including subcarrier grouping, paring and power
allocation, is formulated to maximize the sum transmission rate of source nodes with power constraints.
Simulation analyses show that our proposed algorithm outperforms the other two benchmark algorithms and
reveal the system performance influence of relay location and total transmission power.

INDEX TERMS SWIPT, two-way relay, subcarrier grouping, subcarrier pairing, power allocation.

I. INTRODUCTION
Through energy harvesting, the battery can be powered by
ambient sources, such as wind, solar, and vibration, which
can prolong the lifetime of wireless powered devices effec-
tively. But the wireless devices may be deployed under-
ground or indoor place, which will make the ambient
sources difficult to harvest. Comparing with the unstable
nature resources, radio frequency signals are more obtain-
able and controllable [1]–[3], which can provide continual
stable resource to the wireless devices through wireless power
transfer.
Since the RF signals carry both information and energy,
it is possible to realize simultaneous wireless information
and power transfer (SWIPT), which has received increased
research attention recently [4]. However, in practical sys-
tem, due to the electronic circuits limitation, the receiver
cannot perform information decoding (ID) and energy har-
vesting (EH) on the same RF signals. Thus, it is essential
to divide the received signal into two separate parts. Time
switching (TS) and power splitting (PS) are two practical
schemes to realize SWIPT. In TS scheme, the receiver works
either in ID mode or EH mode. In PS scheme, the receiver
splits the RF signals into two streams using a power splitter,
to perform ID and EH, respectively.
Cooperative relaying has been known as an efficient solu-
tion for cutting down the path-loss effect as well as enhanc-
ing transmission reliability in wireless communications [5].
Decode-and-forward (DF) relaying and amplify-and-forward
relaying are two major cooperative relaying protocols. Since
DF relaying needs to decode information first, then regen-
erates the received signal and forwards it, which makes
its implementation complexity higher than AF relaying.
Integrating cooperative relaying with SWIPT, significant
improvements can be achieved in both flexibility and system
performance.
Various forms of cooperative relaying schemes have been
proposed for SWIPT system. In one-way relaying (OWR)
SWIPT system, a wireless powered relay helps to forward

2169-3536
2018 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 6, 2018 Personal use is also permitted, but republication/redistribution requires IEEE permission. 73223
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

the source signal to the destination [5]–[17]. Reference [5]
proposed time switching relaying (TSR) protocol and power
splitting relaying (PSR) protocol in AF SWIPT system,
in which the delay-tolerant and the delay-limited transmis-
sion modes are adaptively selected according to whether
the destination node is able to buffer the received infor-
mation blocks. Reference [6] compared the ergodic and
outage capacity of PSR and TSR based SWIPT system.
Reference [7] proposed adaptive maximum ratio combin-
ing (MRC) protocol based on the TSR and PSR, where
direct link was taken into consideration by adopting MRC
method with the purpose of maximizing system through-
put. Reference [9] proposed a multiple-input-single-output
(MISO) downlink multiuser SWIPT system, which focuses
on maximizing the system energy efficiency by joint power
splitting and beamforming. Reference [11] proposed three
wireless power transmission (WPT) schemes according to
different energy supply, in which performances of three WPT
schemes at various relay positions are compared. Applying
SWIPT technology to MIMO-OFDM relay networks, [17]
discussed the impacts of the number of antennas and
subcarriers.
Two-way relaying (TWR) protocol is more frequency effi-
ciency than OWR protocol, since TWR protocol only occu-
pies a quarter of the transmission time of OWR protocol
in one transmission [18]–[30]. Reference [19] optimized the
resource allocation of the power splitting and time phase
ratios in a DF-TWR network, in which the goal is to mini-
mize the probability of system outages. Reference [21] stud-
ied power allocation to achieve a balance between spectral
efficiency and energy efficiency (EE) with PS protocol in
an AF-TWR network. Reference [22] studied the problem
of distributed beamforming in an AF-TWR network with
multiple EH-based relays, where the effect of interference
on the performance was also investigated. Reference [23]
combined SWIPT with wireless body area network (WBAN),
and adopted both PS and TS protocols, in which optimal time
and power ratios were studied to maximum system through-
put. Reference [26] proposed a novel dynamic asymmetric
PS scheme in a three-step multiplicative AF-TWR network,
which is simpler than that of two-step on circuitry design and
obtained the optimal solutions by dinkelbach-based iterative
algorithm. The above works mainly focus on power allocation
with different design objectives, relay selection is rarely con-
sidered as [27]. Reference [27] studied a joint relay selection
and optimal power allocation scheme in asymmetric TWR
networks, which selected a relay from K available relays.
OFDM is considered as an key technology for future
wireless networks [31]–[33], because it can allocate sub-
carriers resources flexibly and acquire high data rate in
the frequency-selective fading channel. Therefore, with the
purpose of maximizing the system sum transmission rate,
the objective of this paper is to use OFDM to design SWIPT
scheme for TWR network.
The main contributions of this paper are summarized as
follows:
73224
W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network
FIGURE 1. System model.
• We propose an OFDM based SWIPT scheme for an
AF-TWR network, where a wireless powered relay uti-
lizes two disjoint subcarriers groups to perform ID and
EH, respectively. As the relay only needs subcarriers
index used for ID and EH, no splitter is needed to equip
at the relay. Comparing with PS and TS based SWIPT
scheme, the complexity can be significantly reduced at
the relay.
• We propose an efficient algorithm to solve the joint
subcarrier grouping, paring and power optimization allo-
cation problem for the proposed OFDM based SWIPT
scheme, to maximize system sum transmission rate
while meeting the power constraint.
• Simulation analyses demonstrate the better performance
of our proposed scheme and reveal the impact of relay
location and transmission power on the system perfor-
mance.
The remainder of this paper is organized as follows.
In Section II, we introduce the system model. Optimal solu-
tion is presented in Section III. Simulation results are pro-
vided in Section IV to illustrate the performance of the
proposed algorithms. Finally, we conclude this paper in
Section V.
II. SYSTEM MODEL
As shown in Fig. 1, our system model is consisted of two
source nodes S1 , S2 , and one AF relay node R. All the nodes
operate in half-duplex mode in two equal phase to transfer
signal, which are OFDM modulated. We assume the ideal
OFDM modulation and demodulation in this paper, and chan-
nel state information of two transmission phases is assume
to be known at the corresponding nodes [5]–[10]. In the first
transmission phase, two source nodes transmit its signal x1,k
and x2,k over subcarrier k ∈ K, where K = {1, 2, . . . , K }
denotes the subcarriers set. The received signal at R can be
written as
q q
yR,k = h1,k pIs1,k x1,k + h2,k pIs2,k x2,k + nR,k (1)
where pIsi ,k , i ∈ {1, 2}, denotes the transmit power used for
ID over subcarrier k of source node Si . h1,k and h2,k represent
the channel coefficients of the S1 → R link and S2 → R link
over subcarrier k, respectively. All involved links are block-
fading, which are assumed to be constant in a transmission
phase but vary independently from one phase to another.
nR,k ∼ CN (0, σk2 ), represents the Gaussian random noise at
relay with zero mean and variance σk2 .
VOLUME 6, 2018
W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network

FIGURE 2. Two-way amplify-and-forward OFDM SWIPT relay.

As shown in Fig.2 at the top of the page, the received signal and βk,k 0 is the positive amplification factor
at R is firstly performed by analog-to-digital (A/D) conver- s
sion and serial-to-parallel (S/P) conversion. After removing pr,k 0
βk,k 0 = 2 2 (4)
its cyclic prefix (CP), subcarrier allocation is performed, p h1,k  + pI h2,k  + σ 2
I
 
s1,k s2,k k
in which a part of the subcarriers, denoted as G1 , are utilized
for ID, and the rest subcarriers, denoted as G2 , are utilized for where pr,k 0 is the allocated transmit power over subcarrier k 0
EH. Thus, the harvested energy at R can be written as at R.
X Since Si knows its own signal xi and channel state informa-
Q= ζ (pEs1,k |h1,k |2 + pEs2,k |h2,k |2 + σk2 ) (2) tion, it can decode the information successfully. Therefore,
k∈G2
after two phases transmission, the signal to noise ratio of
where ζ represents the EH conversion efficiency and pEsi ,k , link S1 to S2 and S2 to S1 over subcarrier k and k 0 could be
i ∈ {1, 2}, represents the transmit power of source node Si expressed as
used for EH over subcarrier k.
pIs1,k γ1,k pr,k 0 γ2,k 0
After subcarrier allocation, the relay continues to perform SNR1k,k 0 =   (5)
DFT, Amplification, IDFT, CP insertion, P/S and D/A with pr,k 0 γ2,k 0 + pIs1,k γ1,k + pIs2,k γ2,k + 1
the subcarriers in G1 by utilizing the subcarrier k 0 to pair
with subcarrier k, k ∈ G1 , to forward the source node pIs2,k γ2,k pr,k 0 γ1,k 0
SNR2k,k 0 =  (6)
signals with the harvested energy Q. Let ρk,k 0 ∈ {0, 1} rep-

pr,k 0 γ1,k 0 + pIs1,k γ1,k + pIs2,k γ2,k + 1
resents the indicator of subcarrier pairing, which satisfying
K 2
ρk,k 0 = 1, ∀k. If subcarrier k 0 in the second phase is |h1,k |2 |h2,k |2
P 
h
1,k 0
where γ1,k = , γ2,k = , γ1,k 0 =

and
k 0 =1 σk2 σk2 σk2
paired with subcarrier k in the first phase, ρk,k 0 = 1. And 
h
2
0
one subcarrier k could only be paired with one subcarrier k 0 . γ2,k 0 = 2,k2 .
σk
Then, the received signal at Si can be rewritten as Then, the transmission rate of source node Si with relay R’s
forwarding can be given as
yi,k 0 = βk,k 0 hi,k 0 yR,k + ni,k 0 (3)
K
1  
where ni,k 0 is the noise at Si over subcarrier k 0 , hi,k 0 represents
X X
Rsi = ρk,k 0 ln 1 + SNRik,k 0 (7)
the channel coefficient of the R → Si link over subcarrier k 0 2
k∈G1 k 0 =1

 
X X
L(G, P, ρ) = Rs1 + Rs2 + α1 Ps1 − pIs1,k − pEs1,k 
k∈G1 k∈G2
   
X X K
X X
+ α2 Ps2 − pIs2,k − pEs2,k  + α3 Q − ρk,k 0 pr,k 0  (11)
k∈G1 k∈G2 k∈G1 k 0 =1

VOLUME 6, 2018 73225

 W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network

K
The sum transmission rate of source nodes S1 and S2 can X X
1α3 = (Q − ρk,k 0 pr,k 0 ) (15)
be given as
k∈G1 k 0 =1
Rs = Rs1 + Rs2 (8) We can obtain the optimal dual variables of convergence
by updating α t+1 = α t + ξ t 1α, where ξ t is the step size.
III. OPTIMIZATION PROBLEM The computational complexity of this sub-gradient algorithm
The two source nodes intend to maximize their sum trans- is O(V ν ) [35], in which V is dual variables number and ν is a
mission rate. In this section, we aim to jointly determine nonnegative integer. Because we have three dual variables in
the optimization of subcarrier grouping G = {G1 , G2 }, this paper, the complexity can be denoted by O(3ν ). After get-
subcarrier pairing ρ = {ρk,k 0 } and power allocation P = ting the dual variables, the optimal G, P, ρ can be obtained
(pIs1,k , pIs2,k , pEs1,k , pEs2,k , pr,k 0 ) with the power constraints, through the following two procedures. First, we assume that
so that the sum transmission rate of source nodes is maxi- G and ρ are known to derive the optimal P. Secondly, deriv-
mized. ing the optimal ρ and G.
This joint optimization problem is formulated as
max Rs (9) A. DERIVING THE OPTIMAL P
{G ,P ,ρ} We can obtain the partial derivatives of L(G, P, ρ) to
subject to pEs1,k , pEs2,k pIs1,k , pIs2,k and pr,k 0 , as
∂L
X X
C1 : pIs1,k + pEs1,k ≤ Ps1 2
= −α1 + α3 ζ h1,k 

(16)
k∈G1 k∈G2 ∂ps1,k
E

∂L
X X
C2 : pIs2,k + pEs2,k ≤ Ps2
= −α2 + α3 ζ |h2,k |2 (17)
k∈G1 k∈G2 ∂pEs2,k
K
C3 :
X X
ρk,k 0 pr,k 0 ≤ Q ∂L ∂L ∂L
where , I and are given as (18)-(20) shown
k∈G1 k 0 =1 ∂ps1,k ∂ps2,k
I ∂pr,k 0
K at the bottom of the next page. By Karush-Kuhn-Tucker
conditions, we could know that the optimal solution of the
X
C4 : ρk,k 0 ∈ {0, 1}, ρk,k 0 = 1, ∀k
k 0 =1
Lagrange function can be obtained when the value of partial
derivative is zero. Therefore, the optimal power used for EH,
where C1 and C2 indicate the sum transmission power
pE∗ E∗
s1,k and ps2,k can be obtained as
constraints of two source nodes, respectively. C3 indicates
( 2
that relay cannot consume more energy than it harvested, pmax , α3 ζ h1,k  − α1 ≥ 0

E∗
and C4 represents subcarrier pairing constraint. The above ps1,k = 2 (21)
pmin , α3 ζ h1,k  − α1 < 0

optimization problem is a non-convex optimization prob- ( 2
pmax , α3 ζ h2,k  − α2 ≥ 0

lem. If the problem satisfies the time division condition E∗
and the number of subcarriers is sufficient, then the above ps2,k = 2 (22)
pmin , α3 ζ h2,k  − α2 < 0

problem can be solved equivalently by the Lagrangian dual
method [34]. The specific proof process is given in the where pmax and pmin are the maximum and minimum power
appendix. We denote the Lagrangian dual function of the constraints over each subcarrier, individually. And we can
optimization problem in (9) as obtain the optimal power of every subcarrier k used for ID,
∗ and pI ∗ by getting the positive root of the following
pIs1,k s2,k
g(α) = max L(G, P, ρ) (10) two equations.
{G ,P ,ρ}

where L(G, P, ρ) is shown in (11) at the bottom of previous a4 (pIs1,k )4 +a3 (pIs1,k )3 +a2 (pIs1,k )2 +a1 pIs1,k + a0 = 0 (23)
page, in which α = (α1 , α2 , α3 ) denotes non-negative dual b4 (pIs2,k )4 +b3 (pIs2,k )3 +b2 (pIs2,k )2 +b1 pIs2,k + b0 = 0 (24)
variables vector. And the dual optimization problem can be
expressed as where a4 , a3 , a2 , a1 , a0 and b4 , b3 , b2 , b1 , b0 are shown at the
top of page 6. And M (k) = 1 + pIs2,k γ2,k + pr,k 0 γ1,k 0 , N (k) =
min g(α) (12) 1 + pIs2,k γ2,k + pr,k 0 γ2,k 0 , U (k) = 1 + pIs1,k γ1,k + pr,k 0 γ1,k 0 ,
α
The minimization problem in (12) can be settled by utiliz- V (k) = 1 + pIs1,k γ1,k + pr,k 0 γ2,k 0 .
ing sub-gradient algorithm. The sub-gradients can be easily Similarly, we can obtain the optimal power allocation at
obtained by the following formulas relay R, p∗r,k 0 by getting the positive root of the following
X X equation
1α1 = (Ps1 − pIs1,k − pEs1,k ) (13)
k∈G1 k∈G2
c4 (pr,k 0 )4 + c3 (pr,k 0 )3 + c2 (pr,k 0 )2 + c1 pr,k 0 + c0 = 0 (25)
where c4 , c3 , c2 , c1 , c0 are given at the top of page 6. And
X X
1α2 = (Ps2 − pIs2,k − pEs2,k ) (14)
k∈G1 k∈G2 X (k) = 1 + pIs1,k γ1,k + pIs2,k γ2,k .

73226 VOLUME 6, 2018

W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network

B. DERIVING THE OPTIMAL ρ AND G Algorithm 1 Proposed Algorithm for the Joint Optimization
Substituting the optimal P∗
into (11), and adopting some Problem
manipulation, (11) can be rewritten as (26) shown at page 6, 1: initialize non-negative values {α1 , α2 , α3 }.
in which 2: repeat
3: Calculate the optimal power allocation P ∗ =
1   1  
Ek,k 0 = ln 1+SNR1k,k 0 + ln 1+SNR2k,k 0 −α3 pr,k 0 (ps1,k , pIs2,k
I ∗ ∗ , pE∗ , pE∗ , p∗ ) defined in (21) to (25).
s1,k s2,k r,k 0
2 2 4: Obtaining the optimal subcarrier pairing ρ ∗
(27)
5: Obtaining the optimal subcarrier grouping G ∗ =
Only Ek,k 0 is related to ρ. Therefore, we can get the optimal {G∗1 , G∗2 },
ρ∗ by finding the maximize Ek,k 0 in order. The computational 6: Update {α1 , α2 , α3 } by subgradient algorithm with the
complexity of subcarrier pairing is O(K 2 ). After obtaining subgradients defined in (13) (14) and (15).
ρ∗, we can get 7: until {α1 , α2 α3 } converge.

K
X
Fk = ρk,k 0 Ek,k 0 −α3 ζ (pEs1,k |h1,k |2 + pEs2,k |h2,k |2 + σk2 )
k 0 =1
+ α1 (pEs1,k − pIs1,k ) + α2 (pEs2,k − pIs2,k )
Only Fk is related to G1 . We only need to find all the k who
makes Fk positive. Then, all these k form G∗1 , which can be
expressed as
X
G∗1 = arg max Fk (28)
G1
k∈G1

At last, the optimal G∗2 , subcarrier group utilized for EH,

can be easily determined by G∗2 = K − G∗1 . Obviously,
the involved complexity is O(K). So far, we have got all the
optimal variables with given dual variables, which involve
an overall computational complexity of O(3ν × K 2 × K ) =
O(3ν × K 3 ). We summarise the proposed subgradient-based
algorithm as Algorithm 1. FIGURE 3. Sum transmit rate versus d1 .

IV. SIMULATION RESULTS
In the simulation, we set the distance between two source
nodes to 8 m, and di , i ∈ {1, 2}, denotes the distance between
Si and relay, satisfying d1 + d2 = 8 m. The channel coeffi-
cients follow hi ∼ CN (0, di −m ), where m represents the path
loss index. Other simulation parameters are set as follows:
m = 2, K = 32, ζ = 1, Ps = Ps1 = Ps2 , σ 2 = −30 dBm.
In Fig. 3 and Fig. 4, the performance of our proposed
algorithm is compared with the following two benchmark
algorithms.

∂L 1 −pIs2,k γ2,k pr,k 0 γ1,k 0 γ1,k

= −α1 +
∂pIs1,k 2 (pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ1,k 0 + 1)(pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ1,k 0 + pIs2,k γ2,k pr,k 0 γ1,k 0 + 1)
1 pr,k 0 γ2,k 0 γ1,k (pIs2,k γ2,k + pr,k 0 γ2,k 0 + 1)
+ (18)
2 (pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ2,k 0 + 1)(pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ2,k 0 + pIs1,k γ1,k pr,k 0 γ2,k 0 + 1)
∂L 1 −pIs1,k γ1,k pr,k 0 γ2,k 0 γ2,k
= −α2 +
∂pIs2,k 2 (pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ2,k 0 + 1)(pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ2,k 0 + pIs1,k γ1,k pr,k 0 γ2,k 0 + 1)
1 pr,k 0 γ1,k 0 γ2,k (pIs1,k γ1,k + pr,k 0 γ1,k 0 + 1)
+ (19)
2 (pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ1,k 0 + 1)(pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ1,k 0 + pIs2,k γ2,k pr,k 0 γ1,k 0 + 1)
∂L 1 pIs2,k γ2,k γ1,k 0 (pIs1,k γ1,k + pIs2,k γ2,k + 1)
= −α3 +
∂pr,k 0 2 (pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ1,k 0 + 1)(pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ1,k 0 + pIs2,k γ2,k pr,k 0 γ1,k 0 + 1)
1 pIs1,k γ1,k γ2,k 0 (pIs1,k γ1,k + pIs2,k γ2,k + 1)
+ (20)
2 (pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ2,k 0 + 1)(pIs1,k γ1,k + pIs2,k γ2,k + pr,k 0 γ2,k 0 + pIs1,k γ1,k pr,k 0 γ2,k 0 + 1)

VOLUME 6, 2018 73227

 W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network

a4 = 2α1 γ1,k 3 (1 + pr,k 0 γ2,k 0 )

a3 = 2α1 γ1,k 2 ((1 + pr,k 0 γ2,k 0 )(2M (k) + N (k) + pIs2,k γ2,k pr,k 0 γ1,k 0 ) + N (k))
a2 = 2α1 γ1,k ((1 + pr,k 0 γ2,k 0 )(M (k) (M (k) + 2N (k)) + pIs2,k γ2,k pr,k 0 γ1,k 0 (M (k) + N (k)))
 
+ N (k) 2M (k) + N (k) + pIs2,k γ2,k pr,k 0 γ1,k 0 ) + γ1,k 2 (pIs2,k γ2,k pr,k 0 γ1,k 0 1 + pr,k 0 γ2,k 0 − pr,k 0 γ2,k 0 N (k))


 
a1 = 2α1 ( 1 + pr,k 0 γ2,k 0 M (k)N (k) M (k) + pIs2,k γ2,k pr,k 0 γ1,k 0


 
+ N (k) M (k) (M (k) + 2N (k)) + pIs2,k γ2,k pr,k 0 γ1,k 0 (M (k) + N (k)) )
 
+ γ1,k N (k)(pIs2,k γ2,k pr,k 0 γ1,k 0 (2 + pr,k 0 γ2,k 0 ) − pr,k 0 γ2,k 0 2M (k) + pIs2,k γ2,k pr,k 0 γ1,k 0 )
2α1  
N (k)2 M (k) M (k) + pIs2,k γ2,k pr,k 0 γ1,k 0 + pIs2,k γ2,k pr,k 0 γ1,k 0 N (k) N (k) − pr,k 0 γ2,k 0 M (k)


a0 =
γ1
− pr,k 0 γ2,k 0 M (k)2 N (k)
b4 = 2α2 γ2,k 3 (1 + pr,k 0 γ1,k 0 )
 
b3 = 2α2 γ2,k 2 ((1 + pr,k 0 γ1,k 0 ) 2V (k) + U (k) + pIs1,k γ1,k pr,k 0 γ2,k 0 + U (k))

 
b2 = 2α2 γ2,k ( 1 + pr,k 0 γ1,k 0 V (k) (V (k) + 2U (k)) + pIs1,k γ1,k pr,k 0 γ2,k 0 (U (k) + V (k))
 
+ U (k) 2V (k) + U (k) + pIs1,k γ1,k pr,k 0 γ2,k 0 ) + γ2,k 2 (pIs1,k γ1,k pr,k 0 γ2,k 0 1 + pr,k 0 γ1,k 0 − pr,k 0 γ1,k 0 U (k))



  
b1 = 2α2 (U (k)V (k) 1 + pr,k 0 γ1,k 0 V (k) + pIs1,k γ1,k pr,k 0 γ2,k 0 + (2U (k) + V (k))
 
+ pIs1,k γ1,k pr,k 0 γ2,k 0 U (k)(U (k) + V (k))) − γ2,k (pr,k 0 γ1,k 0 U (k) 2V (k) + pIs1,k γ1,k pr,k 0 γ2,k 0
− pIs1,k γ1,k pr,k 0 γ2,k 0 U (k)(2 + pr,k 0 γ1,k 0 ))
2α2  
U (k)2 V (k) V (k) + pIs1,k γ1,k pr,k 0 γ2,k 0 + pIs1,k γ1,k pr,k 0 γ2,k 0 U (k) U (k) − pr,k 0 γ1,k 0 V (k)


b0 =
γ2
− pr,k 0 γ1,k 0 U (k) V (k)2
2α3   
c4 = 1 + pIs1,k γ1,k 1 + pIs2,k γ2,k γ1,k 0 2 γ2,k 0 2
X (k)
  
c3 = 2α3 γ1,k 0 γ2,k 0 ( 1 + pIs1,k γ1,k 2 + pIs2,k γ2,k γ2,k 0 + (2 + pIs1,k γ1,k )(1 + pIs2,k γ2,k )γ1,k 0 )
  
c2 = 2α3 X (k) − pIs2,k γ2,k γ1,k 0 1 + pIs1,k γ1,k γ2,k 0 2 + (2α3 X (k) − pIs1,k γ1,k γ2,k 0 )(1 + pIs2,k γ2,k )γ1,k 0 2
 
+ 2α3 4 + pIs1,k γ1,k pIs2,k γ2,k + 2pIs1,k γ1,k + 2pIs2,k γ2,k X (k)γ1,k 0 γ2,k 0
    
c1 = 2α3 X (k) − pIs2,k γ2,k γ1,k 0 2 + pIs1,k γ1,k X (k)γ2,k 0 + 2α3 X (k) − pIs1,k γ1,k γ2,k 0 (2 + pIs2,k γ2,k )X(k)γ1,k 0
 
c0 = X (k)2 2α3 X (k) − pIs2,k γ2,k γ1,k 0 − pIs1,k γ1,k γ2,k 0

K
X 2 2
L(G, ρ) = α1 Ps1 + α2 Ps2 + [−α1 pEs1,k − α2 pEs2,k + α3 ζ (pEs1,k h1,k  + pEs2,k h2,k  + σk2 )]
 

k=1
X XK
+ { ρk,k 0 Ek,k 0 − α3 ζ (pEs1,k |h1,k |2 + pEs2,k |h2,k |2 + σk2 ) + α1 (pEs1,k − pIs1,k ) + α2 (pEs2,k − pIs2,k )} (26)
k∈G1 k 0 =1

• Power average allocation (PAA) algorithm: the power of
source and relay nodes are all average allocated while
subcarrier grouping and pairing are the same as our
proposed algorithm.
• Subcarrier allocation based (SAB) algorithm: certain
half of the subcarriers are allocated for ID, whereas the
rest of subcarriers are used for EH. The power allocation
and subcarriers pairing are the same as our proposed
algorithm.
In Fig. 3 we can see the variation of sum transmission
rate with d1 , when Ps is set to be 1mW . We can clearly
observe that our proposed algorithm outperforms the other

73228 VOLUME 6, 2018

W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network
FIGURE 4. Sum transmit rate versus Ps .
FIGURE 5. Power and subcarrier allocation of source node S1 .
two algorithms, because we do not optimize the power of
source and relay nodes in PAA algorithm, and in SAB algo-
rithm the subcarrier grouping is not optimized. In SAB algo-
rithm fixed number of subcarriers are allocated for ID and
EH, which will lead some power and subcarriers sacrificed
for EH. Also in Fig. 3, we can observe that all three algorithms
get the minimum rate when R locates in the middle of the S1
and S2 . It is due to that in this place, the interference caused
from the two source nodes to each other reaches the maximum
value.
Fig. 4 plots the sum transmission rate versus Ps when d1 =
3.2m. Fig. 4 shows that the performance of our algorithm is
better than PAA and SAB algorithms. Besides, we can see the
sum transmission rate becomes larger when Ps increases. It is
easy to understand that when Ps increases, more power can be
allocated for ID and EH, which makes the relay obtain more
power to forward the information of source nodes, resulting
larger sum transmission rate.
Fig. 5 and Fig. 6 illustrate the power allocation and subcar-
rier grouping of two source nodes in the first transmission
VOLUME 6, 2018
FIGURE 6. Power and subcarrier allocation of source node S2 .
FIGURE 7. Power allocation of relay node.
phase, respectively, when Ps = 1mW and d1 = 1.6 m.
We can find that the two source nodes S1 and S2 use the
same subcarrier group for ID and EH, respectively. However,
the power allocated over these subcarriers are different, which
is due to the different channel coefficients between S1 → R
and S2 → R.
Fig. 7 shows the power allocation of relay in the second
transmission phase. From the figure, we can see the number
of relay subcarriers used for forwarding the information is
the same as the number of subcarriers in G1 , which is used
by the source nodes for ID in the first transmission phase.
It is because that one subcarrier k, k ∈ G1 in the first
transmission phase could only be paired with one subcarrier
k 0 in the second transmission phase.
Fig. 8 illustrates power allocation proportion of ID and
EH at two source nodes varies with d1 . We use PIsi and PEsi
to denote the sum power used for ID and EH of Si . As we
can see, power used for EH becomes larger then smaller
when d1 becomes larger. It is because that when d1 is smaller
73229

FIGURE 8. Power allocation proportion used for ID and EH of two source
nodes versus d1 .
than 3.2m, the interference at relay becomes larger when
d1 increases, which lead relay need more power to forward
information. When d1 is larger than 3.2m, the interference
becomes smaller, less power is required for relaying infor-
mation.
V. CONCLUSION
In this paper, an OFDM based SWIPT scheme for an
AF-TWR network was proposed, where a wireless powered
relay uses two disjoint subcarriers groups to perform ID and
EH, respectively. Since only subcarriers index used for ID
and EH is needed by the relay, no splitter is needed to equip
at the relay, which significantly reduced the complexity of
the relay to harvest energy. Moreover, an efficient algorithm
is proposed to solve the joint subcarrier grouping, paring
and power optimization allocation problem for the proposed
OFDM SWIPT scheme, such that system sum transmission
rate is maximized under power constraint. Simulation analy-
ses show that our proposed algorithm has better performance
and provide the insight into the effect of relay location and
total transmission power on system performance.
APPENDIX
We assume that x and y are the optimal solutions for joint
optimization problem with relay energy constraints Qx and
Qy, individually. We denote f (.) as the objective function of
the joint optimization problem in (9), there will always be
a feasible solution z, satisfiying Qz ≥ θ Qx + (1 − θ )Qy and
f (z) ≥ θ f (x) + (1 − θ)f (y), for any 0 ≤ θ ≤ 1. Therefore,
our joint optimization problem meets the so-called ‘‘time-
sharing’’ condition.
We first need to prove that the objective function
of our joint optimization problem is a concave func-
tion of Q. To prove the time-sharing property means
zero duality gap, let Qx, Qy and Qz be the energy con-
straints with Qz ≥ θ Qx + (1 − θ)Qy for some 0 ≤ θ ≤ 1.
73230
W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network
Let x, y and z0 be the optimal solutions to the joint opti-
mization problem with energy constraints Qx, Qy and
Qz, respectively. Since Qz ≥ θ Qx + (1 − θ )Qy, the time-
sharing property implies that there exists a z0 such that
Qz0 ≥ θ Qx + (1 − θ )Qy and f (z0 ) ≥ θ f (x) + (1 − θ )f (y).
Because z0 is a feasible solution for the optimization problem,
f (z) ≥ f (z0 ) ≥ θ Qx + (1 − θ )Qy, which proves that our joint
optimization problem is a concave function of Q. Therefore
the duality gap is zero can be proved by the similar methods
used in [34].
REFERENCES
[1] X. Lu, P. Wang, D. Niyato, D. I. Kim, and Z. Han, ‘‘Wireless networks with
RF energy harvesting: A contemporary survey,’’ IEEE Commun. Surveys
Tuts., vol. 17, no. 2, pp. 757–789, 2nd Quart., 2015.
[2] O. Ozel, K. Tutuncuoglu, S. Ulukus, and A. Yener, ‘‘Fundamental limits of
energy harvesting communications,’’ IEEE Commun. Mag., vol. 53, no. 4,
pp. 126–132, Apr. 2015.
[3] L. R. Varshney, ‘‘Transporting information and energy simultaneously,’’ in
Proc. IEEE Int. Symp. Inf. Theory, Jul. 2008, pp. 1612–1616.
[4] P. Grover and A. Sahai, ‘‘Shannon meets Tesla: Wireless information and
power transfer,’’ in Proc. IEEE Int. Symp. Inf. Theory, Austin, TX, USA,
Jun. 2010, pp. 2363–2367.
[5] A. A. Nasir, X. Zhou, S. Durrani, and R. A. Kennedy, ‘‘Relaying protocols
for wireless energy harvesting and information processing,’’ IEEE Trans.
Wireless Commun., vol. 12, no. 7, pp. 3622–3636, Jul. 2013.
[6] Y. Gu and S. Aïssa, ‘‘RF-based energy harvesting in decode-and-forward
relaying systems: Ergodic and outage capacities,’’ IEEE Trans. Wireless
Commun., vol. 14, no. 11, pp. 6425–6434, Nov. 2015.
[7] N. Zhao, F. Hu, Z. Li, and Y. Gao, ‘‘Simultaneous wireless information and
power transfer strategies in relaying network with direct link to maximize
throughput,’’ IEEE Trans. Veh. Technol., vol. 67, no. 9, pp. 8514–8524,
Sep. 2018.
[8] W. Lu, Y. Gong, X. Liu, J. Wu, and H. Peng, ‘‘Collaborative energy and
information transfer in green wireless sensor networks for smart cities,’’
IEEE Trans. Ind. Informat., vol. 14, no. 4, pp. 1585–1593, Apr. 2018.
[9] Q. Shi, C. Peng, W. Xu, M. Hong, and Y. Cai, ‘‘Energy efficiency
optimization for MISO SWIPT systems with zero-forcing beamform-
ing,’’ IEEE Trans. Signal Process., vol. 64, no. 4, pp. 842–854,
Feb. 2016.
[10] M. Ju, K.-M. Kang, K.-S. Hwang, and C. Jeong, ‘‘Maximum trans-
mission rate of PSR/TSR protocols in wireless energy harvesting
DF-based relay networks,’’ IEEE J. Sel. Areas Commun., vol. 33, no. 12,
pp. 2701–2717, Dec. 2015.
[11] C. Zhang and Y. Chen, ‘‘Wireless power transfer strategies for cooperative
relay system to maximize information throughput,’’ IEEE Access, vol. 5,
pp. 2573–2582, 2017.
[12] Y. Chen and C. Zhang, ‘‘Optimal time allocation for wireless powered
relay systems with joint S-D energy transfer,’’ in Proc. IEEE Veh. Technol.
Conf. (VTC Spring), Nanjing, China, May 2016, pp. 1–5.
[13] G. Zhu, C. Zhong, H. A. Suraweera, G. K. Karagiannidis, Z. Zhang, and
T. A. Tsiftsis, ‘‘Wireless information and power transfer in relay systems
with multiple antennas and interference,’’ IEEE Trans. Commun., vol. 63,
no. 4, pp. 1400–1418, Apr. 2015.
[14] A. A. Nasir, X. Zhou, S. Durrani, and R. A. Kennedy, ‘‘Wireless-powered
relays in cooperative communications: Time-switching relaying proto-
cols and throughput analysis,’’ IEEE Trans. Commun., vol. 63, no. 5,
pp. 1607–1622, May 2015.
[15] L. Mohjazi, S. Muhaidat, M. Dianati, and M. Al-Qutayri, ‘‘Outage prob-
ability and throughput of SWIPT relay networks with differential mod-
ulation,’’ Proc. IEEE 86th Veh. Technol. Conf. (VTC-Fall), Toronto, ON,
Canada, Sep. 2017, pp. 1–6.
[16] Y. Liu and X. Wang, ‘‘Information and energy cooperation in OFDM
relaying: Protocols and optimization,’’ IEEE Trans. Veh. Technol., vol. 65,
no. 7, pp. 5088–5098, Jul. 2016.
[17] K. Xiong, P. Fan, C. Zhang, and K. B. Letaief, ‘‘Wireless information
and energy transfer for two-hop non-regenerative MIMO-OFDM relay
networks,’’ IEEE J. Sel. Areas Commun., vol. 33, no. 8, pp. 1595–1611,
Aug. 2015.
VOLUME 6, 2018
W. Lu et al.: OFDM-Based SWIPT for Two-Way AF Relaying Network
[18] C. Zhong, H. A. Suraweera, G. Zheng, I. Krikidis, and Z. Zhang, ‘‘Wireless
information and power transfer with full duplex relaying,’’ IEEE Trans.
Commun., vol. 62, no. 10, pp. 3447–3461, Oct. 2014.
[19] T. P. Do, I. Song, and Y. H. Kim, ‘‘Simultaneous wireless trans-
fer of power and information in a decode-and-forward two-way
relaying network,’’ IEEE Trans. Wireless Commun., vol. 16, no. 3,
pp. 1579–1592, Mar. 2017.
[20] W. Lu, Y. A. Zhang, M. Wang, X. Liu, and J. Hua, ‘‘Cooperative spectrum
sharing in OFDM two-way relay systems with bidirectional transmis-
sions,’’ IEEE Commun. Lett., vol. 21, no. 6, pp. 1349–1352, Jun. 2017.
[21] X. Zhou and Q. Li, ‘‘Energy efficiency optimisation for SWIPT AF
two-way relay networks,’’ Electron. Lett., vol. 53, no. 6, pp. 436–438,
Mar. 2017.
[22] S. Salari, I.-M. Kim, D. I. Kim, and F. Chan, ‘‘Joint EH time allocation
and distributed beamforming in interference-limited two-way networks
with EH-based relays,’’ IEEE Trans. Wireless Commun., vol. 16, no. 10,
pp. 6395–6408, Oct. 2017.
[23] L. Wang, F. Hu, Z. Ling, and B. Wang, ‘‘Wireless information and power
transfer to maximize information throughput in WBAN,’’ IEEE Internet
Things J., vol. 4, no. 5, pp. 1663–1670, Oct. 2017.
[24] Y. Liu, L. Wang, M. Elkashlan, T. Q. Duong, and A. Nallanathan,
‘‘Two-way relaying networks with wireless power transfer: Policies
design and throughput analysis,’’ in Proc. IEEE Global Commun.
Conf. (GLOBECOM), Dec. 2014, pp. 4030–4035.
[25] S. T. Shah, D. Munir, M. Y. Chung, and K. W. Choi, ‘‘Information process-
ing and wireless energy harvesting in two-way amplify-and-forward relay
networks,’’ in Proc. VTC Spring, May 2016, pp. 1–5.
[26] Y. Ye, Y. Li, Z. Wang, X. Chu, and H. Zhang, ‘‘Dynamic asymmetric power
splitting scheme for SWIPT-based two-way multiplicative AF relaying,’’
IEEE Signal Process. Lett., vol. 25, no. 7, pp. 1014–1018, Jul. 2018.
[27] J. Men, J. Ge, C. Zhang, and J. Li, ‘‘Joint optimal power allocation and
relay selection scheme in energy harvesting asymmetric two-way relaying
system,’’ IET Commun., vol. 9, no. 11, pp. 1421–1426, Jul. 2015.
[28] C. Zhang, H. Du, and J. Ge, ‘‘Energy-efficient power allocation in
energy harvesting two-way AF relay systems,’’ IEEE Access, vol. 5,
pp. 3640–3645, 2017.
[29] N. T. P. Van, S. F. Hasan, X. Gui, S. Mukhopadhyay, and H. Tran, ‘‘Three-
step two-way decode and forward relay with energy harvesting,’’ IEEE
Commun. Lett., vol. 21, no. 4, pp. 857–860, Apr. 2017.
[30] K. Tutuncuoglu, B. Varan, and A. Yener, ‘‘Throughput maximization for
two-way relay channels with energy harvesting nodes: The impact
of relaying strategies,’’ IEEE Trans. Commun., vol. 63, no. 6,
pp. 2081–2093, Jun. 2015.
[31] W. Lu and J. Wang, ‘‘Opportunistic spectrum sharing based on full-
duplex cooperative OFDM relaying,’’ IEEE Commun. Lett., vol. 18, no. 2,
pp. 241–244, Feb. 2014.
[32] L. Zhang, A. Ijaz, P. Xiao, M. M. Molu, and R. Tafazolli, ‘‘Filtered OFDM
systems, algorithms, and performance analysis for 5G and beyond,’’ IEEE
Trans. Commun., vol. 66, no. 3, pp. 1205–1208, Mar. 2018.
[33] P. Guan et al., ‘‘5G field trials: OFDM-based waveforms and
mixed numerologies,’’ IEEE J. Sel. Areas Commun., vol. 35, no. 6,
pp. 1234–1243, Jun. 2017.
[34] W. Yu and R. Lui, ‘‘Dual methods for nonconvex spectrum optimiza-
tion of multicarrier systems,’’ IEEE Trans. Commun., vol. 54, no. 7,
pp. 1310–1322, Jul. 2006.
[35] W. Dang, M. Tao, H. Mu, and J. Huang, ‘‘Subcarrier-pair based resource
allocation for cooperative multi-relay OFDM systems,’’ IEEE Trans. Wire-
less Commun., vol. 9, no. 5, pp. 1640–1649, May 2010.
LU WEIDANG (S’08–M’13) received the Ph.D.
degree in information and communication engi-
neering from the Harbin Institute of Technol-
ogy in 2012. He was a Visiting Scholar with
Nanyang Technology University, Singapore, The
Chinese University of Hong Kong, China, and the
Southern University of Science and Technology,
China. He is currently an Associate Professor with
the College of Information Engineering, Zhejiang
University of Technology, Hangzhou, China. His
current research interests include simultaneous wireless information and
power transfer, wireless sensor networks, cooperative communications, and
physical layer security for wireless systems.
VOLUME 6, 2018
WEILIN ZHAO is currently pursuing the master’s
degree at the Zhejiang University of Technology,
Hangzhou, China. Her current research interests
include simultaneous wireless information and
power transfer, two-way cooperative relaying, and
WSN.
SU HU (S’07–M’11) received the M.S. and Ph.D.
degrees from the National Key Lab on Com-
munications, University of Electronic Science
and Technology of China (UESTC) in 2007 and
2010, respectively. From 2011 to 2012, he was a
Research Fellow with the School of Electrical and
Electronic Engineering, Nanyang Technological
University, Singapore. He is currently a Full Pro-
fessor with UESTC. His research interests include
sequence design with good correlation properties
and physical layer design for wireless communication system, such as filter
bank multicarrier systems and cognitive radio networks.
XIN LIU (M’13) received the M.Sc. and Ph.D.
degrees in communication engineering from the
Harbin Institute of Technology in 2008 and
2012, respectively. He is currently an Associate
Professor with the School of Information and
Communication Engineering, Dalian University of
Technology, China. From 2012 to 2013, he was a
Research Fellow with the School of Electrical and
Electronic Engineering, Nanyang Technological
University, Singapore. From 2013 to 2016, he was
a Lecturer with the College of Astronautics, Nanjing University of Aeronau-
tics and Astronautics, China. His research interests focus on communication
signal processing, wireless sensor networks, spectrum resource allocation,
and broadband satellite communications.
BO LI received the bachelor’s degree in commu-
nication engineering and the master’s and Ph.D.
degrees in information and communication engi-
neering from the Harbin Institute of Technology,
China, in 2007, 2009, and 2013, respectively.
He was a Visiting Ph.D. Student with the School
of EEE, Nanyang Technological University, Sin-
gapore, from 2012 to 2013. Since 2013, he has
been with the School of Information and Electrical
Engineering, Harbin Institute of Technology, Wei-
hai, China, as a Lecturer. His interested research areas are physical-layer
network coding, mobile ad hoc networks, and adaptive modulation and
coding.
YI GONG (S’99–M’03–SM’07) received the
Ph.D. degree in electrical engineering from The
Hong Kong University of Science and Technol-
ogy, Hong Kong, in 2002. He then joined the
Hong Kong Applied Science and Technology
Research Institute, Hong Kong, as a Member of
Professional Staff. He was with Nanyang Tech-
nological University, Singapore, with which he
still has active collaborations. He is currently a
Professor with the Southern University of Science
and Technology, Shenzhen, China. Since 2006, he has served on the Edi-
torial Board for the IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS and
the IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY. His research interests
include multi-antenna and cooperative communications, caching and edge
computing, and physical layer security for wireless systems.
73231