Abstract：This paper presents an improved maximum a posteriori probability (MAP) decoding algorithm for optimal decoding of parallel concatenated convolutional codes (Turbo codes). Compared with the Log-MAP algorithms based on logarithm domain, the proposed algorithm does not introduce logarithm domain, but it can eliminate large amount of exponential and logarithm arithmetic operations required with the standard MAP algorithm in the iterative decoding process. Simulations demonstrate that he proposed algorithm, with optimal error correction performance, can significantly reduce the running time, such that its decoding efficiency is even slightly superior to that of the fastest logarithm domain MAP algorithm(Max-Log-MAP) which sacrifices considerable error correction performance.
2008, Vol. 30 
