Abstract: Two schemes including both linear equalization (LE) and decision feedback equalization (DFE) are introduced based on the adaptive least mean square (LMS) algorithm. Simulations are utilized to compare the equalization performance of the LELMS and the DFELMS in the hybrid mode of training and decision-directed. Moreover, the influence of the length of feedback filter on the DFELMS algorithm is analyzed. Results indicate that in the training period, the LELMS outperforms the DFELMS; in the decision-directed phase, for the channel in good condition, the LELMS has a desirable equalization performance comparatively; in the case of the ill-conditioned channel, the LELMS degrades greatly whereas the DFELMS equalizes the channel better as the length of feedback filter increases.

Key words: adaptive channel equalization, LMS algorithm, linear equalization, decision feedback equalization, ISI