On-line multichannel blind equalization algorithm with nonstationary signals

Multichannel blind equalization is an important task for numerous applications such as speech separation, dereverberation, communication, signal processing and control, etc. In this paper, a cost function is constructed by the knowledge of correlation. Then a novel on-line algorithm is derived with natural gradient search method for multichannel blind source separation of convolved signal mixtures in time domain. Furthermore, equivariance property is possessed by the algorithm such that its asymptotic performance depends only on the correlation of the source signals, and not on the characteristics of the unknown channel. Simulations indicate the ability of the algorithm to perform equalization.