Spectrum Sensing Detection Algorithm Based on Eigenvalue Variance
WANG Zi-li, Xiaoou Song, Wang Xiaorong · 2019
Energy detection as a kind of the traditional spectrum sensing is susceptible to the uncertainty of noise variance. Its performance is poor at low SNR and there is a "signal to noise ratio wall" effect. Eigenvalue detection of spectrum sensing, the same as energy detection, does not require any prior information of the signal and can achieve better detection performance at low SNR. The classical eigenvalue detection methods have maximum-minimum eigenvalue ratio algorithm (MME), maximum-minimum eigenvalue difference algorithm (DMM) and so on. But these algorithms only use part of eigenvalues and do not adequately reflect the characteristics of all eigenvalues. In this paper, a spectrum sensing algorithm based on eigenvalue variance is proposed. The variance which reflects the overall fluctuation of eigenvalues is selected as the observation statistic and the theoretical threshold of the algorithm is derived. The simulation proves that the detection performance of the algorithm is better than the MME algorithm when there is no noise variance uncertainty and the detection performance of this algorithm is better than ED algorithm when there is uncertainty of noise variance.