A fast-converging blind predictive DFE

Blind equalization, which refers to the technique by which an unknown dispersive channel is equalized without the aid of a training sequence, has generated a lot of interest in previous years. However, blind equalizers are notorious for the slow convergence of their filter coefficients to the optimal values, which has prompted a flurry of activity into the development of numerous algorithms and structures to speed up their convergence. This paper is motivated by the approach presented by Labat, Macchi and Laot (see IEEE Transactions on Communications, p.921-30, vol.46, no.7, 1998) and presents a scheme to hasten convergence of the coefficients of the filters of a predictive decision feedback equalizer (PDFE). The proposed scheme is seen to ameliorate many of the drawbacks of the scheme proposed by Labat et al. Numerical results are presented that demonstrate the superiority of the proposed scheme versus the scheme suggested by Tong and Liu (see Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, p.3901-3904, no.5, 1997).