Blind separation of the joint algorithm based on the cyclostationarity optimization

Abstract

For blind source separation algorithm separation accuracy is not high question, this paper proposes a joint algorithm that combining joint approximative diagonalization of eigenmatrix algorithm (JADE) with second-order blind identification (SOBI) algorithm together. At the same time, this paper applies cyclic cumulant to the joint algotithm which takes advantage of cyclic cumulant's suppress noise characteristics. Then this paper proposes a joint algorithm basing on cyclostationarity optimization. The simulation results show that the algorithm can realize effective extraction of the desired signal, higher separation performance, and higher convergence speed.