Recovering the Parameters of an LDPC Code From Noisy Intercepted Sequences

• Published in: IEEE signal processing letters Vol. 29; pp. 617 - 621
• Main Authors: Longqing, Li, Fangqi, Shen, Jing, Zhou
• Format: Journal Article
• Language: English
• Published: New York IEEE 2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Binary system; Blind identification; Buildings; Codes; Collision avoidance; Computational complexity; Detectors; dual codewords; Error correcting codes; LDPC codes; Low density parity check codes; Noise; Noise level; Noise measurement; Noise tolerance; Parameter identification; Parity check codes; Signal processing algorithms
• ISSN: 1070-9908; 1558-2361
• DOI: 10.1109/LSP.2022.3146827

Blind identification of channel codes is a key technology in non-cooperative communication as well as intelligent communication. In this letter, we are interested in recovering the parameters of low-density parity-check (LDPC) codes without any a priori information. We propose to find dual codewords from noisy intercepted sequences and prove that the problem is NP-complete. Afterward, we propose an improved collision-based algorithm that can avoid most of the invalid searches and thus allowing a significant reduction in complexity. The algorithm produces almost no errors when the noise is small. While in the case of non-negligible noise, we propose to use soft information to detect the outputs of the algorithm, and errors in which can therefore be extensively eliminated. After a sufficient number of dual codewords have been found, one can easily recover the code lengths and code rates of LDPC codes, as well as the sizes of sub-matrices for Quasi-Cyclic LDPC codes. Experimental results show that the proposed method has an improved noise tolerance and can be more efficient than existing methods.