Blind classification and parameters estimation of forward error correction codes based on data cleaning

Abstract

The classification and parameters estimation of forward error correction (FEC) codes are significant in non-cooperative communication. Previous solutions suffer from poor noise resistance or only applicable to closed-set recognition. To address these issues, this paper proposes a FEC classification and parameters estimation algorithm based on data cleaning. It utilizes the reliability criterion of soft information as cleaning rules to reduce erroneous codewords, thereby improving the accuracy of rank calculation. A novel classification criterion is proposed, which utilizes all rank information and further improves classification accuracy. Simulation results show the effectiveness of both the data cleaning mechanism and novel classification criteria in enhancing recognition accuracy. Compared with existing algorithms, the proposed algorithm has a higher accuracy and an improvement of approximately 1dB in noise resistance performance.