Blind deconvolution and separation signal processing via inverse model approach

The blind source deconvolution/separation problem is considered based on inverse deconvolution model application. A new approach to inverse model design based on unknown-input observer theory is considered with reference to blind separation problem. The method of reduced-order inverse model design in proposed and the conditions of inverse model parametric design problem solvability are found as a special type of observability. For special case of solvability conditions violation a regularized inverse model is proposed which ensures the possibility of arbitrary pole-placement of inverse model. The method of wave source signals identification and prediction is also proposed.