GLRT based cooperative spectrum sensing with soft combination in heterogeneous networks

As it is not suffered by the well-known noise uncertainty problem of energy detection (ED), generalized likelihood ratio test (GLRT) based spectrum sensing schemes for a single secondary user node have recently been extensively addressed. In this paper, we extend the existing research to an optimal GLRT based cooperative spectrum sensing scheme in a heterogeneous network scenario, where each node can be equipped with different number of antennas and have different signal processing capacity, such as sampling rate. By considering the discrepancy of different secondary user nodes´ sensing reliability due to different hardware configuration, we demonstrate that the optimal cooperative spectrum sensing scheme based on GLRT is a linear combination of each node´s local test statistic. Moreover, the local test statistic is the logarithm of the arithmetic-to-geometric mean ratio of the eigenvalues associated with the sample covariance matrix. Furthermore, the optimal combining coefficient is a simple function of the numbers of the antennas and the received signal samples at each secondary user node, which can be intrinsically provided by each secondary user node without any prior information. Numerical results show that the proposed GLRT based cooperative spectrum sensing scheme performs better than the practical ED based scheme, i.e., equal gain combination based ED, especially under the case where the sensing reliability discrepancy of each secondary user node is significant.