Blind estimation of a complex channel impulse response

Extends to complex channels and data an efficient technique for blind channel identification. The method, which is a variant of maximum likelihood estimation, approaches the Cramer-Rao bound for small (residual) distortion and large sample size, and is recommended as a final step of blind deconvolution. After an analysis for general probability density functions of input data, 'square-' and 'round-shaped' PDFs, and PSK constellations used in digital transmission systems are considered.<<ETX>>