Reduced-search trellis decoding of coded modulations over ISI channels

Optimal Viterbi decoding of trellis codes transmitted over channels exhibiting long intersymbol interference (ISI) can be computationally infeasible. Suboptimal trellis search schemes promise much 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 is shown that two reduced-search breadth-first trellis decoders, namely the M-algorithm and the newer T-algorithm, can provide performance equivalent to RSSE, but at a substantially reduced computational load. This advantage is especially pronounced for (minimum-phase) ISI responses with precursors. With the M or T-algorithms, a more powerful trellis code can be used to achieve a better error rate. These decoders operate on the original trellis without the need for the partitioning required by RSSE