Energy-Efficient Target Channel Sequence Design for Spectrum Handoffs in Cognitive Radio Networks

Xiaolong Yang, Xuezhi Tan

1 Chongqing Key Lab of Mobile Communications Technology, Chongqing University of Posts and Telecommunications,Chongqing 400065, China

2 Communication Research Centre, Harbin Institute of Technology, Harbin 150080, China

I. INTRODUCTION

Cognitive radio has been viewed as a promising technique to alleviate the spectrum scarcity problem caused by the increase of wireless equipment and the appearance of new wireless technologies [1][2]. In order to significantly improve the utilization efficiency of frequency,the secondary user (SU) can be allowed to access the spare licensed spectrum of the primary user (PU) or unlicensed spectrum opportunistically in Cognitive radio networks (CRNs)[3][4]. For preserving the desired quality of service (QoS) of the PU, the SU will impose challenges, mainly including four parts: spectrum sensing, spectrum decision, spectrum sharing, and spectrum mobility [5][6]. In this paper, we focus on spectrum mobility. The SU may be interrupted by the appearance of the PU or the awful channel condition for the high fluctuation in dynamic CRNs. Considered as the main issue in spectrum mobility, spectrum handoff is the process that a SU stay or change its operating channel upon the appearance of a PU and then re-build a new communication connection to resume or repeat its data transmission [7]-[10]. Thus, spectrum handoff,which aims at guaranteeing secondary users’(SUs’) transmission continuity, is a very important functionality of CRNs.

The authors design the target channel sequence for a cognitive radio network in which the secondary users have to perform multiple spectrum handoff utilizing multiple channels in the most energy-efficient manner.

Based on the decision time of selecting target channels, spectrum handoff mechanisms can be generally classified into two categories: proactive-decision and reactive-decision spectrum handoff [11]-[15]. Correspondingly,the target channel selection approaches can be categorized into proactive pre-determined and reactive on-demand channel selection [16].In [17], an optimal target channel sequence was designed by the proactive pre-determined channel selection, which could minimize the cumulative handoff delay in CRNs. However,it did not consider energy efficiency problem during the whole transmission and handoff process. In [18], an energy-efficient design of sequential channel sensing was studied,which included optimal sensing strategy, power allocation and sensing order. However, it did not investigate spectrum handoffs of the SU during communication process. In [19], a handoff delay was optimized by proposing a spectrum handoff scheme based on Recommended Channel Sensing Sequence, subject to the sensing reliability and link maintenance constraints. However, it did not consider energy efficiency and multiple spectrum handoffs problem. In [20], the authors argue that energy efficiency and spectrum efficiency are both important in CRNs. In [21], the energy efficiency opportunistic spectrum access strategy was investigated on the basis of an OFDM-based CRN with multiple SUs. However, both did not consider spectrum handoffs.

In this paper, we consider energy efficiency and multiple spectrum handoffs in totality and propose working solutions. The novelty is that we address a new and more generalized spectrum handoff problem in CRNs which has not been addressed in current literatures, where energy efficiency, multiple spectrum handoffs involving multiple channels, and the effects of primary user’s arrival and service rate are simultaneously considered. Then, we build an energy-based spectrum handoff model and formulate an energy efficiency optimization problem. In order to solve the optimization problem, we firstly transform the ratio-type energy efficiency optimization problem into a parametric formulation by making use of the techniques of fractional programming.Finally, we propose an algorithm combining the dynamic programming algorithm and the bisection (DPB) algorithm to obtain the optimal energy-efficient target channel sequence to maximize the energy efficiency during the whole communication process. Our contributions are that we build an energy-based spectrum handoff model and obtain the optimal energy-efficient target channel sequence.

The rest of this paper is organized as follows. In Section II, the system assumptions and energy efficiency definition are presented and the energy efficiency optimization problem is formulated. In Section III, the energy efficiency optimization problem is solved and the algorithm flow diagram is presented. In Section IV, simulation results are analyzed.Finally, some conclusions are drawn from the present investigation in Section V.

II. SYSTEM DESCRIPTION

We assume that the available spectrum is divided intoLindependent channels. The system is time-slotted. Every SU equips with two sets of antenna, one is directional antenna for transmitting or receiving signals while the other one is omnidirectional antenna for spectrum detection．In order to guarantee QoS of the PU, a SU has to suspend its data transmission and execute in-band detection, with sensing timeτ, to verify whether the channel is busy at the beginning of every time slot. If the sensed channel is idle, the SU will resume its data transmission. Letrepresent the idle and busy status of thechannel, with.Accordingly,denotes the detection result of thechannel, idle and busy, respectively. Hence, the detection probabilityand false alarm probabilityof thechannel can be expressed as

For energy detection, given the detection probabilityand the false alarm probabilitythe minimum sensing time for thekth channel can be written as

In addition, we investigate multiple spectrum handoffs by using preemptive resume priority (PRP) M/G/1 queuing approach.Every channel has two kinds of queues,high-priority and low-priority queue. The PUs and SUs enter the high-priority and low-priority queue respectively, obeying first-comefirst-served (FCFS) principle in each queue.Only when the PU in the high-priority queue accomplishes data transmission, SUs in the low-priority queue can access and start or resume their data transmission. Therefore,a SU may encounter multiple interruptions during its data transmission process. When an interruption happens, the SU has to suspend its data transmission and decide to stay on the current operating channel or switch to another channel based on the target channel sequence.For proactive-decision spectrum handoff, the target channel sequence is determined before data transmission.

Without loss of generality, letMdenote the allowable maximum number of interruptions for a SU andrepresent the target channel after encountering theith interruption. If the number of interruptions is more thanM, the SU connection will be dropped. Note that the initial channelc0is pre-assigned to SU for balancing the payload on every channel through spectrum decision algorithms [23]. Our goal is to design an optimal target channel sequenceto maximize the SU’s energy efficiency during the whole data transmission process. Define energy efficiency as follows [18]:

where the numerator represents the number of delivered bits per hertz and the denominator is the total energy consumption for switching,idling, transmitting and sensing. In order to formulate the specific energy efficiency expression, the denominator expression will be derived for two handoff cases: 1) staying on current operating channel; 2) switching to another channel. Without loss of generality, we investigate the energy consumption expectation betweenith andinterruption.

2.1 Energy efficiency analysis for staying on current operating channel

2.2 Energy efficiency analysis for switching to another channel

The energy consumption expectation for switching iswheredenotes the switching power. For the energy consumption expectation for sensing, we have

In addition, the energy consumption expectation for transmittinghas the same expression asin the first case. So, we have (13) shown in the bottom at this page.

2.3 Energy efficiency analysis for the whole spectrum handoff process

As a result, the total energy consumption expectation of the SU resulting from spectrum handoff can be eventually expressed aswhen the SU stay on the current operating channel; Alternatively, we havewhen the SU switch to another channel. Then, we consider the number of delivered bits per hertzon channelwhich can be expressed as

and

To sum up, the average energy efficiency can be expressed as

III. ENERGY-OPTIMAL TARGET CHANNEL SEQUENCING

In this section, we use the dynamic programming (DP) optimization technique to solve the energy efficiency maximization problem. We view every interruption as a stage.For example, as the SU encounters theinterruption, it means that the SU stays at stageThe strategy at stagecan be represented byand one of possible strategy sets during the whole handoff process is denoted byand the optimal strategy set is denoted byThe initial channel is pre-assigned before data transmission. However, the before-mentioned maximization problem is the ratio of two additive functions, which is difficult to be solved by DP directly. Therefore, we need to transform the ratio-type objective function into a parametric formulation [25] by making use of the solution techniques of fractional programming(FP). The new function at stageis defined as follows:

The physical meaning of the new problem can be interpreted as: ifcan be viewed as the monetary reward of the process, then the parametercan be viewed as the price for consuming per unit of energy. So,can be interpreted as the expected net reward of the whole process. Obviously, different prices for energy consumption may produce different parametric problems, resulting in different optimal target channel sequencesIn order to getwe firstly suppose that the value ofis known. As a result, we have built an additive and separable objective function by the parametric formulation so that DP algorithm can be used directly. Therefore, the new problem can be solved as follows:

In the following, we will find the optimal parameterto maximizeandso thatcan be obtained by applying DP algorithm. Firstly,we will give two propositions and prove them.

Proposition 1:Denote the optimal strategy set of the original problem asThen, we haveif and only if

Proof:See Appendix A.

Proposition 2:is a monotonously decreasing function ofβ.

Proof:See Appendix B.

Apparently, we can use bisection algorithm to explore the maximum ofover an intervalfor its monotonicity.Note that the intervalis known to contain. For eachβwithin the interval, we calculateby applying DP algorithm untilwhereεis a small constant used to take control of the convergence precision of the bisection algorithm. Then, replacingβin Eq. withwe can obtain the optimal energy-efficient target channel sequence for the original problem.

The DPB algorithm combines the DP algorithm and the bisection algorithm. Thus, we record it as DPB algorithm. The flowchart is presented in Fig. 1.

IV. SIMULATION AND DISCUSSION

We consider a SISO system withslow fading channels. On thechannel, we havewheredenotes the received signal andstands for the normalized transmitted signal.is the Rayleigh fading channel coefficient with complex Gaussian distributionis the additive white Gaussian noiseis the average SNR at receive antenna. The worst received SNR of PU’s signal is -15 dB, which is used to determine the minimum sensing time. The sampling frequency is, the bit error rate isand the time slot isbased on the IEEE 802.22 WRAN standard [27]. Let the detection probability and the false alarm probability berespectively. We set,for all channels, andNote that all parameters in this paper are set based on [16]-[18].

Firstly, we investigate the effects of the number of spectrum handoffs on the SU’s data transmission. Letdenote the service time of the newly arriving SU, which is exponentially distributed. Referring to [17],

Thus, the probability that the SU can finishits data transmission withinnspectrum handoffs can be defined as

Algorithm 1 Calculation offor given

Algorithm 1 Calculation offor given

Fig. 1 The exact flow diagram of DPB algorithm

4.1 Eff ects of number of spectrum handoff

Fig. 2 Effects of number of spectrum handoff on probability that SU can finish its data transmission

Fig. 3 Convergence performance of the DPB algorithm

4.2 Convergence performance of the proposed algorithm

Fig. 3 shows the effect of the iteration number of bisection algorithm on the maximum energy effi-ciency, where we set (ρ1,ρ2,ρ3,ρ4)= (6,7,8,9)(dB),βmax= 50(bits/Hz/Joule),βmin= 0(bits/Hz/Joule), andε=10-4. From Fig. 2, we have that whenandλp=0.02 (arrivals /slot), the probability that the SU finishes its data transmission within 6 spectrum handoffs is approximately 1. So, the allowable maximum spectrum handoff timeMis set to be 6. It can be seen from Fig. 3 that the maximum energy efficiency of the SU is determined by the pre-assigned initial channel and becomes bigger with higher SNR. For instance, when the initial channel of the SU is set to channel 4, the maximum energy efficiency converges to 14(bits/Hz/Joule). When the initial channel of the SU is set to channel 2, the maximum energy efficiency converges to 10.4 (bits/Hz/Joule).In addition, the calculation result by applying Eq. will determine whether the line goes up or down until the next literation. For example,at the second iteration, one of the four lines,c0=4, goes up until the third iteration. The reason is thatwhere(bits/Hz/Joule) represents the value ofat the second iteration. Based on the algorithm flow diagram of the DPB diagram, if(bits/Hz /Joule) at the third iteration. Thus,=18.75 (bits/Hz /Joule),and the linec0=4 goes up between the second and third iteration. On the contrary, at the second iteration, the other lines (c0=1,c0=2 andc0=3) go down until the next iteration. Furthermore, no matter which channel is chosen to be the initial channel, the energy efficiency will rapidly converge to the maximum energy efficiency value after approximately 8 iterations.

4.3 Comparisons among diff erent algorithms

4.4 Complexity analysis

Fig. 4 Performance comparisons among different algorithms

Fig. 5 Performance comparisons among different algorithms

Fig. 6 Complexity comparisons among different algorithms

In the following, we analyze the algorithm complexity of before-mentioned algorithms.Based on the analysis of Section III, we obtain that the complexity of the DPB algorithm can be expressed asBy the exhaustive search algorithm, the SU will search for all possible target channel sequences. So, the complexity of the exhaustive search algorithm isAccording to [17], the complexity of the minimum cumulative delay algorithm can be written asAs for the random algorithm, the complexity is directly written asAs shown in Fig. 6, comparing with the exhaustive research algorithm, the proposed DPB algorithm dramatically reduces the complexity and obtains the same performance.Comparing with the minimum cumulative delay algorithm, the ratio-type structure of the DPB algorithm increases the complexity.However, the DPB algorithm has better performance than the minimum cumulative delay algorithm. The random algorithm has the worst performance with the lowest complexity.

V. CONCLUSION

In this paper, we address a new and more generalized spectrum handoff problem in CRNs and simultaneously consider energy efficiency,multiple spectrum handoffs involving multiple channels, and the effects of PU’s arrival and service rate. Then, we design the target channel sequence for a cognitive radio network in which the secondary users have to perform multiple spectrum handoff utilizing multiple channels in the most energy-efficient manner.In our proposed algorithm, we adopt the proactive predetermined channel selection strategy, and a combination of dynamic programming with bisection algorithm to perform the optimization search. Simulation results show that our proposed algorithm is able to find the optimal solution within 8 bisection iteration steps in random slow fading channels, upon which the system is able to achieve higher energy efficiency (in bits/Hz/Joule) than the prior art, with the efficiency gain increasing with the channel SNR. Furthermore, with higher initial channel SNR, the advantage of the proposed DPB algorithm is greater. In our future work, we will investigate energy efficiency of spectrum handoff based on preemptive repeat identical queue model. In addition, potential security issues [28]-[30] are also valuable to be considered.

ACKNOWLEDGEMENTS

We gratefully acknowledge the guidance of Prof. Guan Yong Liang from School of Electrical and Electronic Engineering, Nanyang Technological University. This work is supported by Heilongjiang Province Natural Science Foundation (Grant No. F2016019),National Natural Science Foundation of China(Grant No. 61571162), Major National Science and Technology Project (2015ZX03004002-004) and China Postdoctoral Science Foundation (Grant No. 2014M561347).

Appendix A

Proof of proposition 1:

At first, we prove the sufficiency. Ifwe can obtain that

and

Thus, the maximum energy efficiencycan be achieved by using the optimal target channel sequenceAs a result,the proposition 1 is proved.

Appendix B

Proof of proposition 1:

Assume two arbitrary parametersandTheir corresponding parametric problem can be denoted asandrespectively. Their corresponding optimal target channel sequences are, respectively,

and

Then,

[1] J Mitola，J Guerci，J Reed, et al., “Accelerating 5G QoE via public-private spectrum sharing”,IEEE Communications Magazine, vol. 52, no. 5,pp 77-85, May 2014.

[2] F.F Gao, K.Q Zhang, “Enhanced multi-parameter cognitive architecture for future wireless communications,”IEEE Communications Magazine,vol. 53, no. 7, pp 86-92, July 2015.

[3] Z Li, B.Y Liu, J.B Si, et al., “Optimal satisfaction degree in energy harvesting cognitive radio networks”,Chinese Physics B, vol. 24, no. 12, Article ID: 128401, October 2015.

[4] Z.A Deng, X Liu, M Jia, et al., “Sensing time optimization in energy-harvesting cognitive radio with interference rate control”,China Communications, vol. 13, no. 1, pp 47-56, January 2016.

[5] A.D Domenico, E.C Strinati, M.D Benedetto, “A survey on MAC strategies for cognitive radio networks”,IEEE Communications Surveys &Tutorials, vol. 14, no. 1, pp 21-44, First Quarter 2012.

[6] I Christian, S Moh, C Ilyong, et al., “Spectrum mobility in cognitive radio networks”,IEEE Communications Magazine, vol. 50, no. 6, pp 114-121, June 2012.

[7] J.T Zhao, W Feng, M Zhao, et al., “Coordinated multi-user spectrum sharing in distributed antenna-based cognitive radio systems”,China Communications, vol. 13, no. 1, pp 57-67, January 2016.

[8] Z.N Sun, X Xu, X.M Chai, et al., “The new architecture with time-spatial consistency for 5G networks”,China Communications, vol. 13, no.1, pp 68-79, January 2016.

[9] X.L Yang, X.Z Tan, K Guan, “Spectrum handoff model based on preemptive queuing theory in cognitive radio networks”,Acta Physica Sinica,vol. 64, no. 10, Article ID: 108403, February 2015.

[10] C Yin, X.Z Tan, L Ma, et al., “Global proportional fairness scheduling algorithm based on spectrum aggregation in cognitive radio”,Acta Physica Sinica, vol. 63, no. 11, Article ID: 118402,May 2014.

[11] Z.S Zhang, K.P Long, J.P Wang, “Self-organization paradigms and optimization approaches for cognitive radio technologies: A survey”,IEEE Wireless Communications, vol. 20, no. 2, pp 36-42, April 2013.

[12] V Balaiah, “Modeling and analysis for proactive spectrum handoff in cognitive radio ad hoc networks with improved throughput and security”,International Journal of Wired & Wireless Communications, vol. 2, no. 2, pp 5-9, 2014.

[13] A Azarfar, J.F Frigon, B Sanso, “Delay analysis of multichannel opportunistic spectrum access MAC protocols”,IEEE Transactions on Mobile Computing, vol. 15, no. 1, pp 92-106, March 2015.

[14] S Zheng, X Yang, S Chen, at el., “Target channel sequence selection scheme for proactive-decision spectrum handoff”,IEEE Communications Letters, vol. 15, no. 12, pp 1332-1334, December 2011.

[15] C.W Wang, L.C Wang, “Analysis of reactive spectrum handoff in cognitive radio networks”,IEEE Journal on Selected Areas in Communications,vol. 30, no. 10, pp 2016-2028, November 2012.

[16] L.C Wang, C.W Wang, C.J Chang, “Modeling and analysis for spectrum handoffs in cognitive radio networks”,IEEE Transactions on Mobile Computing, vol. 11, no. 9, pp 1499-1513, September, 2012.

[17] L.C Wang, C.W Wang, and C.L Chang, “Optimal target channel sequence design for multiple spectrum handoffs in cognitive radio networks”,IEEE Transactions on Communications, vol. 60,no. 9, pp 2444-2455, September 2012

[18] Y.Y Pei Y.C Liang, K.C Teh, et al., “Energy-efficient design of sequential channel sensing in cognitive radio networks: Optimal sensing strategy,power allocation, and sensing order”,IEEE Journal on Selected Areas in Communications, vol.29, no. 8, pp 1648-1659, September 2011.

[19] C.Y Wu, C He, L.G Jiang, “Spectrum handoff scheme based on recommended channel sensing sequence”,China Communications, vol. 10,no. 8, pp 18-26, August 2013.

[20] X Hong, J Wang, C.X Wang, et al., “Cognitive radio in 5G: A perspective on energy-spectral efficiency trade-off”,IEEE Communications Magazine, vol. 52, no. 7, pp 46-53, July 2014.

[21] C Xiong, L Lu, G.Y Li, “Energy-efficient spectrum access in cognitive radios”,IEEE Journal on Selected Areas in Communications, vol. 32, no. 3,pp 550-562, March 2014.

[22] H Pradhan, S.S Kalamkar, A Banerjee, “Sensing-throughput tradeoff in cognitive radio with random arrivals and departures of multiple primary users”,IEEE Communications Letters, vol.19, no. 3, pp 1415-418, March 2015.

[23] L.C Wang, C.W Wang, F Adachi, “Load-balancing spectrum decision for cognitive radio networks”,IEEE Journal on Selected Areas in Communications, vol. 29, no. 4, pp 757-769, April 2011.

[24] A Garcia-Armada, “SNR gap approximation for M-PSK-based bit loading”,IEEE Transactions on Wireless Communications, vol. 5, no. 1, pp 57-60, January 2006.

[25] M Sniedovich.Dynamic programming: foundations and principles, 2nd ed. Chemical Rubber Company Press, 2010.

[26] S Boyed, L Vandenberghe.Convex optimization.Cambridge University Press, 2004.

[27] IEEE Std. 802.22-2011,Wireless regional area network (WRAN) specific requirements part 22:cognitive wireless RAN medium access control(MAC) and physical layer (PHY) specifications:policies and procedures for operation in the TV bands, IEEE, July 2011.

[28] Y.L Zou, J Zhu, X.B Wang, et al., “A survey on wireless security: Technical challenges, recent advances, and future trends”,Proceedings of the IEEE, vol. 104, no. 9, pp 1727-1765, September 2016.

[29] G.R Ding, J.L Wang, Q.H Wu, et al., “Robust spectrum sensing with crowd sensors”,IEEE Transactions on Communications, vol.62, no. 9,pp 3129-3143, September 2014.

[30] G.R Ding, Q.H Wu, Y.D Yao, et al., “Kernel-based learning for statistical signal processing in cognitive radio networks: Theoretical foundations,example applications, and future directions”,IEEE Signal Processing Magazine, vol. 30, no. 4,pp 126-136, July 2013.