A novel interpolation method for blind system identification

Abstract

The problem of single channel blind identification/deconvolution is considered, where a degraded observed signal is modelled as the convolution of a non-stationary source signal with a stationary IIR distortion filter. The specific contribution is concerned with the development of the conditions for the resolution of the problem by means of interpolation as opposed to the usual approach through resampling. Recovery of the source input is achieved by introducing a novel interpolation scheme, which is applied to the observed distorted output to create the necessary conditions for blind identification of the filter parameters.