Blind deconvolution of linear MIMO systems driven by arbitrarily distributed sources

This paper addresses the problem of recovering multiple source signals that are transmitted through a linear multiple input multiple output (MIMO) system, without resorting to any prior knowledge. Source signals are assumed to be spatially independent and temporally i.i.d. non-Gaussian sequences. Moreover source signals are allowed to have arbitrary distributions, in contrast to most existing methods where sub-Gaussian sources are considered. We present an unsupervised hybrid network that performs multichannel blind deconvolution for sources with arbitrary distribution. Useful behavior of the proposed method is demonstrated by computer simulations