Avoiding divergence in the constant modulus algorithm for blind equalization of MIMO systems

The multiuser constant modulus algorithm (MU-CMA) can be employed for blind equalization of multiple-input multiple-output (MIMO) communication systems. Due to the multi-modality of the constant modulus cost function, some adverse situations can cause inconsistency in the nonlinear estimates, which in turn can lead the algorithm to diverge. In order to avoid divergence, we propose a dual-mode multiuser algorithm, which works as a normalized multimodulus version of MU-CMA in the first mode, and rejects nonlinear estimates in the second mode. We present a deterministic stability analysis of the proposed algorithm and confirm its good performance by means of numerical simulations.