A robust algorithm for blind space-time equalization

The efficient separation of signals is a frequent problem in multiuser communication systems. Among many algorithms to blind deconvolution of a multiple-input multiple-output (MIMO) systems, the one that utilizes higher-order cumulants has advantages in regards of convergence rate. Inspired on this algorithm and on a stochastic gradient approach, we propose an algorithm with capacity of recovering simultaneously all sources, denoted as MU-SWA (Multiuser Shalvi-Weinstein Algorithm). Based on the steady-state analysis, recently presented by Luo and Chambers for the Multiuser Constant Modulus Algorithm, we derive the expression for the mean-square error of MU-SWA. Simulation results show that MU-SWA presents a more robust behavior with respect to convergence rate and tracking capability when compared to others known algorithms for blind multiuser equalization.