Abstract: In order to enhance the performance of the min-sum decoding algorithm for polar codes, an improved min-sum decoding algorithm was proposed. The improved method corrects the node update formulas of the min-sum decoding algorithm which replaces the function lncosh(x) in the belief propagation (BP) decoding algorithm with the piecewise linear approximation functions. Compared with the min-sum decoding algorithm, the modified decoding algorithm has better performance in case of increasing a little complexity. Compared with the BP decoding algorithm, the algorithm greatly reduces the computational complexity in case of almost no performance loss, and simplifies hardware implementation. The modified algorithm is based on the min-sum and the BP algorithms, and it is a compromise between the complexity and performance. Simulations show that the performance of the improved min-sum decoding algorithm is almost the same as the BP decoding algorithm, and better than the min-sum decoding algorithm.

Key words: Polar code, min-sum decoding algorithm, belief propagation decoding algorithm, piecewise linear function