Cooperative Wireless Energy Harvesting and Spectrum ... - IEEE Xplore

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2016.2579598, IEEE Access 1

Cooperative Wireless Energy Harvesting and Spectrum Sharing in 5G Networks Hongyuan Gao, Waleed Ejaz, and Minho Jo

Abstract—In this paper, we propose a novel best cooperative mechanism (BCM) for wireless energy harvesting and spectrum sharing in 5G networks. Data transfer and energy harvesting are finished in the designed timeslot mode. In the proposed BCM, secondary users (SUs) harvest energy from both ambient signals and primary user’s (PU’s) signals. In addition, the SU’s can act as relay for PUs and harvest energy from PU signals simultaneously. The proposed mechanism allows optimal time duration for data transfer within the timeslot. We formulate an optimization problem based on the proposed BCM with an objective to maximize throughput of PUs and SUs with constraints on data rate and energy harvest save ratios.The effectiveness of the proposed cooperative mechanism is verified by simulations and it is considered as an important stepping stone for future research in this domain. Index Terms—Cooperative, energy harvesting, optimization, 5G networks.

I. I NTRODUCTION The rapid growth of mobile devices and modern communication applications result in the explosive demand for wireless data. Communication networks have to face several challenges such as spectrum sharing and energy scarcity to deal with the dramatic increase in wireless data which shift the focus of research directions to fifth generation (5G) networks. To address these challenges wireless design engineers need to come up with energy and spectrum management solutions in 5G networks. The key technologies for 5G networks are massive MIMO, energy-efficient communications, cognitive radio networks (CRNs), visible light communication, etc [1]– [3]. In general, the aim of 5G networks is to provide high data rate and guaranteed quality-of-service (QoS) [4]. Wireless energy harvesting and spectrum sharing in 5G networks have attracted growing attention to solving the problems caused by the high demand for data and users. Compared to a conventional battery-powered communication systems, wireless energy harvesting can provide an unlimited energy This work was sponsored by the National Natural Science Foundation of China (No. 61102106, No.61571149), Heilongjiang Postdoctoral Fund (No. LBH-Z13054), the Fundamental Research Funds for Central Universities (No. HEUCF150817), the Special China Postdoctoral Science Foundation (No. 2015T80325), and the China Scholarship Council. H. Gao was with the Department of Computer and Information Science, Korea University, Sejong Metropolitan City, S. Korea. He is now with the College of Information and Communication Engineering, Harbin Engineering University, China (email: [email protected]). W. Ejaz is with the Department of Electrical and Computer Engineering, Ryerson University, Toronto, ON, M5B 2K3, Canada (email: [email protected]). M. Jo is with Department of Computer and Information Science, Korea University, Seoul, Republic of Korea (Corresponding author) (e-mail: [email protected]).

supply from ambient radio-frequency (RF) signals which can significantly alleviate energy efficiency [5]–[7]. Simultaneous wireless information and power transfer for two-hop OFDM relay system is proposed in [8], where relay harvest energy from transmitted RF signals by source and then assist information transfer from source to destination using energy harvested. In [9], a simultaneous wireless information and power transfer scheme based on opportunistic communications is proposed to resolve practical challenges in interference alignment networks, which is also a potential technology for 5G networks. In order to present a unified framework of wireless energy harvesting and data transmission in interference alignment networks, a power splitting optimization algorithm is proposed to improve network performance [10]. Therefore, wireless energy harvesting and power transfer are key technologies in 5G networks. On the other hand, spectrum sharing ensures the coverage of 5G network everywhere and all the time. It can support a large number of mobile users and ubiquitous applications [11], [12]. Cognitive radio (CR) technology is considered as a promising technology in 5G wireless systems to solve the lack of spectrum through spectrum sharing. The CR technique allows unlicensed/ secondary users (SUs) to share the idle spectrum of licensed/ primary users (PUs). A public-private spectrum sharing scheme is proposed in order to accelerate 5G Quality of Experience (QoE) [13]. A statistical entropybased methodology is proposed from an information theory perspective to measure the degree of predictability in realworld radio spectrum state dynamics [14]. However, CR devices need to consume more energy to perform exclusive functionalities such as sensing and sharing [15]. Thus, it is more important to apply wireless energy harvesting method in addition to spectrum sharing to improve both energy and spectrum efficiency. Cooperative communication in CRNs allows SUs to actively help each other to look for opportunities in order to achieve better throughput [16]. In this paper, cooperative communication means that when PUs are busy, SUs can work as transmission relays to enhance PUs’ transmission abilities (or throughput). This result in more data transmission opportunities for SUs because PUs can complete their data transmission more rapidly and they release the licensed spectrum for SUs earlier than the traditional case. However, most of the existing cooperative communication mechanisms among users (devices) in CRNs have realized that the cooperation is mainly used for data transfer or spectrum sharing, and do not completely consider simultaneous wireless energy harvesting and data transfer for the whole CR system.

2169-3536 (c) 2016 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.


In this paper, we consider a 5G network that consists of multiple primary networks. We propose a best cooperation mechanism (BCM) for cooperative 5G network that can harvest energy and transmit data simultaneously in timeslot mode. A timeslot is divided into several flexible time intervals to carry out data transfer or/and energy harvesting. Different from the traditional cooperative mechanism in CRNs, in which SUs only cooperate with each other, in our proposed BCM, SUs cooperate with both SUs and PUs. Both SUs and PUs can harvest energy from ambient radio signals. The proposed BCM is the complete study on wireless energy harvesting and spectrum sharing in 5G networks, which aims to improve the throughput of PUs and SUs. We formulate a maximal throughput problem for the coexistence scenario of energy harvesting and data transfer in order to prove the feasibility and efficiency of proposed BCM. The main contributions of this paper are can be summarized as follows: 1) We propose a novel BCM in which SUs act as a relay for simultaneous improvement of throughput and energy efficiency of SUs and PUs in timeslot mode. 2) We introduce a flexible time intervals for data transmission in our proposed mechanism. PU and SU can harvest energy from ambient signals to transmit information with the proposed BCM. 3) We formulate an optimization problem for which objective is to maximize throughput with constraints on data transmission in a slot. Cat swarm optimization (CSO) is used to solve the formulated optimization problem. 4) We conduct simulations to solve the optimization problem in proposed mechanism which verifies the feasibility and superiority of proposed mechanism. This paper is organized as follows: In Section II, the related works of energy harvesting, spectrum sharing, and cooperative mechanism for communications systems are introduced. In Section III, a system model for wireless energy harvesting and spectrum sharing in cognitive 5G networks is presented. Then, in Section IV, we detailed our proposed novel BCM which operates in timeslot mode and an optimization problem is formulated to maximize throughput, and CSO algorithm is used to solve the formulated problem. In Section V, the simulation results are presented and comparison of proposed BCM is done with two existing cooperative schemes in literature. Finally, the conclusion is drawn in Section VI followed by future work and challenges. II. R ELATED W ORK A. Wireless Energy Harvesting and Spectrum Sharing in 5G Networks Wireless energy harvesting and spectrum sharing are promising technologies in 5G networks for prolonged device lifetime and spectrum efficiency respectively. In [17], authors presented an integrated architecture for spectrum and energy harvesting to deal with the challenges (such as data rate requirement) posed by modern and sophisticated applications. A spectrum and energy control mechanism is also proposed that attracts cooperative sensing participants along with the capability of energy harvesting. An optimization problem is

formulated to maximize the energy efficiency of energy harvesting cooperative networks in [18]. A decode-and-forward relay-based cooperative network is considered in which selfenergy recycling is powered by wireless energy harvesting. In [19], authors designed a CRN with wireless energy harvesting with a goal to achieve spectrum and energy efficiency for SUs. The objective is to maximize the total throughput with constraints on energy causality and collision. A novel data transfer protocol was developed in a two-way relay network to improve the outage probability for the primary network in [20]. The energy-efficiency of the whole system is analyzed in order to set up proper parameters to maximize it. In [21], SUs harvest energy from PU transmitters and store harvested energy to transmit data. The relationship between SU throughput and primary transmitter density was given in a stochastic geometry model. In order to obtain the maximal throughput of SUs under energy neutrality constraint and fading channel conditions, a channel selection method is proposed in [22] for energy harvesting in CRNs. In [23], a stochastic geometry model based on the distribution of PUs and SUs is designed to maximize secondary network throughput under the given outage probability constraints in the designed CRNs. A channel access policy for energy harvesting in CRNs to maximize expected long term throughput [24]. In [25], an energy harvesting CR system based on slotted mode is proposed, in which only SU can harvest energy from the ambient environment and system parameters can be optimized for the “harvestingsensing-throughput” tradeoff. B. Cooperative Mechanism for Energy Harvesting and Spectrum Sharing in 5G Networks A cooperative mechanism is a key technology to improve the efficiency of spectrum sharing in 5G networks that gives more opportunities to SUs to transmit on licensed spectrum, not in use of PUs. Compared to non-cooperative spectrum sharing technology, cooperative mechanisms have more advantages and have attracted much attention of researchers. In [26] authors studied the cooperative spectrum sharing with wireless energy harvesting. An optimization problem is formulated to maximize the throughput with constraints on the performance of primary systems. The system performance is analyzed using the stochastic geometry theory. In [27], authors considered two cooperative relay models in a cognitive femtocell network where SUs can make smart decisions when they want to obtain maximal throughput by a generalized Lyapunov optimization algorithm. Beamforming and multiple antennas for SUs can help cognitive cooperation for the PUs. In [28], a zero-forcing (ZF) beamforming method for multi-antenna cognitive base stations is proposed to help a primary system to gain the upper bound in transmitting its own signals. It is important to design a cooperative spectrum sensing scheme to tackle errors and obtain the requirements of PUs and SUs while designing antenna weights and making power allocation. In [29], the authors mentioned an overlay spectrum sharing scheme where the PU leases half of its timeslots to the SUs, in return SUs cooperative relaying the PU’s data using amplify and forward scheme. The SU’s antenna weight design and power



allocation scheme is proposed to tackle a certain error or rate design criterion for both PUs and SUs. To solve the cooperative decentralized stochastic optimization problem of cooperative multiuser CRNs, a decentralized learning algorithm is proposed by [30] to obtain the optimal channel access policies. To boost the primary system’s performance by relaying and providing more opportunities for the secondary system to access the spectrum, three schemes are proposed by [31] for energy harvesting CR system. In [32] and [33], the optimal cooperation protocols are proposed for energy harvesting CR system with timeslot structure to obtain the maximal throughput. Compared to the previous research, the main features of our work is that we do not only propose novel BCM for wireless energy harvesting and spectrum sharing in cognitive 5G networks, but also design a new cooperation strategy that maximizes the throughput of both SUs and PUs by using timeslots and CSO to solve new throughput formulas. III. S YSTEM M ODEL We consider a cognitive 5G network that consists of a PU network and an SU system. The PU networks include a primary transmitter (PT) and a primary receiver (PR) while the SU system includes a secondary transmitter (ST) and a secondary receiver (SR). Fig. 1 (a) shows the cooperative cognitive 5G network for spectrum sharing in a one-to-one (with one PT/PR and one ST/SR) which is considered in our proposed BCM. It can be seen that ST 1 act as a relay for the PU system, i.e., PT 1 and PR 1. Thus, ST 1 and SR 1 can share the licensed spectrum in return for ST 1’s relay for PT 1 and PR 1. Fig. 1 (b) is no-cooperative cognitive 5G network because SR does not play a role as a relay for the PUs. Fig. 1 (c) and (d) are for our future research ideas. In Fig. 1 (c), multiple secondary transmitters (ST 2 and ST3) are used as relay and in Fig. 1 (d), a single secondary transmitter (ST 4) is used for relay for multiple primary transmitters (PT 3 and PT 4). This future work will be discussed more in Section VI. In every timeslot, the PU will transmit data through the licensed channel to the intended receiver. After the PU finishes data transmission, the PU begins to harvest energy from ambient signals and use an energy storage device to store energy for PU’s data transmission for the next timeslot. For the first timeslot, the energy storage device of the PU needs to be charged in advance. However, for the following timeslots, the PU uses the energy harvested during the last timeslot. In order to avoid collision between the SU and the PU, the SU is not allowed to use the licensed spectrum to transmit its own data when the PU utilizes the licensed channel because the PU has priority. Moreover, the SU harvest energy from ambient radio signals first when the licensed channel is used. In this paper, the SU’s transmitter abides by a save-then-transmit protocol for energy half-duplex constraints. That is because the rechargeable energy storage device practically cannot charge and discharge at the same time. We assume that in every timeslot, SU spends a part of a time in the slot for harvesting energy to be consumed for its data transmission. In the proposed BCM, when the PU transmit its data, the SU can also work as a cooperative relay to help the PU’s

SU harvests energy from PU

PU directly transmits its data

PU transmits data to SU and PU receiver

SU transmits data for PU

PU harvests energy from ambient signals

SU transmits its data

SU harvests energy from ambient signals T

1T

 2T

PU 2  2T

SU

Fig. 2. An example of the proposed timeslot structure with BCM for energy harvesting and spectrum sharing.

data transmissions. This cooperative relay of SU not only can improve the throughput of the PU, but also can finish PU data transmission earlier which in turn provide more opportunities for sharing spectrum for SU’s transmission. The cooperative communication between the PUs and the SUs while considering the SU’s relay role can potentially increase the throughput of both SU’s and PU’s. During the cooperative process, the SU should harvest energy in a timeslot mode in order to help PUs to relay its data. IV. B EST C OOPERATIVE M ECHANISM (BCM) USING T IMESLOT FOR E NERGY H ARVESTING AND S PECTRUM S HARING We propose a novel BCM using timeslots for energy harvesting and spectrum sharing in 5G network. The proposed cooperative mechanism allows PU and SU to harvest energy in three modes: the SU harvests energy from ambient signals, the SU harvests energy from the PUs, and the PUs harvest energy from ambient signals, as shown in Fig. 2. Thus, BCM allows SU to harvest energy from both sources, i.e., PU signals and ambient signals. Unlike the existing work, the PU is no longer restricted to transmit a fixed amount of data in a timeslot. Further, simultaneous data and energy transfer technology from PU to SU is proposed in the cooperative mechanism, i.e., SU can receive data and harvest energy from PU simultaneously in the blue/magnet color part of SU timeslot in Fig. 2. The above three different ideas are significantly outstanding cooperative mechanisms compared to the existing methods. A detailed list of symbols used and their description is provided in Table I.



Primary Transmission

Secondary Transmission

Relay Transmission

Command Signal b

a PT1

ST1 PT5 SR1

PR5

ST5 SR5

PR1 Base Station c

d ST2

PR3 SR2

PT3

PR2 PT2 SR3

ST4

SR4

PT4 PR4

ST3 Fig. 1. Cooperative and non-cooperative mechanisms in cognitive 5G networks: (a) cooperation in one-to-one; (b) non-cooperation in one-to-one; (c) cooperation with multiple-to-one; (d) cooperation with one-to-multiple.

In the BCM using timeslot mode shown in Fig. 2, a timeslot duration T is divided into three parts according to different transactions, which can be explained as follows: • At time interval (0, ρ1 T ], when PU transmits its data directly in the non-cooperative style without using SU’s as relay indicating the red color of PU timeslot. ρ1 is the SU’s energy harvest save ratio from ambient signals. At this time, SU needs to harvest energy from ambient radio signals and PU signals indicated in the aqua color of SU timeslot. • At time interval (ρ1 T, (ρ1 + 2ρ2 )T ], the SU works as a cooperative relay for the PU to improve the throughput and to finish the PU’s transmission earlier than normal time. ρ2 is the SU’s energy harvest save ratio from PU signals. In the first half of the cooperative interval (ρ1 T, (ρ1 + ρ2 )T ], when the PU’s transmit its data to the receiver in blue color of PU timeslot, the SU receives data and harvests energy simultaneously in blue/magnet colors of SU timeslot. In the second half of the cooperative interval ((ρ1 + ρ2 )T, (ρ1 + 2ρ2 )T ], while the PU harvests energy from ambient signals in yellow color of PU timeslot, the SU relays the PU’s data to the PU receiver in the dark cyan color of SU timeslot. • At time interval ((ρ1 + 2ρ2 )T, T ], after the PU completes the transmission in blue color of PU timeslot, then the idle licensed channel (shown in yellow color of PU timeslot) can be used by the SU and the SU will transmit its own data by using the idle licensed channel in green color of SU timeslot. Through all this, the PU keeps harvesting

energy from ambient sources, and the harvested energy of the PU will be used in the next timeslot. Since we focus on short-term optimization of throughput, we assume that the SUs have to use up the harvested energy during each timeslot. For the proposed BCM, we formulate an optimization problem with an objective to maximize throughput of both PUs and SUs. We introduce the energy harvest save ratio definition which is a portion of energy harvesting time within a timeslot. Let Rpt and Rst represents throughput of the PU and the SU in each timeslot respectively. Rpt depends on Rp and Rc , where Rp represents PU’s instantaneous non-cooperative transmission rate during one timeslot and Rc denotes PU’s instantaneous cooperative transmission rate during one timeslot. Rp is affected by the SU’s energy harvest save ratio from ambient signals (ρ1 ), the SU’s energy harvest save ratio from PU signals (ρ2 ), the channel-power-gain-to-noise-power ratio of the transmitter and receiver of PU (γp ), and current available energy which including energy supply of energy equipment (can be equal to 0) and harvested energy of the last timeslot. Rp can be expressed as Xp (1 − ρ1 − ρ2 ) γp , Rp = log2 1 + Yp + ρ1 + ρ2

(1)

where ρ1 and ρ2 are SU’s energy harvest save ratio from ambient signals and PU signals in last time slot respectively. Yp represents energy supply rate of PU and Xp represents PU’s energy harvest rate in the last timeslot.



TABLE I D ESCRIPTION OF THE SYMBOLS USED IN THE BCM.

We formulate an optimization problem for maximization of throughput of the SU and the PU in each timeslot as:

Symbol Description T Qpmin Rc Rp

Timeslot duration PU’s required minimal data transmission amount in each timeslot PU’s instantaneous cooperative transmission rate PU’s instantaneous non-cooperative transmission rate SU’s instantaneous non-cooperative transmission rate Energy supply rate of PU PU’s energy harvest rate in the last timeslot SU’s energy harvest rate from ambient signals SU’s energy harvest save ratio from ambient signals in the current timeslot SU’s energy harvest save ratio from PU signals in the current timeslot SU’s energy save ratio from ambient signals in the last timeslot SU’s energy save ratio from PU signals in the last timeslot SU’s allocated power for cooperative relay Channel-power-gain-to-noise-power ratio of the transmitter and receiver of PU Channel-power-gain-to-noise-power ratio between PU’s transmitter and SU’s transmitter Channel-power-gain-to-noise-power ratio of the link between SU’s transmitter and PU’s receiver Channel-power-gain-to-noise-power ratio between the SU’s transmitter and receiver

max : Rpt + Rst ,

ρ1 ,ρ2 ,ws

Subject to : Xs ρ1 + Xsp ρ2 − 2ρ2 ws ≥ 0, 1 − ρ1 − 2ρ2 ≥ 0, Rpt ≥ Qpmin ,

(4)

where Rst = (1 − ρ1 − 2ρ2 )Rs and Rpt = ρ1 Rp + 2ρ2 Rc . Qpmin represents PU’s minimal data transmission amount in each timeslot. The first constraint refers to energy causality Yp constraint that the energy used for relay transmission should Xp be less than the energy harvested by the SU. For the SU, the Xs energy harvested in the current timeslot should be consumed ρ1 by SU in the current timeslot. In the case of PU, the energy harvested by PU in the last timeslot should be consumed ρ2 in the current timeslot. The second constraint refers to the cooperative transmission duration and the non-cooperative ρ1 transmission duration must be less than the duration of the timeslot. In the third constraint, the throughput of PU must be ρ2 bigger than the PU’s minimal objective throughput. We assume that SU should not start its data transmission ws before PU finishes its data transmission. In other words, γp SU cannot share the licensed channel when the channel is occupied by PU. Under all these constraints, we need to find γs the best values for energy harvest save ratios for SUs, i.e., ρ1 and ρ2 , and allocated power for the cooperative relay (ws ) to rp maximize the sum rate of PU’s and SU’s, i.e., sum of Rpt and Rst . rs We solve the optimization problem for our proposed BCM using CSO which is an optimization algorithm. CSO is an opportunistic method with fast global convergence [34]. The initial population of CSO was created randomly from the Rc has a relation with energy harvesting of the last timeslot, definition domain. In CSO, each cat’s position represents a channel-power-gain-to-noise-power ratio between PU’s trans- potential solution for a cooperative spectrum sharing system in mitter and SU’s transmitter (γs ), channel-power-gain-to-noise- 5G networks. The target of the maximal throughput optimizapower ratio of the link between SU’s transmitter and PU’s tion using CSO is the maximization of the objective function. receiver (rp ), SUˆas allocated power (ws ) for the cooperative If a cat’s position (a latent solution) satisfies the constraints, relay. Rc can be expressed as the fitness of the cat is equal to objective function value. If the cat’s position does not satisfy the constraint conditions, a penalty factor is used for the constraint optimization problem 1 Xp (1 − ρ1 − ρ2 ) Rc = min log2 1 + Yp + γs , to guarantee the convergence, i.e., if the solution does not 2 ρ1 + ρ2 (2) satisfy constraints then it will be eliminated by cat swarm Xp (1 − ρ1 − ρ2 ) in the future, and the fitness of the cat is equal to the log2 1 + Yp + γp + ws rp }. ρ1 + ρ2 multiplication of objective function value and the penalty Rs represents SU’s instantaneous non-cooperative transmis- factor which is set as 0.1. The description of CSO for the sion rate. In the designed BCM, SU’s cooperative transmission BCM is given in Algorithm 1. For the detailed evolutionary rate term is not needed because the SU only receives the data process of CSO, see [34]. from the PU as a relay in the cooperative mode. Rs has a relation with SU’s energy harvest rate (Xsp ) from the PU’s V. P ERFORMANCE A NALYSIS signal, SU’s energy harvest rate (Xs ) from ambient signals, In this section, we evaluate the performance of proposed channel-to-noise power gain ratio between the SU’s transmitter BCM. For simulations, we use a cognitive 5G network a and receiver (rs ). Rs can be expressed as PU network and an SU system. We consider a scenario for spectrum sharing and energy harvesting in a one-to-one way Xs ρ1 + Xsp ρ2 − 2ρ2 ws RS = log2 1 + rs . (3) (with one PT/PR and one ST/SR) as shown in Fig. 1 (a). 1 − ρ1 − 2ρ2 However, we can extend this prototype simulation model Rs



Algorithm 1 Generale description of CSO According to the system requirements, define the objective function and solution space for CSO Initial a population of H cats with random variants which includes velocities and positions t t t , . . . , viD ), xti =(xti1 , xti2 , . . . , xtiD ), , vi2 vti =(vi1 i = 1, 2, . . . , H Compute the fitness of every cat and store the global best position ptb =(ptb1 , ptb2 , . . . , ptbD ) while (t < Maximal Generation) do for i = 1 : H do if (rand < MR) then % rand represents uniform random variant between 0 and 1 Use seeking mode to generate new solutions Make SMP copies Allow SMP-1 copies to mutation based on SRD and CDC Compute the fitness of every copy and choose the best copy to update the position of the cat else Use tracing mode to generate new velocities and positions vit+1 =wt vit + c × rand × (ptb − xti ) xt+1 =xti + vit+1 i end if end for Find current global best position pt+1 b end while

to a more realistic model. We compare our proposed BCM with the existing schemes presented in [33]. From [33], the non-cooperative mode is named as Type A and the simple cooperative mode is named Type B. In Type A, SU starts its transmission only after PU finishes its transmission to PU’s receiver. In Type B cooperative protocol, SU can cooperate with PU by a save ratio of battery and allocate battery power for the throughput of SU. We use the following simulation parameters for CSO: constant factor (c=2), counts of dimensions to change (CDC=0.5), seeking memory pool (SMP=6), seeking a range of selected dimension (SRD =0.2), and mode ratio (MR=0.8). Inertia weight (wt ) will decrease from 0.9 to 0.4 and population size (H) is set to 300. The maximal iteration number is set to 100. Detailed simulation parameters are given in Table II. Some parameters definition and processes are adopted from [34] and [35]. For type A and type B, the throughput of PU must be fixed which is the minimum required, i.e., Qpmin = 4. For the sake of simplicity, we set T = 1 and Xp =Xp . For the proposed BCM, we assume that PU’s minimal throughput equal or superior to type A and type B in each timeslot for PU’s throughput. In addition to special explanation, the environmental parameters of cognitive 5G network are set as the same for BCM, Type A, and Type B methods in simulation experiments. Unless specifically stated, we set γp =0.4, γs =100, rp =200, rs =40, Xp =0, 50, 100, 150, Yp =100, Xsp =30, Xs =20:4:60,

TABLE II S IMULATION PARAMETERS FOR CSO.

parameter

Value

Constant factor (c) Counts of dimensions to change (CDC) Seeking memory pool size (SMP) Seeking range of selected dimension (SRD) Mixture ratio (MR) Inertia weight (wt )

2 0.5

Population size (H) Maximal iteration number

6 0.2 0.8 decrease from 0.9 to 0.4 300 100

and ws ∈ [10, 200]. A. PU’s Energy Provided by Energy Harvesting and Fixed Energy Resource Fig. 3 (a) and (b) show achievable sum throughput of both PU and SU, and achievable throughput of SU versus energy harvesting rate of SU respectively for BCM (with varying PU rate), Type A, and Type B. It is obvious that the proposed BCM can satisfy the throughput requirement of the PU. We can also notice that the achievable throughput of SU will be increasing with the increase of energy harvesting rate of the PU and the SU. We can see that even energy harvesting rate of PU is 0, the proposed BCM is superior to the traditional non-cooperative and cooperative schemes. Thus, the proposed BCM is an efficient scheme for wireless energy harvesting and spectrum sharing in 5G networks. Fig. 4 (a) and (b) give maximal throughput simulation results in 30 timeslots with the different probability distribution of energy harvesting rate and channel condition. We set the variables as Xp ∼ Γ(40, 5), Xs ∼ Γ(20, 5), Xsp ∼ Γ(10, 5), γs ∼ EXP (200), rp ∼ EXP (100), and rs ∼ EXP (40), where Γ represents the Gamma distribution function and EXP represents the exponential distribution function, since a Rayleigh fading channel is considered during the simulations. We can notice that the proposed cooperative scheme is superior to the traditional non-cooperative scheme Type A and simple cooperative scheme Type B. Compared to Type A and Type B, our proposed mechanism can harvest more energy through both PU and SU and thus more power can be used for data transmission of SU and PU and for cooperative relay of SU. Fig. 5 (a) and (b) show achievable sum throughput of both PU and SU, and achievable throughput of SU versus energy harvesting rate of PU respectively for BCM, Type A, and Type B. We set the variables as Yp =40:6:100, Xp = 100, Xs = 40, and Xsp = 40. It is obvious that the achievable throughput of BCM is superior to other two methods. It is because of that our proposed BCM can harvest more energy through both PU and SU and thus more power can be used for data transmission of SU and PU, and for the cooperative relay of SU. In addition, it can be concluded that if energy harvesting rate of PU and



10

Achievable sum throughput of PU and SU


8.5

8

7.5

7

Type A Type B BCM with X =0 p

BCM with X p =50

6.5

BCM with X =100 p

9.5 9 8.5 8 7.5 7 6.5

Type A Type B

6

BCM

BCM with X p =150

6 20

5.5 25

30

35

40

45

50

55

60

0

5

10

Energy harvesting rate of SU

15

20

25

30

25

30

Timeslot

(a)

(a)

5.5

4.5

Achievable throughput of SU


5 4

3.5

3

Type A Type B BCM with X p =0 BCM with X p =50

2.5

4.5

4

3.5

3 Type A

BCM with X p =100

2.5

Type B BCM

BCM with X p =150

2 20

2 25

30

35

40

45

50

55

60


0

5

10

15

20

Timeslot

(b)

(b)

Fig. 3. Relationship of SU’s throughput and energy harvesting rate of SU (a) Sum throughput of both SU and PU, and (b) SU’s throughput.

Fig. 4. Comparison of throughput traces for three different mechanisms (a) Sum throughput traces of both SU and PU, and (b) Throughput traces of SU.

SU are fixed in some special environment, we can improve battery capacity of PU to transmit big data of SU. For SU, when Yp = 40, the throughput of BCM is about 5 times of Type B. From Figs. 3-5, we can see that the proposed BCM can satisfy the requirement of both PU and SU. Compared to existing methods, our proposed BCM can harvest more energy through both PU and SU and thus more power can be used for data transmission of SUs, PUs, and for the cooperative relay of SUs. We also can obtain satisfactory SU throughput by setting of PU’s minimal transmission data. So, energy harvesting and data transmission have an intimate relationship, the BCM can resolve the conflict of energy harvesting and data transmission by resolving the optimal parameters. B. PU’s Energy provided by Only Energy Harvesting Here, the PU energy is provided by energy harvesting and there is no need of additional energy equipment (i.e., Yp = 0).

We use BCM to harvest energy for PU and SU. For Type A and Type B, PU must be supplied energy by the traditional powered system (Yp = 100) and only SU can harvest energy from ambient signals. Fig. 6 (a) and (b) show achievable sum throughput of both PU and SU, and achievable throughput of SU versus energy harvesting rate of SU from ambient signals and PU’s energy harvesting rate respectively for BCM (with varying Xp ), Type A, and Type B. We set the variables as Xp = 120, 170, 220, 270, and Xsp = 30. If energy harvesting rate of PU is high, the achievable throughput of BCM is superior to the existing Type A and Type B methods. It is because of the fact that PU in proposed BCM can harvest energy in idle time to speed PU’s and SU’s data transmission. So, it is feasible that PU and SU can work well by energy harvesting without additional energy supply equipment. Fig. 7 (a) and (b) give maximal throughput simulation results in different channel-power-to-noise-power ratio of the



8



8.8 7.5 7 6.5 6 5.5 5

Type A Type B

4.5 4 40

BCM

50

60

70

80

90

Type B

8.6

BCM with X p =120

8.4

BCM with X p =170

8.2

BCM with X p =220 BCM with X =270 p

8 7.8 7.6 7.4 7.2 7 6.8 20

100

Type A

25

30

Energy supply rate of PU

35

40

45

50

55

60


(a)

(a)

4

3.5



4 3

2.5

2

1.5

Type A Type B

Type A

3

Type B BCM with X p =120 BCM with X p =170

BCM

1

3.5

2.5

BCM with X =220 p

BCM with X p =270

0.5 40

50

60

70

80

90

100

Energy supply rate of PU

20

25

30

35

40

45

50

55

60


(b)

(b)

Fig. 5. Comparison of throughput in different energy supply rate of PU for three mechanisms (a) Sum throughput of both SU and PU, and (b) SU’s throughput.

Fig. 6. Comparison of throughput traces for three mechanisms (a) Sum throughput of both SU and PU, and (b) SU’s throughput traces.

transmitter and receiver of PU (γp ) for BCM, Type A, and Type B. We set the variables γp =0.2:0.03:0.5, γs = 60, rp = 80, Xp = 200, Xs = 40, and Xsp = 40. It is obvious that the achievable throughput of BCM is superior to the other two methods. For higher values of γp , Type B has the same throughput as Type A, and the cooperative relay of Type B does not produce good result. However, the proposed BCM has obvious performance advantages.And it is easy to understand that BCM is superior to other two mechanisms in the same simulation environment with different channel-power-to-noisepower ratio of the transmitter and receiver of PU. Fig. 8 (a) and (b) give maximal throughput simulation results in different channel-power-gain-to-noise-power ratio of the transmitter and receiver of SU for BCM, Type A, and Type B. We set the variables as rs =10:3:40, γs = 110, Xp = 200, Xs = 40, and Xsp = 40. It is obvious that the achievable throughput of BCM is superior to other two methods, because PUs and SUs in BCM can harvest

more energy for data transmission and thus, obtain excellent throughput performance which has relationship with power of PU and SU. The channel condition is a key factor for data transmission. Thus, in order to obtain high throughput in the same channel conditions, BCM should be used in 5G systems, and BCM can transmit more data in high noise environment than other existing Type A and Type B methods in low noise environment. Fig. 9 (a) and (b) give maximal throughput simulation results in different energy harvesting rate of SU from PU for BCM, Type A, and Type B. We set the variables as Xp = 100, Xsp =10:3:40, and Xs = 20, 25, 30, 35. It can be seen that the achievable throughput will be increasing with the increase in SU’s energy harvesting rate from ambient signals and from PU signals. The maximal throughput has an intimate relationship with two kinds of SU’s energy harvesting rate. We can see that the achievable throughput increase with the increase of energy harvesting rate of SU. In order to obtain more throughput, we



8

8

7.5

7

6.5

Type A Type B BCM

6

5.5



8.5

5 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Channel−power−gain−to−noise−power ratio of transmitter and receiver of PU

Type A Type B BCM 7.5

7

6.5 10 15 20 25 30 35 40 Channel−power−gain−to−noise−power ratio of the transmitter and receiver of SU

(a)

(a)

4.5

4 Type A Type B BCM

3.5

3

2.5

2

Type A Type B BCM

1.5



4

3.5

3

2.5 1 0.2 0.25 0.3 0.35 0.4 0.45 0.5 Channel−power−gain−to−noise−power ratio of transmitter and receiver of PU

10 15 20 25 30 35 40 Channel−power−gain−to−noise−power ratio of the transmitter and receiver of SU

(b)

(b)

Fig. 7. Comparison of throughput with different Xs for three mechanisms (a) Sum throughput of both SU and PU, and (b) SU’s throughput.

Fig. 8. Comparison of throughput with different rs for three mechanisms (a) Sum throughput of both SU and PU, and (b) SU’s throughput.

need to use the high rate of SU’s energy harvesting. When SU’s energy harvesting rate from ambient signals is fixed, we can control the SU’s energy harvest rate from PU’s signals to speed data transmission rate of SU to some extent. Fig. 10 (a) and (b) gives maximal throughput simulation results in different energy harvesting rate of SU from ambient signals and channel-power-gain-to-noise-power ratio between SU’s transmitter and PU’s receiver. We set the variables as γp = 1, γs = 200, rp =60:10:160, Xp = 100, Xs = 20, 25, 30, 35, and Xsp = 20. It is obvious that the achievable throughput increase with the increasing of SU’s energy harvesting rate and rp . We can obtain more throughput by improving SU’s energy harvesting rate and channel condition of rp . According to Figs. 6-10, it is obvious that the proposed optimal cooperation scheme can satisfy the throughput requirement of PU by energy harvesting. We not only give the comparison of the proposed BCM and previous mechanisms, but also show the relationship of throughput and system

parameters. Compared to previous methods and mechanisms, we can obtain excellent performance, because the proposed mechanism can harvest more energy through both PU and SU and thus more power can be used for data transmission of SU and PU and for cooperative relay of SU. It should be noticed that 5G networks based on BCM can attain excellent performance without additional energy supplication equipment in this section. VI. C ONCLUSION AND F UTURE R ESEARCH C HALLENGES In this paper, we considered wireless energy harvesting and spectrum sharing in 5G networks which operate in timeslot. In our proposed mechanism using timeslot, SU can harvest more energy from both ambient signals and PU signals when compared to the traditional mechanisms. Further, SUs actively cooperate with PUs to relay their signals, and simultaneously work as relay and harvest energy to gain more opportunities for transmission and harvesting. The proposed mechanism also offer flexible size of transmitted data. We formulated


This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2016.2579598, IEEE Access

7.2

8

7.1

7.9

Xs=20 Achievable sum throughput of PU and SU


10

7 6.9 6.8 6.7 X s =20

6.6

X =25 s

X =30 s

6.5

7.8

Xs=25 Xs=30 Xs=35

7.7 7.6 7.5 7.4 7.3

X =35 s

6.4 10

15

20

25

30

35

7.2 60 80 100 120 140 160 Channel−power−gain−to−noise−power ratio of SU’s transmitter and PU’s receiver

40

Energy harvesting rate of SU from PU

(a)

3.2

4

3

3.9

2.8

3.8



(a)

2.6 2.4 2.2 2 X =20 s

Xs=25 Xs=30 Xs=35

3.7 3.6 3.5 3.4

X =25

1.8

s

3.3

X =30 s

1.6 1.4 10

Xs=20

X s =35

15

20

25

30

35

40

3.2 60 80 100 120 140 160 Channel−power−gain−to−noise−power ratio of SU’s transmitter and PU’s receiver

Energy harvesting rate of SU from PU

(b) Fig. 9. Relationship of SU’s throughput and different Xsp (a) Sum throughput of both SU and PU, and (b) SU’s throughput.

an optimization problem to maximize the throughput of both SUs and PUs and solve it using CSO algorithm. The proposed mechanism has been verified to be superior to the traditional schemes by simulation. Thus the proposed mechanism can be an important step stone of doing more perspective research work in the future. There are many research problems in this area which are yet to be studied, we highlight some interesting research topics for cooperation mechanism in 5G networks energy harvesting and wireless information-energy transmission. Since the proposed mechanism has proved to obtain the maximal throughput, we can extend the current prototype to the real world problem by increasing the number of SUs and PUs, and by providing more detailed channel information. As shown in Fig. 1, there are more diversified network models for PUs and SUs in an actual system. Fig. 1 (c) and (d) are the cases of different cognitive 5G models for our future research. In Fig. 1 (c), called “cooperation with multiple-to-one,” multiple

(b) Fig. 10. Relationship of SU’s throughput and different rp for three mechanisms (a) Sum throughput of both SU and PU, and (b) SU’s throughput.

secondary transmitters (ST 2 and ST3), are used for relay. A single primary transmitter PT2 cooperates with two secondary transmitters ST2 and ST3 for the relay. In this case, we expect that the throughput of PU can be increased. In Fig. 1 (d), a single secondary transmitter (ST 4) is used as a relay for multiple primary transmitters (PT 3 and PT 4). In this case, the secondary transmitter ST4 can have higher opportunities of sharing the licensed channel and thus the secondary transmitter can have higher capacity. However, if the number of relay requests by two primary transmitters is increased, the throughput of the secondary transmitter can be decreased. The mix model of cooperation with multiple-toone and cooperation with one-to-multiple can be considered as another practical cognitive 5G network case for the future research. In this research, we didn’t verify how much the proposed mechanism harvests energy efficiently. Thus this specific topic can be considered to be future research work.



R EFERENCES [1] M. Iwamura, “NGMN View on 5G Architecture,” in IEEE 81st Vehicular Technology Conference (VTC Spring). Glasgow, Scotland: IEEE, May 2015, pp. 1–5. [2] H. Droste, G. Zimmermann, M. Stamatelatos, N. Lindqvist, O. Bulakci, J. Eichinger, V. Venkatasubramanian, U. Dotsch, and H. Tullberg, “The METIS 5G Architecture: A Summary of METIS Work on 5G Architectures,” in IEEE 81st Vehicular Technology Conference (VTC Spring). Glasgow, Scotland: IEEE, May 2015, pp. 1–5. [3] P. Agyapong, M. Iwamura, D. Staehle, W. Kiess, and A. Benjebbour, “Design considerations for a 5G network architecture,” IEEE Communications Magazine, vol. 52, no. 11, pp. 65–75, Nov. 2014. [4] N. ul Hasan, W. Ejaz, N. Ejaz, H. S. Kim, A. Anpalagan, and M. Jo, “Network Selection and Channel Allocation for Spectrum Sharing in 5G Heterogeneous Networks,” IEEE Access, vol. 4, pp. 980–992, Mar. 2016. [5] S. Sudevalayam and P. Kulkarni, “Energy harvesting sensor nodes: Survey and implications,” IEEE Communications Surveys & Tutorials, vol. 13, no. 3, pp. 443–461, Jul. 2011. [6] W. Ejaz, S. Kandeepan, and A. Anpalagan, “Optimal placement and number of energy transmitters in wireless sensor networks for rf energy transfer,” in IEEE 26th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC). Hong Kong: IEEE, Sep. 2015, pp. 1238–1243. [7] G. Du, K. Xiong, Y. Zhang, and Z. Qiu, “Outage analysis and optimization for time switching-based two-way relaying with energy harvesting relay node,” KSII Transactions on Internet and Information Systems, vol. 9, no. 2, pp. 545–563, Feb. 2015. [8] X. Di, K. Xiong, Y. Zhang, and Z. Qiu, “Simultaneous wireless information and power transfer in two-hop ofdm decode-and-forward relay networks,” KSII Transactions on Internet and Information Systems, vol. 10, no. 1, pp. 152–167, Jan. 2016. [9] N. Zhao, F. Yu, and V. C. Leung, “Opportunistic communications in interference alignment networks with wireless power transfer,” IEEE Wireless Communications, vol. 22, no. 1, pp. 88–95, Feb. 2015. [10] N. Zhao, F. R. Yu, and V. Leung, “Wireless energy harvesting in interference alignment networks,” IEEE Communications Magazine, vol. 53, no. 6, pp. 72–78, Jun. 2015. [11] A. El-Mougy, M. Ibnkahla, G. Hattab, and W. Ejaz, “Reconfigurable wireless networks,” Proceedings of the IEEE, vol. 103, no. 7, pp. 1125– 1158, Jul. 2015. [12] W. Ejaz and M. Ibnkahla, “Machine-to-Machine Communications in Cognitive Cellular Systems,” in IEEE International Conference on Ubiquitous Wireless Broadband (ICUWB), Montreal, QC, Oct. 2015, pp. 1–5. [13] J. Mitola, J. Guerci, and J. a. Reed, “Accelerating 5G QoE via publicprivate spectrum sharing,” IEEE Communications Magazine, vol. 52, no. 5, pp. 77–85, May. 2014. [14] G. Ding, J. Wang, and Q. a. Wu, “On the limits of predictability in real-world radio spectrum state dynamics: From entropy theory to 5G spectrum sharing,” IEEE Communications Magazine, vol. 53, no. 7, pp. 178–183, Jul. 2015. [15] W. Ejaz, G. A. Shah, H. S. Kim et al., “Energy and throughput efficient cooperative spectrum sensing in cognitive radio sensor networks,” Transactions on Emerging Telecommunications Technologies, vol. 26, no. 7, pp. 1019–1030, Jul. 2015. [16] Q. Zhang, J. Jia, and J. Zhang, “Cooperative relay to improve diversity in cognitive radio networks,” IEEE Communications Magazine, vol. 47, no. 2, pp. 111–117, Feb. 2009. [17] Y. Liu, Y. Zhang, R. Yu, and S. Xie, “Integrated energy and spectrum harvesting for 5G wireless communications,” Network, IEEE, vol. 29, no. 3, pp. 75–81, Jun. 2015. [18] S. Hu, Z. Ding, Q. Ni, W. Yu, and Z. Song, “Energy efficiency in energy harvesting cooperative networks with self-energy recycling,” in IEEE 20th International Workshop on Computer Aided Modelling and Design of Communication Links and Networks (CAMAD). Guildford, UK: IEEE, Sep. 2015, pp. 59–63. [19] S. Park, H. Kim, and D. Hong, “Cognitive radio networks with energy harvesting,” IEEE Transactions on Wireless Communications, vol. 12, no. 3, pp. 1386–1397, Mar. 2013. [20] Z. Wang, Z. Chen, Y. Yao, B. Xia, and H. Liu, “Wireless energy harvesting and information transfer in cognitive two-way relay networks,” in IEEE Global Communications Conference (GLOBECOM). Austin, TX: IEEE, Dec. 2014, pp. 3465–3470.

[21] S. Lee, K. Huang, and R. Zhang, “Cognitive energy harvesting and transmission from a network perspective,” in IEEE International Conference on Communication Systems (ICCS), Singapore, Nov. 2012, pp. 225–229. [22] J. Jeya Pradha, S. S. Kalamkar, and A. Banerjee, “Energy harvesting cognitive radio with channel-aware sensing strategy,” IEEE Communications Letters, vol. 18, no. 7, pp. 1171–1174, Jul. 2014. [23] S. Lee, R. Zhang, and K. Huang, “Opportunistic wireless energy harvesting in cognitive radio networks,” IEEE Transactions on Wireless Communications, vol. 12, no. 9, pp. 4788–4799, Sep. 2013. [24] T. T. Nguyen and I. Koo, “Throughput maximization for a primary user with cognitive radio and energy harvesting functions,” KSII Transactions on Internet and Information Systems, vol. 8, no. 9, pp. 3075–3093, Sep. 2014. [25] S. Yin, Z. Qu, and S. Li, “Achievable throughput optimization in energy harvesting cognitive radio systems,” IEEE Journal on Selected Areas in Communications, vol. 33, no. 3, pp. 407–422, Mar. 2015. [26] C. Zhai, J. Liu, and L. Zheng, “Cooperative Spectrum Sharing with Wireless Energy Harvesting in Cognitive Radio Network,” IEEE Transactions on Vehicular Technology, vol. PP, no. 99, pp. 1–13, Jul. 2015. [27] R. Urgaonkar and M. J. Neely, “Opportunistic cooperation in cognitive femtocell networks,” IEEE Journal on Selected Areas in Communications, vol. 30, no. 3, pp. 607–616, Apr. 2012. [28] S. Song, M. O. Hasna, and K. Letaief, “Prior zero forcing for cognitive relaying,” IEEE Transactions on Wireless Communications, vol. 12, no. 2, pp. 938–947, Feb. 2013. [29] R. Manna, R. H. Louie, Y. Li, and B. Vucetic, “Cooperative spectrum sharing in cognitive radio networks with multiple antennas,” IEEE Transactions on Signal Processing, vol. 59, no. 11, pp. 5509–5522, Nov. 2011. [30] D. T. Hoang, D. Niyato, P. Wang, and D. I. Kim, “Performance Optimization for Cooperative Multiuser Cognitive Radio Networks with RF Energy Harvesting Capability,” IEEE Transactions on Wireless Communications, vol. 14, no. 7, pp. 3614–3629, Jul. 2015. [31] G. Zheng, Z. Ho, E. A. Jorswieck, and B. Ottersten, “Information and energy cooperation in cognitive radio networks,” IEEE Transactions on Signal Processing, vol. 62, no. 9, pp. 2290–2303, Sep. 2014. [32] S. Park and D. Hong, “Achievable throughput of energy harvesting cognitive radio networks,” IEEE Transactions on Wireless Communications, vol. 13, no. 2, pp. 1010–1022, Feb. 2014. [33] S. Yin, E. Zhang, Z. Qu, L. Yin, and S. Li, “Optimal cooperation strategy in cognitive radio systems with energy harvesting,” IEEE Transactions on Wireless Communications, vol. 13, no. 9, pp. 4693–4707, Sep. 2014. [34] S. K. Saha, S. P. Ghoshal, R. Kar, and D. Mandal, “Cat swarm optimization algorithm for optimal linear phase FIR filter design,” ISA transactions, vol. 52, no. 6, pp. 781–794, Nov. 2013. [35] L. Pappula and D. Ghosh, “Linear antenna array synthesis using cat swarm optimization,” AEU-International Journal of Electronics and Communications, vol. 68, no. 6, pp. 540–549, Jun. 2014.
