Non-uniform quantized exponential entropy-based spectrum sensing algorithm in cognitive radio

Spectrum sensing is one of the key technologies in cognitive radio system. Sensitivity to noise uncertainty is a fundamental limitation of current spectrum sensing strategies in detecting the presence of primary users in cognitive radio. Because of noise uncertainty, the detection performance of traditional detectors such as energy detector, matched filter and even cyclostationary detectors deteriorate rapidly at low signal-to-noise ratio. Without accurate estimation of noise power, an absolute `SNR wall' exists in traditional detectors below which robust detection is impossible, no matter how long the observations are. Shannon entropy-based detection schemes arouse widespread attention in recent years due to its property of effective anti-noise uncertainty. However, the Rayleigh distribution of spectral entropy is non-maximum entropy distribution in the absence of primary users, and the detection failure phenomenon caused by misconvergence of Shannon entropy estimation are the two major serious defects in Shannon entropy-based detectors. These problems restrict the improvement of detection performance. In this paper, we first address the problem of the maximum entropy distribution problem in the absence of primary users by quantizing sections of spectrum amplitude in a non-uniform manner. Subsequently, the concept of exponential entropy is introduced into the spectrum sensing to avoid detection failure problem in Shannon entropy, and a novel non-uniform quantized exponential entropy (NQEE) detector is proposed. Simulation results verify that the detection performance of the improved local entropy-based detector is superior to previous Shannon entropy-based detectors and is proved to be robust to noise power uncertainty. In addition, the novel non-uniform quantized exponential entropy-based scheme need no prior information of the primary users' signal.