Modified Enhanced Max-log-maximum a Posteriori Algorithm Using Variable Scaling Factor

Abstract

Using a variable scaling factor (VSF), a modified enhanced max-log-maximum a posteriori (EMLMAP) algorithm is proposed to improve the performance of the conventional EMLMAP algorithm which uses a fixed SF. The VSF, maximising the mutual information exchanged between component decoders in a turbo decoder, is adaptively varied with a priori information of each component decoder. For a large size of interleaver, it is shown that the proposed algorithm improves the bit error rate performance and speed of decoding convergence of the conventional EMLMAP algorithm. The improved speed of the decoding convergence reduces the average number of iterations and results in low-power implementations of a turbo decoder.