Non-minimum phase AR identification using blind deconvolution methods

Identification of the parameters of non-minimum phase AR processes is formulated in the framework of the Super Exponential blind deconvolution scheme described in [4]. We show that the vector of the AR parameters lies in the range of a suitable (rank one) matrix formed using second and higher order statistics measured from the received data. The resulting algorithm is similar to the blind deconvolution scheme presented in [5], which here is shown to be directly obtained from the optimality criterion underlying the Super Exponential blind deconvolution algorithm.