Fading channel estimation by mutual information minimization for Gaussian stochastic processes

In this paper, we propose a method for estimating the fading channel coefficients of the desired user in a code-division multiple access (CDMA) system. The essential assumption made is that the transmission channel is Gaussian. An independent component analysis approach based on complexity minimization is used for separating the signal of interest. Both the time structure of the fading channel and the independence of different transmission paths are taken into account. Simulations with downlink data, propagated through a Rayleigh fading channel, shows noticeable performance gains compared to blind MMSE channel estimation.