Blind source separation with CMA+AMA equalizer over correlated MIMO channels

The blind estimation of digital communication channels together with the problem of signal separation in correlated MIMO communication links is considered here. In this work, a method of blindly estimating the impulse response of a frequency selective digital channel is proposed and used for MIMO equalization. The most powerful data source is recovered from received signal mixture by using a modified CMA equalization method. The channel impulse response is then estimated based on the autocorrelation of the received signal block and the equalizer weight vector. Monte Carlo simulations demonstrate the effectiveness of the separation method. Equalizer performance is evaluated in terms of the MSE (mean square error) and SER (symbol error rate) of the recovered sequences.