Pipelined implementation of the adaptive canceller-equalizer

Due to the feedback path, pipelining of the decision feedback equalizer (DFE) is not a simple matter. Delayed update algorithms and relaxed look-ahead methods have been successfully applied to pipelining implementations. Both approaches are based on modifying the update algorithm, resulting in a structure that is very sensitive to statistical characteristics of the channel. Also, because of pipeline delays in the feedback path, subtraction of exact ISI components is not possible. In this paper, we propose a method to achieve high performance pipelined decision-aided equalization. With the adaptive canceller-equalizer (ACE) structure, the feed-forward filter is decomposed into two small sub filters. Using a tentative decision device with one of these short filters will require a smaller number of delays for re-timing and will allow the use of the output of a tentative decision device as an input to the feedback filter. This improves the performance of the pipelined structure while maintaining the throughtput.