Subspace-based blind channel identification of SISO-FIR systems with improper random inputs

In this paper, we consider the problem of blindly estimating the impulse response of a nonminintum phase single-input single-output channel, by resorting only to the second-order statistics (SOS) of the channel output. On the basis of a general treatment of the problem, we show that a SOS-based subspace procedure can be applied to the problem at hand if the transmitted signal is an improper process, exhibiting some additional properties. After characterizing and discussing these properties, we show that they are exhibited by many improper digital modulation schemes of practical interest. Moreover, based on our unifying framework, we devise subspace-based algorithms for blind channel identification, by addressing in particular the related identifiability issues. Finally, the theoretical expression of the mean-squared error of the channel estimate is derived, and numerical simulations are carried out for its validation and for comparing the performance of the proposed algorithm with that of a conventional subspace-based method. (C) 2004 Elsevier B.V. All rights reserved.