Iterative detection and decoding on nonlinear ISI channels

This paper examines various techniques for nonlinear intersymbol interference (NL-ISI) mitigation, including the RAM-Search Viterbi detector, the RAM-Search BCJR algorithm, and two iterative detection/decoding algorithms. The two RAM-Search algorithms are only presented to lay a foundation and to provide a basis for comparison. The primary contribution in this paper is the comparison of the two iterative detection/decoding algorithms. The first is simply a generalization of turbo equalization to NL-ISI channels. The second considers the NL-ISI channel to be an inner code in a serial concatenation scheme, in which case interleaver gain is possible whenever this inner code is recursive. We show that by the simple insertion of a differential encoder prior to the NL-ISI channel to achieve recursiveness, this second scheme is far superior to the turbo equalization approach, but with equal complexity.