Precoded Polar Product Decoder Based on Soft-Output SCL Decoding and Maximization of Generalized Mutual Information

Iterative decoding of product codes with precoded polar component codes is optimized by combining two approaches. First, bitwise soft messages are generated by considering all valid decoding paths of a successive cancellation list (SCL) decoder. Second, the soft information during message passing is scaled with offline-computed coefficients that maximize the generalized mutual information of the channel input and the outgoing message in each half iteration. The method significantly improves the error-correcting performance compared to heuristic scaling and soft information generation based solely on the candidate list of the decoder. An extrinsic SCL decoder is used with a Monte Carlo density evolution analysis to derive iterative decoding thresholds that accurately predict the decoder performance.