Reduced-complexity offset min-sum check node unit for layered 5G LDPC decoder

• Published in: Telfor Journal Vol. 13; no. 1; pp. 7 - 12
• Main Authors: Petrović, Vladimir, El, Mezeni
• Format: Journal Article
• Language: English
• Published: Telecommunications Society, Academic Mind 2021
• Subjects: 5g new radio; layered decoding; low density parity check (ldpc) codes; min-sum (ms) algorithm
• ISSN: 1821-3251; 2334-9905; 2334-9905
• DOI: 10.5937/telfor2101007P

This paper presents a novel approach for the reduced-complexity Min-Sum (MS) decoding of low density parity check (LDPC) codes in the partially parallel layered decoder architecture, which contains a large number of serial check node processors. Reduced complexity is obtained by using the variant of the single-minimum Offset Min-Sum (smOMS) algorithm that approximates a second minimum with the addition of the variable weight parameter to the minimum value. Although the reduced-complexity MS algorithms primarily reduce hardware resources in fully parallel implementations, the results showed that a considerable reduction can be obtained if serial check node processors are used. The paper also proposes a better subminimum estimation for irregular codes from 5G new radio (5G NR). The method uses smaller subminimum estimation weights in check nodes with a higher degree and higher weights in check nodes with a smaller degree, which leads to the significant improvement in the SNR performance. Additionally, it is shown that SNR performance can be further improved by applying offset before minimum calculation, which differs from conventional Min-Sum approaches.