Iterative Decoding of Non-binary Turbo Codes Using Symbol Based SOVA Algorithm

Bandwidth efficient turbo coding schemes attract increasing research interest these years, most of these schemes are constructed by non-binary convolutional codes and need to be decoded symbol by symbol instead of bit by bit. In this paper, a modified symbol based SOVA decoding algorithm for non-binary turbo codes is proposed. Using our symbol based SOVA algorithm, soft output information from the prior iteration can be accepted as a priori soft input information for current iteration. The experimental results show that our algorithm can achieve close performance to MaxLogMAP algorithm for large interleaver size with lower complexity, and the performance loss of the SOVA compared to the optimal MAP algorithm is only about 0.3 dB at a BER of 10^-4