Intermountain Engineering, Technology and Computing (IETC)
A Low-Complexity Noise-Aided Selective-Symbol Flip Decoder for Non-Binary LDPC Codes over QAM Channel
Zinnia Muntaha Mowri, Chris Winstead
Abstract
Non-binary low-density parity-check (NB-LDPC) codes offer strong error-correction performance for short to moderate block lengths yet incur high decoding complexity. Symbol-flipping decoding (SFD) algorithms offer reduced implementation complexity compared to soft-decision decoding schemes, such as belief-propagation (BP) and extended min-sum (EMS). However, this reduction comes at the expense of degraded error-rate performance. Motivated by the success of low-complexity, noise-perturbed bit-flipping decoders in the binary domain, our work investigates whether similar benefits can be achieved for NB-LDPC codes operating over QAM channels. We propose a low-complexity noise-aided selective-symbol flipping decoder that incorporates controlled stochastic perturbations into a symbol-level reliability metric combining Euclidean-distance channel likelihood and non-binary syndrome reinforcement. The proposed method preserves the low computational complexity and implementation-friendly structure of conventional SFD algorithms, while relying solely on simple scalar metric computations and syndrome evaluation. Simulation results provide insight into the role and limitations of noise perturbation in non-binary symbol-flipping decoders and highlight a practical, low-complexity noise-aware decoding approach for higher-order modulated systems.