Fast Convergent LDPC-like BP List Decoding Algorithm for Polar Code Based on Partially Permutation Sparse Graph

• Zhang Chenfei,
• Guo Rui,

Vol. 19, No. 2, February 28, 2025
10.3837/tiis.2025.02.016, Download Paper (Free):

• Polar codes
• Belief Propagation
• LDPC-like BP decoding
• sparse graph
• Partially permute

Abstract

Although the LDPC-like BP decoding algorithm based on sparse graph has lower complexity, its performance is inferior to basic BP decoding’s. To improve the performance, LDPC-like BP list decoding algorithm based on permuted sparse graphs (SG-BPL) has been proposed by utilizing multiple parallel LDPC-like BP decoders. However, most general permuted schemes are based on "full permutation" of interstage of factor graph before pruning, which leads to slower iterative convergence. In this article, a kind of fast convergent SG-BPL decoding algorithm is proposed. This algorithm introduces the partially permuted scheme into sparse graph for the first time, which accelerates the iteration speed on single sparse graph by permuting subgraphs and preserving partial information, allowing decoder can traverse more sparse graphs within the specified iteration limit and improve performance. Simulation results show that compared with LDPC-like BP and other general SG-BPL based on fully permuted scheme, the proposed fast convergent SG-BPL decoding algorithm improves the performance of decoder by traversing more sparse graphs; compared with basic BPL, fast convergent SG-BPL has significantly reduced complexity and delay.

Statistics

Cite this article