A Reliability-based Stopping Criterion for Turbo Product Codes

摘要

This paper introduces a novel reliability-based stopping criterion for iterative decoding of turbo product codes. The proposed stopping criterion is working by first exploiting the least confident Log-Likelihood Ratio (LLR) input information to soft decision decoder, according to an optimized reliability threshold, and if their count is less than a certain limit, the soft decision iterative process stops and the next half iteration will be carried out using only algebraic decoding. Simulation results that compares the proposed stopping criterion with other stopping criteria, devised so far, show that our proposed stopping criterion, with very limited loss in coding gain, can achieve a superior reduction in the average number of iterations.