Bayesian Energy Detection Based on Temporal Persistence

Abstract

Abstract. Energy detection is one of the classical methods for spectrum sensing in Cognitive radio (CR). Previous research on energy detection is almost based on single time slot, while the communication process of the primary user (PU) is hardly completed in one time slot. In this study, the authors consider 2-slot temporal persistence (TP) that PU maintains the same state (absence or presence) for at least 2 slots. Since we cannot know the actual state of PU in spectrum sensing, two kinds of TP results are obtained, based on which an improved TP-based Bayesian Energy Detection (ITPBED) is proposed. Simulation results show that, compared with TPBED, ITPBED scheme can achieve significant reduction in false alarm probability, missed detection probability and Bayesian cost when signal-to-noise ratio (SNR) is less than -10 dB; in other SNR regions, the performance of ITPBED scheme is also superior to Bayesian Energy Detection (BED) scheme. IntroductionIn recent years, with the development of wireless communication technology, the majority of spectrum resource has been allocated for specific use. According to the spectrum report published in 2010 by The Federal Communications Commission (FCC) [1], the available spectrum resources are on the verge of extinction. However, many researchers analyze and monitor the use of radio spectrum, finding that significant amount of spectrum remains underutilized, and even completely idle [2]. In order to solve this problem, authors in [3, 4] put forward secondary market of spectrum to improve the spectral efficiency, in which context cognitive radio (CR) was proposed by Mitola and Maguire [5].CR is a promising solution to the spectrum scarcity issue [6] via enabling the secondary user (SU) to take chances to access the unutilized spectrum without causing interference to the primary user (PU). Consequently, SU has to continuously and reliably detect whether PU is present or not. There are three classical spectrum sensing (SS) technologies, matched filter detection [7], cyclostationary feature detection [8] and energy detection (ED) [9]. Among them, ED is found to be the simplest and the most widely used SS mechanism since it does not require prior information about PU signals. ED senses the presence of PU signal by accumulating the energy of the received signal over a specific time interval [10]. There exist many drawbacks for ED, the threshold highly depends on the environment conditions [11], robustness is poor and the estimation error due to noise may degrades sensing performance significantly at low signal-to-noise ratio (SNR) values [10,12].In [13], the authors propose the concept of PU's temporal persistence (TP) and prove that the prediction probabilities of PU's state are at least not worse than the prior probabilities when considering TP. The concept of TP derived from the research of target recognition in the field of image processing. In practice, PU can hardly complete its communication process in a single slot, which means that if PU becomes present in a slot to begin ...