An Covariance Spectrum Sensing Algorithm Based on DMM / MME Mixture Features

With the research of Random Matrix Theory (RMT), many RMT -based covariance detection algorithms have been proposed and applied to spectrum sensing. As a blind detection algorithm, the detection performance in practice is often susceptible to environmental factors due to the approximation of threshold calculation. In view of this problem, some scholars have proposed to use machine learning to improve the covariance detection, so as to obtain high environmental adaptability. In this paper, we propose a covariance spectrum sensing algorithm with judgment statistics in maximum-minimum eigenvalue algorithm (MME) and difference between the maximum eigenvalue and the minimum eigenvalue algorithm (DMM) as hybrid features, which takes the both the quotient and difference of the maximum minimum eigenvalue of the covariance matrix as the feature vector for classification, and then trained and classified by SVM. Simulation results show that the proposed algorithm improves significantly compared to the traditional covariance detection. Moreover, compared with the existing covariance in combination with SVM detection algorithm, the two features selected in this paper have better differentiation and better environmental adaptability under low SNR.