A Review on the Evolution of Eigenvalue Based Spectrum Sensing Algorithms for Cognitive Radio

Abstract

Spectrum scarcity has been encountered as a leading problem when launching new wireless services. To overcome this problem, cognitive radio is an optimistic solution. Spectrum sensing is a prominent task of cognitive radio. Over the past decade, numerous spectrum sensing algorithms have been proposed. In this paper, we present a comprehensive survey of
evolutionary achievements of eigenvalue based spectrum sensing algorithms. The correlation between signal samples due to oversampling, multipath or multiple receivers gets reflected on the eigenvalues of the covariance matrix. It has been observed that different combinations of
eigenvalues are used as test statistics and the distribution of eigenvalues and derivation of probability of detection is based on RMT (Random Matrix Theory). The main advantage offered by these algorithms is their robustness to noise uncertainty which severely affect other methods. Furthermore, they do not require accurate synchronization. These detections can be used for different signal detection applications without any prior information of signal or noise. To evaluate the performance of eigenvalue based spectrum sensing techniques under fading channels, we have simulated maximum to minimum eigenvalue based Detection
(MME) and maximum eigenvalue based detection (MED) estimation for Rician fading channel. Simulation results shows that MME is much better than MED.