Spectrum Sensing in Low SNR Regime via Stochastic Resonance

arXiv:0906.0739v1 [cs.IT] 3 Jun 2009 Spectrum Sensing in Low SNR Regime via Stochastic Resonance Kun Zheng, Husheng Li, Seddik M. Djouadi and Jun Wang Abstract—Spectrum sensing is essential in cognitive radio to enable dynamic spectrum access. In many scenarios, primary user signal must be detected reliably in low signal-to-noise ratio (SNR) regime under required sensing time. We propose to use stochastic resonance, a nonlinear filter having certain resonance frequency, to detect primary users when the SNR is very low. Both block and sequential detection schemes are studied. Simulation results show that, under the required false alarm rate, both detection probability and average detection delay can be substantially improved. A few implementation issues are also discussed. Keywords—spectrum sensing, stochastic resonance I. INTRODUCTION In cognitive systems [1], spectrum sensing is one of the key issues. In order to effectively use the scarce spectrum resource, secondary users (SUs), which have no license to the frequency spectrum band, need to opportunistically share the spectrum with primary users (PUs) having license. An important requirement in cognitive radio systems is that SUs should not cause significant interference to PUs. This requires that SUs have the capability to sense the spectrum environment periodically and detect PUs’ presence reliably. The most challenging problem in spectrum sensing is how to detect very weak signals of PUs, or in other words, how to improve the signal-to-noise ratio (SNR) of received signal. One of the reasons why it is necessary to sense weak signals in cognitive radio systems is the existence of the Hidden Primary User Problem [2]. An example is illustrated in Fig. 1, where the PU receiver suffers substantial interference from the SU transmitter while the SU transmitter is unaware of the existence of the PU receiver. In traditional wireless networks, the hidden user problem can be alleviated by a clear-to-send (CTS) signaling from the receiver. However, in cognitive radio systems, PU systems cannot be modified and the risk of violat- ing a dummy PU receiver cannot be alleviated via signaling. The only reliable approach is to improve the sensitivity of the SU’s spectrum sensing such that it can detect the PU transmitter’s signal once it is located within the interference range of the PU receiver. Thus it requires SUs to detect weak signals of PUs under required sensing time and achieve desired K. Zheng, H. Li and S. M. Djouadi are with the department of electrical engineering and computer science, the University of Tennessee, Knoxville, TN. J. Wang is with the National Key Lab of Communications, Univer- sity of Electronic Science and Technology, Chengdu, P. R. China. This work was supported by the National Science Foundation under grant CCF- 0830451, High-Tech Research and Development Program of China under grant 2007AA01Z209 and National Basic Research Program (973) under grant 2009CB320405. probabilities of detection and false alarm. For example, in 802.22 standard, the signal of PU must be detected in no more than 2 seconds with less than 10% probability of false alarm and greater than 90% probability of detection. Meanwhile, the IDT (Incumbent Detection Threshold) are -107 dBm and -116 dBm for wireless microphones and TV services respectively [3]. SU SU PU PU Interference Fig. 1. Illustration of Hidden Primary User Problem Several typical approaches can be used for spectrum sensing in cognitive radio systems: • Energy Detection: It is extensively used in radiometry and is the most popular approach of spectrum sensing due to its low computational and design complexities [4] [5] [6]. Although energy detection based spectrum sensing is simple to implement and can be applied to any type of PU signal, there are still many drawbacks: (1) The energy detector does not distinguish signals between PUs and other SUs. Thus it may result in false alarm due to interference from other SUs’ transmission, especially when SUs are not well time synchronized. (2) The usage of energy detection is limited by SNRwall [7], which results from the noise uncertainty. Under this SNR threshold, the signal is found to be completely non- detectable by energy detection [7]. • Feature Detection: If we already know the specific char- acteristics of primary signal, such as cyclostationarity, pilot, radio identification and etc., feature detection can be applied [5] [8] [9]. For example, ATSC signal has many features that can be applied to feature detection. The main problem for feature detection method is the required large number of samples, thus incurring significant detection delay. • Matched Filter: When transmitted signal is known at the TABLE I COMPARISON OF TRADITIONAL DETECTION APPROACHES AND WHY SR CAN BE APPLIED Energy Detection Pros simple to implement, applicable to any signal type Cons suffer to low SNR can’t distinguish signal and interference How SR works increase input SNR to overcome SNR

…(Full text truncated)…

This content is AI-processed based on ArXiv data.