A Viterbi based decoding algorithm for multiple description variable length codes

In the presence of bit errors, variable length (VL) codes often suffer from a loss of synchronization, which leads to spans of symbol errors. It is of our interest to investigate whether the redundancy introduced by multiple description (MD) coding is useful for improving performance. We consider a sequence of i.i.d. source symbols of known length, first quantized, then coded using MD VL codes and transmitted over a binary symmetric channel (BSC). We propose a maximum a posteriori probability (MAP) decoder, in which the optimal sequence with the right number of symbols and bits is found using the Viterbi algorithm. We compare the MD VL code performance against a conventional single description (SD) VL entropy code, and against a single description (SD) parity code