Performance analysis of spectrum sensing schemes based on energy detector in generalized Gaussian noise

Rui Gao,Peihan Qi,Zhenghua Zhang

DOI: https://doi.org/10.1016/j.sigpro.2020.107893

IF: 4.729

2021-04-01

Signal Processing

Abstract:<p>Spectrum sensing in the presence of non-Gaussian noise is a challenging task for cognitive radio networks (CRNs). Among all spectrum sensing approach, energy detection (ED) is the dominantly used one and becomes the object of comparison of others. Motivated by these, in this paper, we provide a performance analysis of ED in generalized Gaussian noise (GGN). We make full use of the stochastic characteristics of generalized Gaussian distribution and the central limit theorem (CLT) to derive the probabilities of detection and false alarm. By using the Berry-Esseen theorem, the convergence of generalized Gaussian distribution to Gaussian distribution is further given. Based on the related Berry-Esseen bound, the number of samples required to ensure a given approximation error was derived. Besides, a performance analysis about the influence of noise uncertainty is proposed, which shows that the SNR Wall of ED in GGN is the same as that in Gaussian noise. For cooperative spectrum sensing with ED in GGN, we provide targeted approach for different cases. Simulation results validate our analysis, and show that the proposed fusion rule can improve ED in CRNs.</p>

engineering, electrical & electronic