Performance analysis of multi-antenna GLRT-based spectrum sensing for cognitive radio

This paper addresses the generalized likelihood ratio test (GLRT) eigenvalue based detector with an arbitrary number of receive antennas. We investigate the optimum decision threshold, the minimum sensing time and the achievable sensing throughput trade-off of the secondary network. First, we derive the generalized asymptotic distributions of the test statistic. Second, we investigate the optimal decision threshold that can minimize the total error rate with constraints. Third, we provide the algorithm to find out the shortest sensing time that enables the minimum total error rate to achieve the target value. Finally, we formulate the achievable sensing throughput trade-off for the secondary network and investigate the optimal sensing time which can maximize the achievable throughput for the GLRT detector with multiple antennas under the absence and presence of the noise uncertainty. The accuracy of the derived theoretical models is supported by simulations. The results have shown that the optimized decision threshold and the minimum sensing time can satisfy the target value of the minimum total error rate speedily while both the interests of primary and secondary users are guaranteed simultaneously. In addition, the chosen optimal sensing time maximizes the throughput with and without noise uncertainty.