Multichannel signal modeling and separation

Separation of multiple signals from their superposition recorded at several sensors is addressed. The methods employ polyspectra of the sensor data in order to extract the unknown signals and estimate the exact coupling systems via a linear equation based method. Proposed schemes are also useful for multichannel blind deconvolution even when the input signals are colored with (possibly) overlapping spectra. Simulation results corroborate the applicability of the algorithm.<>