Reduced-search SOVA for block turbo codes

In this paper, we propose an efficient reduced-search SOVA (soft output Viterbi algorithm) for block turbo codes. To reduce complexity, the decoder selects a part of the existing paths using the statistics of the path metrics. In addition, we compensate for the performance degradation incurred from soft output values at the parity part of the trellis that were too optimistically estimated, so that the performance of the reduced-search decoder almost approximates that of a full-search decoder. Our simulation results reveal that the proposed reduced-search method can reduce the complexity by about 10 times with just about 0.1dB performance degradation in coding gain.