Enhanced Bit-Flipping Decoding Using Refined Checksum Weights and Flip Metrics for LDPC Codes

Bit-flipping (BF) algorithms for decoding low-density parity-check (LDPC) codes have attracted considerable attention due to their low complexity and high parallelism. While recent improvements have enabled certain BF decoders to approach the performance of belief-propagation (BP) decoding, many of these enhancements rely on complex mechanisms such as loop detection, decoder switching, or costly online computations. In this paper, we propose enhanced BF decoding algorithms that integrate refined checksum weighting and flipping function designs. The proposed weighting scheme jointly exploits soft-valued channel information and hard-decision consistency to better reflect the reliability of each checksum, while maintaining negligible online complexity. Additionally, we propose a flipping function design that implicitly discourages repeated flipping by incorporating the accumulated flip count into its computation, thus suppressing decoding loops without requiring explicit loop detection or decoder switching. Simulation results demonstrate that the proposed decoding strategies yield notable performance gains over existing BF decoders, with only a minor complexity increase.