Dynamic-Confined Iterative Guessing Codeword Decoding for Product Codes

Abstract

In this letter, we propose a low-complexity universal decoder for product codes based on guessing codeword decoding (GCD). Recognizing that errors typically concentrate at the intersections of rows and columns where the component decoder fails, we propose to generate test error patterns (TEPs) of component GCD specifically at these intersections. We also introduce new stopping criteria for the component GCD. The resulting decoder, named dynamic-confined iterative GCD (DC-IGCD), dynamically refines its focus to reduce unnecessary computations. To further mitigate the impact of miscorrections, we incorporate enhanced anchor decoding (EAD) into the framework. Numerical evaluations demonstrate that the proposed DC-IGCD with EAD based on the new stopping criteria achieves a significant reduction in decoding complexity compared to IGCD with EAD and DC-IGCD with EAD with the original stopping criteria while maintaining nearly identical error-correction performance.