Low-complexity MAP decoding for turbo codes

Two simplified MAP algorithms for iterative decoding of turbo codes are presented. By using a parameter "decoding depth", our algorithms do not require computation of the a posteriori probability of each decoded information bit throughout the entire trellis, thus reducing the computational complexity and decoding delay considerably. One of the algorithms can achieve performance very close to the conventional MAP algorithm; the other performs better than SOVA, while retaining a comparable process cost. Their advantages over the MAP and SOVA algorithms are demonstrated in both theoretical analyses and computer simulation assuming a Rayleigh fading channel.