Blind equalization based on second and fourth order statistics

A novel blind and adaptive equalization scheme is introduced for nonminimum phase communication channels. A classical least-mean-squares type predictor is used for the identification and equalization of the amplitude information of the channel. Independently, phase correction is carried out with a method based on noncausal autoregressive moving average modelling of the fourth-order cumulants of the received signal sequence. It is shown by means of computer simulation examples that the proposed equalizer has very good convergence speed and low start-up symbol error rate.<<ETX>>