Natural Gradient Blind Equalization Algorithm Based on Blind Signal Separation

Where the approximate processing leads error problems. In this paper, a new natural gradient blind equalization algorithm based on BSS is proposed. The underlying prior knowledge of source signals constellation is made full use of and a multi-stage clustering method is adopted to solve multi-peak problem in the new algorithm. Simulation results show that the new algorithm has higher accuracy than those only based on BBS and has higher convergence speed and less inter-symbol interference.