Threshold-Based List Decoding for PAC Codes

Abstract

Recently, polarization-adjusted convolutional (PAC) codes proposed by Arikan were shown to approach theoretical bound with sequential decoding and arouse the interest of the channel coding community. Then list decoding was shown to achieve similar performance with distinct advantages over sequential decoding in low-SNR regimes, but with higher complexity in high SNRs. To further reduce the complexity of list decoding, in this paper, we first adapt the list decoding with a threshold for use to decode PAC codes. Specifically, we introduce a threshold-based splitting rule, which avoids unnecessary splitting of branches with reliability higher than a preset threshold. Simulation results show that, the threshold-based list decoding achieves a balance between the performance and complexity, by choosing threshold value carefully, the threshold-based list decoding exhibits lower complexity than list decoding but with similar error-correction performance.