Blind identification of MIMO-FIR systems: A generalized linear prediction approach

We address the blind identification of multiple input multiple output (MIMO ) polynomial transfer functions with the number of inputs strictly less tha n the number of outputs. According to previous results, the knowledge of th e second-order statistics of the output series allows to reduce the convolu tive mixture to an instantaneous mixture of the inputs under mild assumptio ns. Our approach consists of second-order deconvolution via the so-called l inear prediction approach with the possibility of further instantaneous mix ture separation. Contrary to the standard linear prediction, we propose a s tatistically optimal least-squares identification procedure. The latter rev eals a simple sub-optimal approach characterized by high estimation accurac y as well as robustness to the model order choice. The statistical inferenc e about the asymptotic performance of these methods is validated by compute r simulations. (C) 1999 Elsevier Science B.V. All rights reserved.