A blind equalization algorithm with variable step-size procedure

Abstract

A blind equalization algorithm with a variable step-size procedure leading to faster convergence speed is considered. An earlier work on an enhanced version of the Decision Directed Modified Constant Modulus Algorithm(DD-MCMA), used with a variable step-size procedure showed a very good convergence speed (compared to the same algorithm with a fixed step-size) on different, types of channels and different kinds of modulation schemes. Different step-size update schemes were considered, a prediction error dependant scheme were adapted to change the step-size according to the auto-correlation of the prediction error. The step-size parameters were optimized using particle swarm optimization method which yield solutions that gave good results in terms of the step-size affect on the speed of convergence.