Adaptive Paraunitary Filter Methods For The Blind Equalization of Frequency Selective Multiple-Input Multiple-Output Channels

In this paper we introduce fast paraunitary matrix obtaining approaches to be used for the blind deconvolution of all blindly equalizable, frequency selective MIMO (multiple-input multiple-output) systems. The main purpose for offering these approaches is to extend various instantaneous blind source separation (BSS) approaches to handle the convolutive BSS case. It was aimed that the offered algorithms have fast convergence properties and low computational complexities so that they can be realized real-time in the communication systems having high sampling rates. These adaptive algorithms which were developed according to the gradient search on specific sets, were compared to the afore introduced alternating projections algorithm in terms of speed and computational complexity