학술논문
A UnetNND-BP Architecture for Channel Decoding Under Correlated Noise
Document Type
Periodical
Source
IEEE Communications Letters IEEE Commun. Lett. Communications Letters, IEEE. 28(4):823-826 Apr, 2024
Subject
Language
ISSN
1089-7798
1558-2558
2373-7891
1558-2558
2373-7891
Abstract
In this letter, we introduce a Unet-based neural network denoiser-belief propagation (UnetNND-BP) architecture with two training modes to improve the decoding performance of low-density parity-check (LDPC) codes. In the supervised learning mode, UnetNND-BP achieves better performance than benchmark schemes, while in the self-supervised learning mode, it achieves comparable performance.