Elsevier

Computer Networks

Volume 58, 15 January 2014, Pages 87-98
Computer Networks

Sequential sensing based spectrum handoff in cognitive radio networks with multiple users

https://doi.org/10.1016/j.comnet.2013.08.025Get rights and content

Abstract

Spectrum handoff occurs when the primary users appear in the licensed spectrum temporarily occupied by the secondary users and aims to help the secondary users to vacate the spectrum rapidly and to resume transmission on new selected available channels. However, a spectrum handoff policy that comprehensively considers spectrum sensing, target channel selection as well as spectrum estimation has yet to be developed. In this paper we present a sequential sensing based spectrum handoff policy for multiple-user cognitive radio networks. First, we select the appropriate candidate channels for each secondary user, then their associated optimal sensing order together with the best target handoff channel is determined through sequential sensing based on Dynamic Programming (DP). Note that many spectrum handoff will occur during one secondary user transmission and our objective is to minimize the total number of spectrum handoff. The sequential sensing based spectrum handoff policy is evaluated through a comprehensive simulation study. The results reveal significant improvements in the system performance by reducing the number of spectrum handoff over conventional approaches. Moreover, our proposed DP method can significantly lower the computational complexity compared to exhaustive search and common DP (performing sequential sensing over all the channels in the system using Dynamic Programming).

Introduction

The rapid growth of wireless technologies has led to a dramatic increase in demand for spectrum. However, according to a report by Federal Communications Commission (FCC), most of the allocated spectrum is significantly under-utilized [1] [2] due to the inefficient fixed spectrum allocation policy. Cognitive radio is an emerging technique to mitigate the spectrum scarcity by allowing the secondary users to temporarily utilize the licensed spectrum unoccupied by the primary user [3] [4]. Due to the high priority of the primary user, the secondary user is required to vacate the occupied channel when the primary user appears and determine a new suitable channel to resume its unfinished transmission. This process is referred to as spectrum handoff. Compared with other major functionalities: spectrum sensing, spectrum decision and spectrum sharing, spectrum handoff is less well explored in the research community.

In general, the spectrum handoff can be categorized into two types [5]: (1) Proactive spectrum handoff which decides the target channel before the interruption occurs according to the long term traffic statistics [6], [7], [8]. (2) Reactive spectrum handoff which selects the target channel when it is required via spectrum sensing [9], [10]. One issue related to proactive spectrum handoff is that the preselected channel may be no longer available at the moment that spectrum handoff is initiated. While for reactive spectrum handoff, additional time is required for spectrum sensing to search an idle channel, which results in an increase in handoff delay. In our paper, we propose a spectrum handoff mechanism which can reliably determine an idle target channel through sequential spectrum sensing. This mechanism can help to resolve the obsolescent channel issue in proactive spectrum sensing handoff where the preselected channel may no longer be available.

On the other hand, our spectrum sensing handoff is triggered when the residual idle time (which is defined as the duration from the time instant that the channel is detected to be idle and able to be utilized by the secondary user until the time instant that the interrupting event occurs and a spectrum handoff is required) of the current utilized channel reaches a threshold. By doing so, spectrum sensing and spectrum analysis can be overlapped with the ongoing transmission. As a consequence, the high handoff delay occurring in the reactive spectrum handoff can be addressed. Compared with other work on spectrum handoff where the secondary user greedily selects the target channel which either results in minimum transmission latency [11], [12], [13] or has the highest probability of being idle [17], in our paper, we choose the target channel with maximum residual idle time as our overall objective to reduce the number of spectrum handoff as much as possible.

To implement a better spectrum handoff in the event of primary user activity with seamless communication, wideband sensing is essential for designing a maximally effective cognitive network. It can detect multiple opportunities and enable the choice of the best available channel. However, the literature of wideband spectrum sensing for cognitive radio networks is very limited. In [33], sequential sensing is introduced, in which a wideband radio channel is sensed using tunable narrowband bandpss filter at the RF front-end to sense one narrow frequency band at a time. There have also been studies on sensing different frequency bands simultaneously. In [34], multiband joint detection approach is proposed where the wideband channel is divided into K non-overlapping narrow subbands. In [35], an optimal algorithm is presented for wideband spectrum sensing with the aim of maximizing the achievable throughput of the secondary user while keeping the interference with the primary network bounded to a reasonably low level.

One important issue related to wideband spectrum sensing is the optimal sensing order problem. Most prior work on sensing order issue [14], [15], [16] only considers single user or two-user scenario. In [14], the channel sensing problem is formulated as an optimal stopping rule problem and it is shown that the optimal sensing order does exist in some special scenarios for single user case. In [16], the authors extend the sensing order issue to two-user multi-channel case. The problem is still open for multiple secondary users case. In [17], [18], the authors propose the spectrum switching algorithm according to descending order of channel idle probability. When primary user appears on the current operating channel, the secondary user must vacate the channel immediately and resume its transmission on the channel which has the highest idle probability. This approach has proven that the spectrum switching delay can be significantly reduced. However, if multiple secondary users perform spectrum sensing at the same time, switching channel according to the descending order of channel idle probability may no longer be optimal, since these selections will cause collisions among secondary users.

Recently, study on the optimization of spectrum sensing in cognitive radio network has attracted a lot of attention. There are two approaches that are commonly used to schedule the spectrum sensing [15]. The first one is periodic spectrum sensing [19], [20], [21], [22], in which the secondary users perform spectrum sensing at the beginning of each frame and transmit if the channel is detected as idle. Otherwise, the secondary users have to wait until the next frame. The other approach is sequential spectrum sensing [15], [16], [23]. In this case, spectrum is searched by sensing the channels one by one until an idle channel with satisfied quality is detected. This approach allows secondary users to explore diversity in the licensed spectrum. Hence, in the case that one channel is sensed to be busy, the secondary user can quickly continue to sense the next spectrum opportunity without waiting until the next frame as in the case of periodic sensing. Obviously, sequential sensing can have better performance than periodic sensing. Currently, the spectrum sensing techniques can be mainly classified into matched filter, covariance matrix based detection, cyclostationary feature detection and energy detection [36], each has its own disadvantages and advantages. Matched filter requires perfect knowledge of the PUs’ signaling features such as operating frequency, modulation type and order, which is not practised in reality. Cyclostationarity feature detection not only requires a priori knowledge of the signal characteristics, but also needs a much longer sensing duration for a channel which may be more than 20 ms [37]. Comparing with them, energy detection is the most common way of spectrum sensing due to its low computational and implementation complexity and less sensing time. In our paper, each secondary user may sequentially perform sensing up to N channels in a slot. Thus, a short sensing duration is reasonable. Due to the above reasons, the energy detection based sequential sensing is of our concern in this paper. We propose a sequential sensing based spectrum handoff scheme to determine the optimal sensing order as well as the best target handoff channel for each secondary user in a multiple-user cognitive network using DP.

Whenever a secondary user needs to switch channel, it must perform sequential sensing to discover a new opportunity with minimum delay so that the secondary user in the network can resume its communication quickly. However, one has to account for the fact that finding the optimal sensing order as well as the best target channel has huge computational complexity. To address this problem, instead of performing sequential sensing over all the channels in the network, we select candidate channels for each secondary user, which analyzes the tradeoff among the three key characteristics: (1) keep the probability of detecting at least one idle channel high; (2) reduce the sensing overhead and computational complexity as much as possible; and (3) avoid collision with other secondary users. Therefore, each secondary user only scans its associated candidate channels every time when the spectrum handoff is triggered. It is shown that the computational complexity is significantly reduced while the system performance is maintained. The contributions and significance of this paper are listed as follows:

  • (1)

    Instead of only considering single user or two users in a network, we consider multiple secondary users contending the spectrum for handoff. For the case of single user, it is shown that sensing the channels according to the descending order of primary-free probability is optimal. However, when multiple users are investigated, some factors e.g., activities of primary users and contention among secondary users, will together affect the optimal sensing order. We propose a sequential sensing based spectrum handoff for multiple-user cognitive radio networks. By overlapping the spectrum sensing and spectrum analysis with the ongoing transmission, the high handoff delay can be reduced, while the optimal sensing order as well as a reliable target channel with maximal residual idle time can be found.

  • (2)

    Although DP search can be used to find the optimal sensing order as well as the best target channel for each secondary user, it incurs huge complexity. We determine candidate channels for sequential sensing, which exploits the tradeoff between the spectrum opportunity and sensing overhead and computational complexity. It is shown that the system performance is maintained while the computational complexity is significantly reduced.

  • (3)

    To evaluate our proposed sequential sensing based spectrum handoff strategy, a comparison with conventional sequential sensing CSS [25] is conducted. Simulation results show that our proposed strategy outperforms CSS due to the fact that the optimal target handoff channel is selected, which results in a reduction in the number of spectrum handoff.

The rest of this paper is organized as follows. The system model is introduced in Section 2. The key part: the detailed design of our proposed sequential sensing based spectrum handoff strategy is elaborated in Section 3, in which a dynamic programming method is proposed to determine the optimal sensing order as well as the best target handoff channel with maximal residual idle time for each secondary user. The effect of error rate is evaluated in Section 4. Simulation results and evaluation are given in Section 5. Finally, Section 6 concludes the paper.

Section snippets

System model

We consider a cognitive radio network consisting of M secondary users and N channels (primary users), with channel indices (1, 2,  , N). The primary users are licensed holders and thus have absolute priority to interrupt the transmission of secondary user. On the other hand, the secondary user can only utilize the channel opportunistically, and as soon as a primary user is detected, the secondary user is forced to vacate the occupied channel. Thus spectrum handoff is initiated when the primary

The design of our proposed policy

The intelligent cognitive technologies can allow the secondary users to temporarily utilize the unused licensed spectrum. In order to enhance the utilization, our proposed approach required three major capacities: spectrum estimation, sequential spectrum sensing and spectrum handoff. Spectrum estimation is to measure the residual idle time of the current transmitting channel and determine when to trigger a spectrum handoff; sequential spectrum sensing is to discover a new spectrum opportunity

The effect of sensing error

Sensing errors such as false alarm and missed detection will degrade the performance of secondary user and primary user. Since we already set a target detection probability, the primary users will receive a satisfied protection. In the following, we will investigate the effect of false alarm on the actual residual idle time of spectrum handoff.

When a false alarm occurs in the sequential spectrum sensing, the idle channel will be declared as busy. The only decision at this stage is to sense the

Simulation

In this section, we present the simulation results and discussions for our proposed sequential sensing based spectrum handoff strategy. The parameters used in this system are as follows: the sampling frequency is fixed and set to fs = 6 MHz, the target detection probability α = 0.9, the upper and lower bounds of idle probability are set as Pmin = 0.9 and Pmax = 0.95, respectively. We assume that there are M = 4 secondary users in the network, while the number of channels varies between 2  N  10.

Conclusions

In this paper, we have studied the design of sequential sensing based spectrum handoff strategy in multiple-user cognitive radio networks. Most of the prior work on sequential sensing only considers single or two-secondary user scenario. We propose a sequential sensing which can determine optimal sensing order for each secondary user as well as best target channel for spectrum handoff by Dynamic Programming (DP). In addition, we select candidate channels for each secondary user, which jointly

Wenjie Zhang received his B.E. in Institute of Application Mathematics from University of Electronic Science and Technology of China, Chengdu, China in 2008. He is currently working toward the PhD degree at School of Computer Engineering, Nanyang Technological University, Singapore. His research interests include spectrum sensing and management in challenged cognitive networks, and throughput and delay analysis in wireless networks.

References (37)

  • L. Yang et al.

    Proactive channel access in dynamic spectrum networks

    Elsevier Phys. Commun.

    (2008)
  • Federal Communications Commission, Spectrum Policy Task Force Report, FCC 02-155, November...
  • G. Staple et al.

    The end of spectrum scarcity

    IEEE Spectrum

    (2004)
  • J. Mitola

    Cognitive radio: an integrated agent architecture for software defined radio

    (2000)
  • J. Mitola et al.

    Cognitive radio: making software radios more personal

    IEEE Pers. Commun.

    (1999)
  • L.C. Wang et al.

    Spectrum handoff for cognitive radio networks: reactive sensing or proactive sensing

    IEEE IPCCC

    (2008)
  • S. Srinivasa et al.

    The throughput potential of cognitive radio: a theoretical perspective

    IEEE Commun. Mag.

    (2007)
  • Q. Shi et al.

    Link maintenance protocol for cognitive radio system with OFDM PHY

    IEEE DySPAN

    (2007)
  • H. Su et al.

    Channel-hopping based single transceiver MAC for cognitive radio networks

    IEEE Ann. Conf. Inform. Sci. Syst. (CISS)

    (2008)
  • D. Willkomm et al.

    Reliable link maintenance in cognitive radio systems

    IEEE DySPAN

    (2005)
  • J. Tian, G. Bi, A new link maintenance and compensation model for cognitive UWB radio systems, in: International...
  • L.-C. Wang et al.

    A queueing-theoretical framework for QoS-enhanced spectrum management in cognitive radio networks

    IEEE Wirel. Commun. Mag.

    (2010)
  • L.-C. Wang et al.

    Modeling and analysis for spectrum handoffs in cognitive radio networks

    IEEE J. Sel. Areas Commun.

    (2011)
  • L.-C. Wang et al.

    Load-balancing spectrum decision for cognitive radio networks

    IEEE Trans. Mobile Comput.

    (2011)
  • H. Jiang et al.

    Optimal selection of channel sensing order in cognitive radio

    IEEE Trans. Wirel. Commun.

    (2009)
  • Y. Pei et al.

    Energy-efficient design of sequential channel sensing in cognitive radio networks: optimal sensing strategy power allocation and sensing order

    IEEE J. Sel. Areas Commun.

    (2011)
  • R. Fan et al.

    Channel sensing-order setting in cognitive radio networks: a two-user case

    IEEE Trans. Veh. Technol.

    (2009)
  • L. Wang, C. Wang, Spectrum handoff for cognitive radio networks: reactive-sensing or proactive-sensing, in: Proc. IEEE...
  • Cited by (0)

    Wenjie Zhang received his B.E. in Institute of Application Mathematics from University of Electronic Science and Technology of China, Chengdu, China in 2008. He is currently working toward the PhD degree at School of Computer Engineering, Nanyang Technological University, Singapore. His research interests include spectrum sensing and management in challenged cognitive networks, and throughput and delay analysis in wireless networks.

    Chai Kiat Yeo received the B.E. (Hons.) and M.Sc. degrees in 1987 and 1991 respectively, both in electrical engineering, from the National University of Singapore and the Ph.D. degree from the School of Electrical and Electronics Engineering, Nanyang Technological University (NTU), Singapore, in 2007. She was a Principal Engineer with Singapore Technologies Electronics and Engineering Limited prior to joining NTU in 1993. She has been the Deputy Director of Centre for Multimedia and Network Technology (CeMNet) in Nanyang Technological University (NTU), Singapore. She is currently Associate Chair (Academic) with the School of Computer Engineering, NTU. Her research interests include ad hoc and mobile networks, overlay networks, speech processing and enhancement.

    View full text