Optimal cooperative sensing for sensors equipped with multiple antennas

This paper considers multi-sensor multi-antenna spectrum sensing. First, it is assumed that all users are able to send their raw data to the fusion center. In this case the global optimial solution is the likelihood ratio test (LRT) using all the raw data. A simple closed-form expression for the LRT is found. Although LRT is optimal, it is hardly useful in practice due to its reliance on the knowledge of primary user's channels and noise powers of all users. Thus a method using the estimated channels and noise powers is proposed, which is called generalized LRT (GLRT). Secondly, the optimal fusion scheme (OFS) is found if each user computes its test statistic based on an eigenvalue based detection and sends the test statistic to the fusion center. Both GLRT and OFS need the SNR information of all users. To make the detections more practical, two totally blind detections, namely, approximated OFS and approximated GLRT, are proposed. Simulations are provided to support the results.