A comparison of optimal decoding of convolutional codes over fast fading channels

It is well known that the Bahl, Cocke, Jelinek and Raviv (BCJR) and Viterbi decoding algorithms for convolutional codes are optimal over additive white Gaussian noise (AWGN) channels. The BCJR algorithm is optimal in the sense that it maximizes the a posteriori probability of each bit and Viterbi algorithm is optimal in that it maximizes the likelihood of each decoded codeword. However, experimental results showed that the difference in their performances over pure AWGN channels with respect to both bit error rate (BER) and word error rate (WER) are negligible. In this paper, the performance of BCJR and Viterbi decoding algorithms over fast changing time-varying wireless channels where channel state information may not be available is investigated. Simulation results show that, with the channel condition changing fast, Viterbi decoding offers better performance than BCJR decoding with respect to both BER and WER. It is also shown that, although the knowledge of channel state information does help BCJR decoding, it still underperforms the Viterbi decoding.