Parzen window entropy based spectrum sensing in cognitive radio

Abstract

In this paper, we propose a Parzen window entropy based spectrum sensing algorithm for enhancing the signal-to-noise ratio (SNR) wall of cognitive radio primary user detection. We compute the information entropy using a non-parametric Kernel Density Estimation (KDE) method. Single node sensing is extended to cooperative sensing using the weighted gain combining (WGC) fusion method. The weights of WGC technique are computed using a Differential Evolution(DE) algorithm and compared with the log-likelihood ratio (LLR) method. In addition, the detection performance of the proposed Parzen window entropy is compared with Shannon entropy and energy detection techniques. We consider a DVB-T signal with Additive White Gaussian Noise (AWGN) subjected to Rayleigh fading under noise uncertainty as a primary user signal for simulation. The simulation result reveals that in the case of a single node and cooperative sensing, the proposed method achieves SNR wall of −19 dB and −24 dB respectively at the probability of false alarm 0.1.