Effect of spatial correlation on blind source separation in MIMO systems

Abstract

In this work, we investigate the effect of spatial correlation in MIMO channels on the performance of multistage CMA+AMA blind equalization and signal separation method. A modified Constant Modulus Algorithm (CMA), which minimizes a linear combination of the CMA and Alphabet Matched Algorithm (AMA) is used as the blind equalizer for blocks of QAM signals. It has been shown that a multistage CMA+AMA equalizer with a blind channel estimator among stages successfully equalizes and separates multiple data signals when the channel model used is that of identically and independently distributed (iid) Rayleigh fading. In this paper a different channel model from work of H.Bolcskei is adopted to investigate the effectiveness of multistage CMA+AMA equalizer and the channel estimator on the channels which are spatially correlated. The performance of the multistage equalization and estimation scheme is compared with uncorrelated iid channels versus transmit and/or receive correlated channels and presented in terms of symbol error rate (SER) and mean square error MSE) for the signals and for the channel estimate.