An Improved Cognitive Radio System Based on WSDM

This paper proposes an improved cognitive radio system based on Wireless Statistic Division Multiplexing (WSDM) which is the modified version of the new-style cognitive radio system proposed by Li. The maximal user number is limited by the antenna number at receiver in Li's system. In this paper, we relax this limitation in multiplexing subsystem and propose a modified minimum description length (MDL) algorithm based on the fourth-order cumulant by adding the virtual noise which can estimate the source number even when the source number is more than antenna number at receiver. After estimating the source number, mixing matrix is obtained by Second-Order Blind Identification of Underdetermined Mixtures (SOBIUM) and then the sources can be recovered by minimum mean squared error-based (MMSE) beamforming approach. Simulation results show the feasibility of the proposed cognitive radio system and the superiority of the proposed source number estimation algorithm over the existing source number estimation algorithm based on fourth-order cumulant.