Blind identification in non-Gaussian impulsive environments

A new algorithm is proposed for blind channel identification in the impulsive signal environments, where the signals are modeled as symmetric /spl alpha/-stable processes. The alpha-spectrum, a new spectral representation based on fractional lower-order moments, is developed. Conditions for blind identifiability of any FIR channel (non-minimum phase, unknown order) are established using the properties of the alpha-spectrum.