SBL-Based GLRT for Spectrum Sensing in OFDMA-Based Cognitive Radio Networks

摘要

In this letter, a novel sparse Bayesian learning (SBL)-based spectrum sensing technique is proposed for an orthogonal frequency-division multiple access-based cognitive radio network. The SBL framework is employed to acquire the temporally sparse channel estimates, which are subsequently incorporated in the generalized likelihood ratio test (GLRT) to obtain the novel decision statistics. In addition, two other GLRT-based schemes are developed, considering the known and unknown sparsity information, respectively, of the wireless multipath channel at the secondary user. In this context, closed form expressions are derived to characterize the theoretical probability of false-alarm and the probability of detection performance. Finally, some interesting inferences are presented with respect to the derived asymptotic GLRT bounds.