GLRT-Based Spectrum Sensing for Cognitive Radio

Abstract

In this paper, we propose several spectrum sensing methods designed using the generalized likelihood ratio test (GLRT) paradigm, for application in a cognitive radio network. The proposed techniques utilize the eigenvalues of the sample covariance matrix of the received signal vector, taking advantage of the fact that in practice, the primary signal in a cognitive radio environment will either occupy a subspace of dimension strictly smaller than the dimension of the observation space, or have a spectrum that is non-white. We show that by making various assumptions on the availability of side information such as noise variance and signal space dimension, several feasible algorithms result which all outperform the standard energy detector.