On Maximum-Likelihood Decoding of Time-Varying Trellis Codes

Decoding complexity of convolutional and trellis codes by Viterbi decoder can be reduced by applying suggested merging algorithm to the Forney code trellis. The algorithm can be applied for every trellis section separately, which is convenient for time-varying codes, and it outputs the minimal trellis of the section. In case of convolutional codes, the same minimal trellis of every section can be obtained from the syndrome trellis of proposed split code.