IEEE Communications Letters
Enhanced Guessing Random Additive Noise Decoding via Bit Marking for Product Codes
Yi Lei, Minjie Xu, Alex Alvarado, Bin Chen
Abstract
Iterative GRAND (IGRAND) is attractive for its universality and low complexity, but its exhaustive search over all bits can lead to redundant attempts and potential miscorrections. To address these issues, this letter proposes an enhanced IGRAND decoder for product codes based on bit marking, termed BM-IGRAND. Specifically, the decoder exploits soft information to classify received bits into three reliability levels via two thresholds, thereby eliminating highly reliable bits from the error search space and restricting the search to unreliable bits (UBs) and highly unreliable bits (HUBs). The test error pattern (TEP) ordering is further optimized by considering the error probabilities of HUBs/UBs and the size of the candidate combinations. Simulation results show that at a BER of$10^{-5}$, BM-IGRAND achieves up to 1 dB gain over hard-decision IGRAND while reducing the number of TEP attempts by over 50%.