Spectrum sensing with Gaussian approximated eigenvalue ratio based detection

Eigenvalue based detection is an ideal solution to spectrum sensing problem in cognitive radio networks. However, results on its sensing performance are very limited. In this paper, we analytically investigate the sensing performance of the eigenvalue ratio based detection. Firstly, a simple and accurate Gaussian approximation to the extreme eigenvalue distribution is proposed. Secondly, a closed-form expression for the optimal decision threshold is derived. The proposed analytical framework yields a useful approximation in realistic sensing scenarios.