Enhanced Detection Algorithms Based on Eigenvalues and Energy in Random Matrix Theory Paradigm

This paper considers the problem of spectrum sensing in multi-antenna cognitive radio networks. Energy detection (ED) method for spectrum sensing does not require any information of the source signal and channel, as well as it is suitable for detecting independent identically distributed signals. Since covariance matrix catches the signal correlations well, the maximum eigenvalue detection (MED) method is more competitive than the ED method for correlated signals. Under the framework of random matrix theory, this paper firstly proposes two enhanced detection algorithms based on the maximum eigenvalue and energy of the signal to achieve performance improvement while preserving the advantages of the two algorithms. The proposed algorithms are a generalization of the ED and MED methods. To render the proposed algorithms more practical, we propose two other new blind spectrum sensing algorithms based on the maximum likelihood estimate of unknown noise variance. Using random matrix theory, the theoretical analysis on detection probability, false alarm probability and threshold are given. Finally, simulation results show the effectiveness and robustness of the proposed algorithms.