A Learning Algorithm with Distortion Free Constraint for Multi-Channel BSS Systems and Performance Analysis Based on Their Structures

Feed-Forward (FF-) and FeedBack (FB-) structures have been proposed for Blind Source Separation (BSS) systems. FF-BSS systems have some degree of freedom in the solution space, and signal distortion is likely to occur in convolutive mixtures. Previously, a condition for complete separation and distortion free has been derived for 2-channel FF-BSS system. This condition has been applied to the learning algorithms as a dis-tortion free constraint in both the time and frequency domains. In this paper, the condition is further extended to multiple channel FF-BSS systems. This condition re-quires a high computational complexity to be applied to the learning process as a constraint. An approxi-mate constraint is proposed in order to relax the high computational load. In comparison with the original constraint, computer simulations have demonstrated that the approximation can obtain similar performances with respect to source separation as well as signal dis-tortion using speech signals. Furthermore, the per-formances can be improved compared to the conven-tional methods. On the other hand, FB-BSS structure is hard to cause signal distortion. However, FB-BSS system requires a condition for the transmission time dierence in the mixing process. FF-BSS systems and FB-BSS system are compared based on the transmis-sion time dierence in the mixing process. Location of the signal sources and the sensors are rather limited in the FB-BSS system