Blind identification and equalization of linear multi-input, multi-output FIR channels

Blind identification of linear MIMO channels is addressed using high-order spectra of the non-Gaussian observations. The proposed eigen-decomposition schemes are applicable to a class of FIR channels with (possibly) unknown order and identify the channel frequency response uniquely. Alternate correlation and kurtosis based adaptive schemes that are applicable to a different class of channels are also proposed. These algorithms are based on multi-channel decision feedback equalizers. Simulation results are provided to demonstrate the viability of the proposed algorithms with various types of channels.