Reduced-search trellis decoding of coded modulations over ISIchannels

摘要

Optimal Viterbi decoding of trellis codes transmitted overchannels exhibiting long intersymbol interference (ISI) can becomputationally infeasible. Suboptimal trellis search schemes promisemuch better general performance than decision feedback equalization(DFE) or linear equalization, M. Eyuboglu and S. Qureshi's (1989)reduced-state sequence estimation (RSSE) is the most general. It isshown that two reduced-search breadth-first trellis decoders, namely theM-algorithm and the newer T-algorithm, can provide performanceequivalent to RSSE, but at a substantially reduced computational load.This advantage is especially pronounced for (minimum-phase) ISIresponses with precursors. With the M or T-algorithms, a more powerfultrellis code can be used to achieve a better error rate. These decodersoperate on the original trellis without the need for the partitioningrequired by RSSE