Cooperative spectrum sensing and allocation with LDA for cognitive radio networks

Spectrum sensing is a key approach for the realization of cognitive radio and needs to efficiently detect the presence of the licensed users to eliminate interference and find the spectrum holes for transmission. Even though the available holes have been successfully detected, how to allocate remains a challenge. In this paper, we model a scheme combining spectrum sensing and spectrum allocation to overcome these challenges. Specifically, we frame a training period and introduce Linear Discriminant Analysis (LDA) aiming at improving the accuracy of the spectrum sensing in the data fusion process. Also a channel allocation scheme is carefully designed to guarantee the fairness among secondary users (SUs) and improve the sensing performance. Besides, with this information, fusion center can balance sensing performance and spectrum utilization so as to maximize global utility. Simulations results illustrate a significant improvement in detect accuracy and the utility of fusion center and secondary users ensured by the proposed scheme.