Adaptive minimum variance methods for direct blind multichannel equalization

In this paper, stochastic gradient and RLS-based methods are presented for designing direct adaptive equalizers. Self-recovering solutions are obtaine d by minimizing the equalizer's output Variance subject to appropriate cons traints. The constraints are chosen to guarantee no desired signal cancella tion and are also jointly and adaptively optimized to improve performance. The resulting algorithm may be interpreted as an optimal version of earlier linear prediction-based approaches. It is shown that the algorithm enjoys global convergence. Moreover, the constraint parameters converge to the cha nnel parameters at high SNR. Comparisons with other blind and trained metho ds are presented. (C) 1999 Elsevier Science B.V. All rights reserved.