A Continuous Time Balanced Parametrization Approach to Multichannel Blind Deconvolution

Abstract

In this paper, we will apply a balanced parametrization approach to multichannel blind deconvolution problem with the mixer being modelled as a continuous time linear time invariant system. Such an approach has the advantages of (a) being a computationally robust method, compared with the controller canonical form representation or observer canonical form representation of the linear time invariant continuous time system, and (b) allowing the determination of the number of states required in the demixer. Our approach is validated through a computer simulation example using speech signals.