Energy Splitting for SWIPT in QoS-constraint MTC Network: A Non-Cooperative Game Theoretic Approach

Energy Splitting for SWIPT in QoS-constraint MTC
Network: A Non-Cooperative Game Theoretic Approach
Kang Kang, Zhenni Pan, Jiang Liu, Shigeru Shimamoto
This paper studies the emerging wireless energy harvesting algorithm dedicated for machine type com-
munication (MTC) in a typical cellular network where one transmitter (e.g. the base station, a hybrid
access point) with constant power supply communicates with a set of users (e.g. wearable devices,
sensors). In the downlink direction, the information transmission and power transfer are conducted si-
multaneously by the base station. Since MTC only transmits several bits control signal in the downlink
direction, the received signal power can be split into two parts at the receiver side. One is used for
information decoding and the other part is used for energy harvesting. Since we assume that the users
are without power supply or battery, the uplink transmission power is totally from the energy harvest-
ing. Then, the users are able to transmit their measured or collected data to the base station in the
uplink direction. Game theory is used in this paper to exploit the optimal ratio for energy harvesting
of each user since power splitting scheme is adopted. The results show that this proposed algorithm is
capable of modifying dynamically to achieve the prescribed target downlink decoding signal-to-noise
plus interference ratio (SINR) which ensures the high reliability of MTC while maximizing the uplink
throughput.
KEYWORDS
Energy harvesting, decoding SINR, uplink throughput maximization
1. INTRODUCTION
Recent years, wireless power transfer is emerging as a possible solution to address the power supply
issue in wireless communications. Conventionally, terminals in wireless networks are powered by fixed
energy supplies, e.g. power lines and batteries. But for most mobile terminals and remote deployed
terminals (e.g. temperature sensor and wind speed monitor), they are mainly supplied by batteries,
which extremely limits the lifetime [1]. However, sometimes because of the unreachable deployment
location, it is impossible or extremely difficult to replace or charge the battery. Thus, wireless energy
transfer is considered as an alternative solution to extend the working time of the network by transfer
unlimited power via a wireless method.
1.1. Related Works
In [2], Wireless power transfer techniques are summarized into three categories: RF energy transfer,
resonant inductive coupling and magnetic resonance coupling. Compared with the other two coupling
related techniques, RF energy transfer outperforms in terms of effective distance. Note that in this paper,
DOI: 10.5121/ 105
International Journal of Computer Networks & Communications (IJCNC) Vol.10, No.5, September 2018
Graduate School of Fundamental Science and Engineering, Waseda University, Japan
ABSTRACT
ijcnc.2018.10506

we only focus on this RF energy transfer technique. In [3], an orthogonal frequency division multiple
access (OFDMA) system with simultaneous wireless information and power transfer is considered.
The tradeoff between energy efficiency, system capacity, and wireless power transfer are achieved by
developing suboptimal iterative resource allocation algorithms. In [4], the authors consider both delay
and mobility with wireless energy harvesting. Three issues including optimal transmission policy for
the mobile node, energy management strategy and deployment of wireless power sources are solved. In
[5], a point-to-point wireless link over the narrowband flat-fading channel with time-varying co-channel
interference is considered. Time switcher is used at the receiver side to switch between information
decoding and energy harvesting based on the instantaneous channel and interference condition. Various
trades-offs between wireless information transfer and energy harvesting are achieved.
Additionally, multiple-input multiple-output (MIMO) is widely used in wireless energy transfer to
improve the performance of wireless energy transfer. In [6], an efficient channel acquisition method
for a point-to-point MIMO wireless energy transfer system is designed by exploiting the channel reci-
procity. The dedicated reverse-link training from the energy receiver is used by the energy transmitter
to estimate the CSI. In [7], wireless energy transfer with MIMO is considered and a general design
framework for a new type of channel learning method based on the energy receivers energy measure-
ment feedback is proposed. In [8], the frame is divided into downlink for wireless energy transfer and
uplink for wireless information transmission and the optimal time allocation for downlink and uplink is
obtained.
Game theory has been widely used as an effective approach to solving resource allocation problems.
A Multiservice Uplink Power Control game (MSUPC) is formulated in [9] and the cost function of each
user is designed to maximize its own utility. Besides, some optimal power allocation schemes using
game theory are proposed in [10], [11] and [12] by considering different network scenarios. In addition,
game theory is also used to handle joint resource allocation problems. [13] considers both utility-based
uplink transmission power and rate allocation in a non-cooperative game model with pricing. [14]
proposes a utility function representing its perceived satisfaction with respect to its allocated power and
rate.
1.2. Contributions
In this paper, game theory is utilized to derive the balancing ratio for energy harvesting out of the total
received power for future MTC networks. Most of the existing works discuss optimal time alloca-
tion, throughput maximization and various trades-offs between energy efficiency, capacity and wireless
power transfer. Different from these existing papers, our main contributions are listed below.
• We first use game theory to address wireless energy transfer problem. Game theory enables us
to obtain a balancing solution under various constraints. In this paper, in order to ensure the
downlink received decoding SINR and the uplink throughput of each user, a non- cooperative
game model is formed to seek a balancing solution between energy for decoding and energy for
harvesting.
• We first design a wireless energy transfer algorithm dedicated for MTC networks. Different from
common mobile communications, the base station broadcasts control signal, which is only sev-
eral bits size, to the machines in the downlink direction while in the uplink direction the machines
transmit large amount data to the base station periodically. So these require an extremely high de-
coding SINR level in the downlink direction to ensure the control signal is decoded correctly and
a maximum throughput in the uplink direction. Based on these requirements, a target decoding
SINR is set, upon which the uplink throughput is maximized.
106

Figure 1: System model
• Our solution indicates that in order to ensure the high reliability of MTC, a cell edge machine
with poor channel condition utilizes the majority of received energy to do harvesting while a cell
center machine with good channel condition utilizes majority of received energy to do informa-
tion decoding. As a result, the uplink throughput of a cell edge machine is lower than that of a
cell centre machine since the harvest-then-transmit protocol is used here. Additionally, the value
of target decoding SINR aﬀects the performance of this system as well. When the target decoding
SINR increases, the uplink throughput decreases so that more received energy can be used for
decoding.
The rest of this paper is organized as follows: Section 2 shows the system model of this wireless energy
transfer MTC networks. In Section 3, the wireless energy transfer algorithm is derived based on game
theory. The simulation results are presented in Section 4. Finally, this paper is concluded in Section 5.
2. SYSTEM MODEL
As shown in Fig. 1, this paper considers a wireless cellular network with mixed wireless informa-
tion transmission and wireless energy transfer in the downlink direction and pure wireless information
transmission in the uplink direction. The system consists of one base station and N devices denoted by
MTCDi, i = 1, · · · , N. It is assumed that the base station and all devices operate on the same frequency
band. Diﬀerent from those MIMO systems, we assume that one single antenna is equipped at the base
station and all devices. We further assume that all devices are without batteries or power supplies.
Therefore, the devices have to harvest energy from the received signals transmitted by the base station
in the downlink direction, upon which the devices use the harvested energy to transmit data to the base
station in the uplink direction. The downlink and uplink channel gains of ith device are denoted as GDL
i
107

Figure 2: Energy harvesting frame
and GUL
i , respectively. It is assumed that the downlink and uplink channels are Rayleigh fading. Since
no battery embedded in the devices, the harvest-then-transmit protocol is used in this network. On the
device side, power splitter is connected to two functionality units: energy harvester and information
decoder as shown in Fig. 2. The power splitter splits the received power into two parts: θ × 100%
received power for information decoding.
It is known that MTC networks require high reliability. Therefore, the downlink decoding SINR
γd
i is crucial because the downlink signal contains control message which may cause misoperation if
decoding fails. Thus, a target decoding SINR, which ensures the high reliability, is prescribed as γtar
i
which should satisfy
γd
i ≥ γtar
i (1)
where subscript i indicates the ith device. Since (1 − θi) × 100% of the received power is used to decode
the downlink signal, the downlink decoding SINR of ith device is written as
γd
i =
PtGDL
i (1 − θi)
N0
(2)
where Pt is transmission power of base station and N0 is noise. And θi ×% of the received power is used
for harvesting so after this harvesting, the power used for uplink transmission is PUL
i = ηiPtGDL
i θiGUL
i
where ηi represents the harvesting eﬃciency which is set to be a constant value in this paper. The
corresponding uplink SINR is shown as
γUL
i =
ηiPtGDL
i θiGUL
i
Ii + N0
(3)
where Ii is the interference from other devices and it is written as Ii = j i ηjPtGDL
j θjGUL
j . From (2)
and (3), we can see that if θi increases, the downlink decoding SINR γd
i decreases and uplink SINR γUL
i
increases. And vice versa. Thus, there must be a θi balancing this conﬂict.
3. GAME PROBLEM FORMULATION
In this section, we formulate the energy harvesting problem using non-cooperative game model, upon
which the energy harvesting ratio iteration function is derived. In addition, we prove the existence of
Nash equilibrium of this algorithm as well as the convergence of this iteration function with certain
constraints.
3.1. Utility Function and Nash Equilibrium Derivation
A game model consists of three essential elements: player, strategy and utility function. In this paper,
each device is regarded as a player participating in this game. The energy harvesting ratio θi is assumed
108

to be the strategy of ith device and the utility function of ith device is defined as Ui(θi, Ii(Θi)), where
Θi := [θ1, · · · , θi−1, θi+1, · · · θN]. Note that when the Nash equilibrium is achieved, the utility function
is presented as Ui(θ∗
i , Ii(Θ∗
i )). The definition of Nash equilibrium indicates that no one can increase its
utility by deviating from the Nash equilibrium solution θ∗
i . Thus, θ∗
i satisfies the following equation
Ui(θ∗
i , Ii(Θ∗
i )) ≥ Ui(θi, Ii(Θ∗
i )) (4)
∀θi, ∀i = 1, 2, · · · , N
As we illustrated above, there exists a conflict between downlink decoding SINR and uplink throughput.
So, the objective of the proposed algorithm in this paper is to find out a balancing energy harvesting
ratio θi for each device so that the uplink throughput of each device is maximized while the down-
link decoding SINR of each device achieves the prescribed target SINR which ensures the reliability.
Therefore, we take downlink decoding SINR γd
i , prescribed target SINR γtar
i and uplink throughput
log(1 + γUL
i ) into consideration and propose the following utility function:
Ui(θi, Ii(Θi)) = ai(γd
i − γtar
i )2
+ bi(log(1 + γUL
i )) (5)
where ai and bi is two non-negative coefficients. The value of this prescribed target decoding SINR
γtar
i depends on reliability level, e.g. high level for military and industrial applications, low level for
civil and home applications. Then the first order partial derivative of this utility function is derived as
∂Ui
∂θi
= 2ai(γd
i − γtar
i )
∂γd
i
∂θi
+
bi
ln2
1
1 + γUL
i
∂γUL
i
∂θi
= −2ai
PtGDL
i
N0
(
PtGDL
i (1 − θi)
N0
) +
bi
ln2
1
θi + Ii+N0
ηiPtGDL
i GUL
i
(6)
then we let ∂Ui
∂θi
= 0 and obtain
γd
i = γtar
i +
bi
2ailn2
N0
PiGDL
i θi + Ii+N0
ηiGUL
i
(7)
it is clear that when bi = 0, γd
i = γtar
i holds. This indicates that if the uplink throughput is not taken
into consideration or we put no emphasize on uplink throughput, the solution is achieved by γd
i = γtar
i .
Recalling (2), we obtain
θi = 1 −
N0
PtGDL
i
γtar
i −
bi
2ailn2
N2
0
(PiGDL
i )2θi +
PiGDL
i (Ii+N0)
ηiGUL
i
(8)
Here we define
PiGDL
i
N0
= γDL
i and associated with (3), we can get
γDL
i
γUL
i
=
Ii + N0
ηiGUL
i N0θi
(9)
So that (8) can be written as
θi = 1 −
γtar
i
γDL
i
−
bi
2ailn2
1
(γDL
i )2θi(1 + 1
γUL
i
)
(10)
It can be seen that θi = f(γDL
i , γUL
i ) with a given γtar
i .
109

3.2. Energy Harvesting Algorithm
Providing that the interference and noise can be measured at the receiver side, we can get
θ(k+1)
i = 1 −
γtar
i
γDL
i
−
bi
2ailn2
1
(γDL
i )2θ(k)
i (1 + 1
(γUL
i )(k) )
(11)
where θ(k)
i represents the energy harvesting ratio of the ith device at the kth iteration, and (γUL
i )(k)
represents the SINR of the ith device at the kth iteration which includes the I(k)
i experienced by the ith
device at the kth iteration. Note that if and only if θ(k)
i > 0, this algorithm makes sense. Otherwise, there
is no received energy for harvesting at all. In order to analyze the convergence, we deﬁne the iteration
function as θ(k+1)
i = f(k)
i (θ(k)
i ) and rewrite it as
f(k)
i (θ(k)
i ) := θ(k+1)
i = 1 −
γtar
i
γDL
i
−
bi
2ailn2
1
(γDL
i )2θ(k)
i (1 + 1
(γUL
i )(k) )
(12)
So far, we have obtained the energy harvesting control iteration function and the convergence is
proved in the next subsection.
3.3. Convergence
For an iterative algorithm, it needs to achieve a stable value after some iterations. In [15], the authors
show that when convergence is achieved, three properties need to be satisﬁed. The proof of convergence
is showed as follows
• Positivity: f(θ) > 0
Since θ represents the energy harvesting ratio, it needs to follow a more strict constraint, 0 < θ <
1. By putting (10) into this constraint, we obtain
0 <
γtar
i
γDL
i
+
bi
2ailn2
1
(γDL
i )2θ(k)
i (1 + 1
(γUL
i )(k) )
< 1 (13)
in order to satisfy this inequality, the ratio value between ai and bi need to be
−γtar
i θi

γDL
i +
γDL
i
γUL
i

 <
bi
2ailn2
< γDL
i − γtar
i θi

γDL
i +
γDL
i
γUL
i

 (14)
Since ai and bi are both positive, (14) can be rewritten as
0 <
bi
ai
< 2ln2 γDL
i − γtar
i θi

γDL
i +
γDL
i
γUL
i

 (15)
This constraint indicates that bi
ai
has to be carefully designed and it need to be a very small value
so that this positivity is guaranteed.
• Monotonicity: θ > θ
′
→ f(θ) > f(θ
′
)
In order to prove this monotonicity, we have to prove f(θi) is a monotonically increasing function.
So
∂ f(θi)
∂θi
=
bi
2ailn2 θiγDL
i
2
1 + 1
γUL
i
(16)
Since bi
ai
> 0, ∂f(θi)
∂θi
> 0 holds.
110

• Scalability: f(αθ) < α f(θ), ∀α > 1 In order to determine the scalability, we use (10) and obtain
f(αθi) − α f(θi) = (α − 1)


γtar
i
γDK
i
+
bi
2ailn2(γDL
i )2 1 + 1
γUL
i
α + 1
αθi
− 1


(17)
It is clear that the ﬁrst term α − 1 > 0. Thus we need to prove the latter term is negative so that
f(αθ) < α f(θ), ∀α > 1 is ensured. So we need to solve this following inequality
bi
2ailn2(γDL
i )2 1 + 1
γUL
i
α + 1
αθi
< 1 −
γtar
i
γDK
i
(18)
After several mathematical derivations, the solution of this inequality is shown as
bi
ai
< ln2 γDL
i − γtar
i θi

γDL
i +
γDL
i
γUL
i

 (19)
So far, in order to satisfy the convergence requirements, we have discussed three properties and derive
constraints of bi
ai
. Combined with (15) and (19), we limit bi
ai
as
0 <
bi
ai
< ln2 γDL
i − γtar
i θi

γDL
i +
γDL
i
γUL
i

 (20)
3.4. Existence of Nash Equilibrium
In this section, the Implicit Function Theorem [14] is used to prove the existence of a unique solution
for this iterative function. According to the Implicit Function Theorem, a new function is established
by moving the left-hand side term of (8) to the right-hand side and obtain
Fi(θi, Ii,GDL
i ,GUL
i , ai, bi, N0) = 1 − θi −
N0
PtGDL
i
γtar
i −
bi
2ailn2
N2
0
(PiGDL
i )2θi +
PiGDL
i (Ii+N0)
ηiGUL
i
(21)
According to Implicit Function Theorem, if there exists an unique Nash Equilibrium, the Jacobian
matrix must be non-singular. Then, we obtain
∂Fi(θi, Ii,GDL
i ,GUL
i , ai, bi, N0)
∂θk
=
∂F1
∂θ1
∂F1
∂θ2
· · · ∂F1
∂θN
∂F2
∂θ1
∂F2
∂θ2
· · · ∂F2
∂θN
...
...
...
...
∂FN
∂θ1
∂FN
∂θ2
· · · ∂FN
∂θN
=
A1 B12 · · · B1N
B21 A2 · · · B2N
...
...
...
...
BN1 BN2 · · · AN
(22)
where
Ai = −1 +
bi
2ailn2
1
PtGDL
i
N0
θi + Ii+N0
ηiGUL
i N0
(23)
and
Bij =
bi
2ailn2
ηjP2
t GDL
i GDL
j GUL
j
PtGDL
i
N0
2
+
PtGDL
i (Ii+N0)
ηiGUL
i N2
0
2
(24)
111

It is clear that the value of this matrix is determined by each element. In order to make this matrix
non-singular, the ratio value bi
ai
should be carefully defined. Otherwise, this matrix cannot be ensured to
be non-singular. So far, we have proved the convergence as well as the existence of the unique solution
of this energy harvesting iterative algorithm under constraint of bi
ai
.
1 2 3 4 5 6 7
Iteration
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
θi
Figure 3: Convergence of energy harvesting ratios
4. NUMERICAL RESULTS
In this section, the simulation results of this proposed algorithm are presented in Matlab. We con-
sider a cell with one base station communicating with N randomly located MTC devices denoted
byMTCDi, i = 1, · · · , N. The channel that every device experiences is defined as quasi-static flat-
fading. The available bandwidth is 20MHz and the transmission power at the base station, which is
denoted as Pt, is 30dBm. The energy harvesting efficiency ηi is set to be 0.8. The ratio value of bi
ai
is set
to be 0.025, which is small enough to ensure the convergence and existence of the unique Nash solution
of this iterative algorithm.
Fig. 3 shows the convergence of θi with N = 50. It is the core parameter of this iterative algorithm
since it directly determines how much received energy is used for energy harvesting. The initial θ(0)
i
is set to be 0.001. It is very clear that after 3 iterations, the θiof each device achieves a stable state.
Note that when reaching values within 0.01% of the steady state values the convergence of the iterative
algorithm is obtained. There are 50 lines representing the 50 devices, respectively. θ varies from 0 to
1 because each device experiences a different channel so that each individual device needs different
energy to reach the prescribed target SINR.
Fig. 4 shows the downlink decoding SINR gap between the average downlink decoding SINR and
target decoding SINR with different number of devices (N) in terms of different target decoding SINR.
112

10 20 30 40 50 60 70 80 90 100
Number of devices (N)
0
0.05
0.1
0.15
0.2
0.25
0.3
E(γi
d
)-γtar
(dB)
Target decoding SINR = 0 dB
Target decoding SINR = -2 dB
Figure 4: Downlink decoding SINR gap with different number of devices in terms of different target
decoding SINR
In MTC networks, the control information is carried by the downlink transmission. And at the receiver
side, the device has to successfully decodes this control information so that functionality of the device
can be conducted correctly. This requires a high reliability for the decoding part. Here we define a
target decoding SINR denoted as γtar
i . We can see that all these six lines are positive regardless of the
target decoding SINR. In another word, this proposed algorithm is capable of dynamically modifying
the energy for decoding so that the reliability requirement can be satisfied. With the increase of target
decoding SINR, the SINR gap is becoming narrow. This is because compared with a high target decod-
ing SINR, a lower target decoding SINR is much easier to achieve. In this paper, the downlink
received energy at the UE side is split into two parts. θi × 100% of received energy is used for energy
harvesting and (1−θi)×100% is used for information decoding. So θi actually represents the amount of
energy for harvesting. Further more, the value of θi is affected by its corresponding channel gain, which
is shown in Fig. 5. Note that stable theta means θi achieves its stable state after several iterations. It is
clear that with the improvement of channel gain, the stable θi increases regardless of the different target
decoding SINR. The reason is that when the channel is good, the device is capable of reaching the target
decoding SINR with less energy (low (1 − θi)). Therefore, more energy (high θi) can be used for energy
harvesting. On contrary, when channel is poor, the device has to assume more energy (high (1 − θi)) to
do information decoding so that the target decoding SINR is ensured. In this case, less energy (low θi)
is used for energy harvesting. In addition, target decoding SINR also make an influence on θi. As we
can see that when channel gain is fixed, a higher θi is obtained with a lower target decoding SINR. It
is straightforward that reaching a low target decoding SINR always assumes less energy (low (1 − θi))
compared with reaching a high target decoding SINR.
In this proposed algorithm, θi not only determines the energy splitting for both information decoding
113

0 2 4 6 8 10 12
Channel Gain (dB)
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
EnergyHarvestingRatioθ
Target SINR = -1 dB
Target SINR = -3 dB
Target SINR = -5 dB
Target SINR = -7 dB
Target SINR = -9 dB
Figure 5: Achievable stable θi under different channel gain in terms of target decoding SINR
and energy harvesting but also affects the uplink throughput of each device. Fig. 6 shows the effects
that θi makes on throughput. As it is discussed before, θi actually represents the amount of energy for
harvesting. And all of this harvested energy is used for uplink transmission since there is no battery
for energy storage. So it is clear that with the increase ofθi , the throughput increases as well. It is
worth noting thatθi varies in different regions under different target decoding SINRs. This is because
a device experiencing a poor channel with a high target decoding SINR has to utilize more received
energy to reach this target decoding SINR compared with a device experiencing a poor channel with a
low target decoding SINR. This explains the starting points locations of these five lines. Additionally,
for the devices with good channels, regardless of the target decoding SINRs, all these five lines achieve
similar throughput. This is because for a device with a good channel, the target decoding SINR can be
easily achieved by consuming little-received energy. As a result, most of the received energy is used
for uplink transmission. Therefore, they are capable of achieving the same throughput. This explains
the endpoints locations of these five lines.
5. CONCLUSION
In this paper, an energy harvesting control algorithm is proposed for future wireless powered MTC
networks by involving a non-cooperative game model. This algorithm takes both downlink and uplink
transmission into consideration. In the downlink direction, the device splits the received energy into
two parts: the information decoding part and the energy harvesting part. Because of the harvest-then-
transmit protocol, the harvested energy is in turn used for uplink transmission. Since the high reliability
of MTC has to be satisfied with the highest priority, we firstly ensure the downlink decoding SINR
achieve the prescribed target decoding SINR, and then we improve the uplink throughput as much as
114

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Theta
10-2
10-1
100
101
102
Throughput(Mbps)
Figure 6: Individual throughput under stable θi in terms of different target decoding SINR
possible. The simulation and numerical results show that the decoding SINR of each individual device
is slightly greater than the target decoding SINR. By varying the target decoding SINR, we evaluate
the algorithm performance with different channel gain in terms of stable θ which represents the amount
of received energy for harvesting. And also this algorithm reveals the potential relation between the
stable θ and throughput with different target decoding SINR. For wireless powered MTC networks, this
algorithm is capable of providing a solution which satisfies the reliability of MTC and maximizes the
throughput of each device.
References
[1] P. T. The, B. H. Mai, N. T. Tuan and T. C. Hung , (2017) Improving Distributed Energy-efficient
Clustering Algorithm to Save Lifetime for Heterogeneous WSN, International Journal of Com-
puter Networks Communications, Vol. 9, No. 7, pp. 81-96.
[2] X. Lu, P. Wang, D. Niyato, D. I. Kim, and Z. Han, (2015) Wireless networks with rf energy
harvesting: A contemporary survey, IEEE Communications Surveys & Tutorials, Vol. 17, No. 2,
pp. 757-789.
[3] D. W. K. Ng, E. S. Lo, and R. Schober, (2013) Wireless information and power transfer: Energy
efficiency optimization in ofdma systems, IEEE Transactions on Wireless Communications, Vol.
12, No. 12, pp. 6352-6370.
[4] D. Niyato and P. Wang, (2014) Delay-limited communications of mobile node with wireless
energy harvesting: performance analysis and optimization, IEEE Transactions on Vehicular Tech-
nology, Vol. 63, No. 4, pp. 1870-1885.
115

[5] L. Liu, R. Zhang, and K.-C. Chua, (2013) Wireless information transfer with opportunistic energy
harvesting, IEEE Transactions on Wireless Communications, Vol. 12, No. 1, pp. 288-300.
[6] Y. Zeng and R. Zhang, (2015) Optimized training design for wireless energy transfer, IEEE
Transactions on Communications, Vol. 63, No. 2, pp. 536-550.
[7] J. Xu and R. Zhang, (2016) A general design framework for mimo wireless energy transfer with
limited feedback, IEEE Transactions on Signal Processing, Vol. 64, No. 10, pp. 2475-2488.
[8] H. Ju and R. Zhang, (2014) Throughput maximization in wireless powered communication net-
works, IEEE Transactions on Wireless Communications, Vol. 13, No. 1, pp. 418-428.
[9] Tsiropoulou, E. E., Katsinis, G. K., Papavassiliou, S, (2012) Distributed uplink power control in
multiservice wireless networks via a game theoretic approach with convex pricing, IEEE Trans-
actions on Parallel and Distributed Systems, Vol. 23, No. 1, pp. 61-68.
[10] Zheng, J., Wu, Y., Zhang, N., Zhou, H., Cai, Y., Shen, X, (2017) Optimal power control in
ultra-dense small cell networks: A game-theoretic approach, IEEE Transactions on Wireless
Communications, Vol. 16, No. 7, pp. 4139-4150.
[11] Liu, Z., Li, S., Yang, H., Chan, K. Y., Guan, X, (2017) Approach for power allocation in two-tier
femtocell networks based on robust non-cooperative game, IET Communications, Vol. 11, No.
10, pp. 1549–1557.
[12] Kang, K., Ye, R., Pan, Z., Liu, J., Shimamoto, S, (2017) Game-Theory-Based Distributed Power
Splitting for Future Wireless Powered MTC Networks, IEEE Access, Vol. 5, pp. 20124–20134.
[13] Tsiropoulou, E. E., Vamvakas, P., Papavassiliou, S Energy eﬃcient uplink joint resource alloca-
tion non-cooperative game with pricing, in Proc. 2012 Wireless Communications and Networking
Conference (WCNC 2012), 2012, 2352–2356.
[14] Tsiropoulou, E. E., Vamvakas, P., Papavassiliou, S, (2017) Supermodular game-based distributed
joint uplink power and rate allocation in two-tier femtocell networks, IEEE Transactions on
Mobile Computing, Vol. 16, No. 9, pp. 2656–2667.
[15] R. D. Yates, (1995) A framework for uplink power control in cellular radio systems, IEEE Journal
on selected areas in communications, Vol. 13, No. 7, pp. 1341-1347.
116

Energy Splitting for SWIPT in QoS-constraint MTC Network: A Non-Cooperative Game Theoretic Approach

More Related Content

What's hot (19)

Similar to Energy Splitting for SWIPT in QoS-constraint MTC Network: A Non-Cooperative Game Theoretic Approach (20)

More from IJCNCJournal (20)

Recently uploaded (20)

Energy Splitting for SWIPT in QoS-constraint MTC Network: A Non-Cooperative Game Theoretic Approach