Blind identification of multi-channel systems with single input and unknown orders

In commonly used SOS-based methods for blind identification of single input multiple output (SIMO) systems, such as subspace (SS) and least squares (LS), the highest order of unknown FIR channels is assumed to be known or is overestimated. In this paper we propose a new method based on modified Levinson algorithm (MLA) that makes no such assumptions. This method determines which FIR channel is minimum-phase, and estimates every channel impulse response with correct order when at least one of the unknown FIR channels is invertible. We first develop a MLA-based method for estimating a LTI SISO stable IIR system of correct orders when the input is an arbitrary stationary signal. We then use the results of such method and propose a solution to the SIMO blind system identification problem when orders of FIR channels are unknown. Performance of the proposed method is compared with those of the SS and power of R (POR) methods using different signal to noise ratio.