A numerically robust blind equalization scheme applied to MIMO communication systems.

In this paper, we propose a numerically robust blind equalization scheme to mitigate cochannel and intersymbol interferences and separate signals from a convolutive mixture in multiple-input multiple-output (MIMO) communication systems. The robustness of the proposed scheme is achieved through a particular dual-mode operation applied to a constant-modulus-based space-time algorithm that corrects phase rotations. In the first mode of operation, the consistency in the nonlinear estimates of the transmitted signals is ensured. In the second mode, the Euclidean norms of the equalizer coefficient vectors decrease along the iterations, which allows the algorithm to return to the first mode of operation, where the estimation of the transmitted signals really occurs. In particular, we apply this scheme to a multimodulus version of the multiuser Shalvi-Weinstein algorithm (MU-SWA). By means of a deterministic stability analysis assuming the persistence of excitation condition, we show that the dual-mode MU-SWA does not diverge.