CMA-based blind multichannel equalizer for a multivariate linear prediction scheme

In recent years, it has been shown that an oversampled communication or an array receiver improves the capability of the blind equalization task. One of these approaches leads us to a multivariate linear prediction scheme. This article presents an alternative formulation of this approach designing a proper cost function (based on the well known constant modulus algorithm (CMA)) for a feasible implementation. Our main contribution is the proposal of an adaptive scheme to solve the multivariate linear prediction by using a linearly constrained CMA-based algorithm with also an iterative gain adjustment. The robustness to overdetermination and to the noise presence, the low computational requirements and its fast convergence provide a competitive algorithm for wireless communications. Some simulations support the ability of our proposal in a wireless scenario.