A fast online separation algorithm for convolutive mixture model in WSDM

Wireless statistic division multiplexing (WSDM) is a multiplexing technology that can transmit multiple signals simultaneously in the same frequency band over wireless channels. Aiming at the convolutive mixture model in WSDM, this paper proposes a fast online separation algorithm. This algorithm firstly transforms noisy convolutive mixture into a noiseless instantaneous linear mixture by employing a Time-Domian Blind Deconvolution Method (TD-BDM). Then, the sources are recovered online by a modified variable step-size Equivariant Adaptive Separation via Independence (EASI) algorithm using the difference value of kurtosis to adjust step-size. Several simulation results show that the proposed separation algorithm has the faster convergence rate and better separation performance than the conventional fixed step-size and the classical variable step-size algorithm. The results prove that the proposed separation algorithm can be better applied to the WSDM system.