Adaptive Viterbi decoding of reed-muller codes

Abstract

For years Viterbi algorithm is considered one of the most efficient decoding algorithm for decoding trellis based Reed-Muller codes. As a result of the nature of the trellis structure of block codes, the number of eliminated branches is far less than that eliminated in convolutional codes. This paper adopts the use of an adaptive Viterbi algorithm to Reed-Muller codes such that branch elimination can take place not only at merging states but also at any trellis level. The used adaptive algorithm not only preserves the error performance, but also can achieve significant complexity reduction for both hard decision decoding and soft decision decoding.