Adaptive error constrained LMS algorithms and its blind equalization method

In this paper, the adaptive error constrained least mean square (AECLMS) algorithm is proposed through the extension and generalization of the noise constrained LMS (NCLMS) algorithm and its performance analysis is presented. By using a constrained optimization technique, the assumption that the noise variance is known is eliminated. Therefore, the proposed constrained optimization method can be easily applied to blind equalization methods. The proposed constrained method is also applied to the constant modulus criterion. The proposed method can accelerates the convergence speed of the conventional steepest descent-type training procedure by several times.