Robust subsystems for iterative multichannel blind system identification and equalization

This paper presents an iterative approach to multi-channel blind system identification. The concept includes two subsystems for channel identification and equalization which are specifically tailored for robust mutual interaction. This robustness is a natural prerequisite for the convergence of iterative systems when no suitable a priori information about the channels or the input signal is available. For the channel identification subsystem, we introduce a supervised, variable stepsize LMS-type adaptive algorithm which is able to identify the channels from an estimated input signal. We then show that matched filter arrays can be utilized as the equalization subsystem to estimate the input signal on the basis of the identified channels. Eventually, we demonstrate that the iterative coupling of both subsystems converges to a solution, the quality of which can be predicted from our comprehensive analysis of the independent subsystems.