Contrast with reference a new approach to the blind equalization of MIMO channels

Abstract

Contrast-based separation of sources have a number of advantages. Among others, they are optimal in presence of noise. In this work, we propose a new contrast by considering a so-called reference signal that allows not only to obtain the optimal numerical algorithm but also yields better performance in terms of variance of the estimated mixing filter. This contrast needs source kurtosis to have the same sign. Thus, it’s appropriate to multichannel blind deconvolution.