An efficient implementation of sequential detector in spectrum sensing under correlated observations

Improvement on throughput of cognitive radio networks is possible by spectrum sensing when the required sample number (sensing time) is relatively small. This is due to the fact that when the sample number for sensing is smaller, time for a cognitive radio user to transmit is larger. This paper presents a scheme of sequential detector which has average sample number lower than its counterpart, a quadratic detector with a fixed sample number. A sequential detector as per se has a high computational complexity under correlated observations due to the need to compute the inverse and determinant of the signal covariance matrix for every sample acquired before finally deciding. The paper studies how to reduce the computational complexity using circulant matrix approximations of the signal covariance matrix, so that the calculations of the inverse and the determinant are replaced by simpler methods. Reductions on computational complexity are quantified and the performances are evaluated by simulations.