Joint source and channel decoding with symbol re-alignment

Abstract

The maximum a-posteriori probability (MAP) decoding of Huffman variable-length codes (VLCs) has been widely used in the joint source and channel decoding. This method exploits both source memory information and channel status information, but does not consider the symbol re-alignment as the conventional Huffman decoding does. In this letter, we show that self-synchronization strings (SSSs) of VLCs can be readily found and used with the MAP decoding. As a result, the performance of symbol error rate based on Levenshtein distance can be improved, with reduced decoding complexity.