移动通信2024，Vol.48Issue(5)：83-88,6.DOI:10.3969/j.issn.1006-1010.20240322-0001
相关噪声下基于深度学习的LDPC码联合降噪译码算法设计
Joint Denoising and Decoding of LDPC Codes Under Correlated Noise Based on Deep Learning
摘要
Abstract
To enhance the decoding performance of wireless communication systems utilizing low-density parity-check codes and high-order modulation in correlated noise environments,we propose a deep learning-based joint denoising and decoding algorithm.The denoiser incorporates a residual shrinkage building unit,and the decoder employs a neural network-based min-sum decoding algorithm using recurrent neural networks(RNNs).In our joint denoising and decoding(JDD)approach,a complex residual shrinkage convolutional neural network(CRSCNN)capitalizes on the superiority of complex neural networks over real-valued networks for processing complex signals.The received complex signals are directly fed into the CRSCNN,utilizing a novel multi-task learning strategy that jointly optimizes denoising and decoding loss functions to improve decoding performance.Simulation results demonstrate that the CRSCNN-based JDD algorithm achieves superior decoding performance compared to the neural network-based min-sum decoding algorithm using RNNs.关键词
低密度奇偶校验码/高阶调制/相关噪声/深度学习/残余收缩模块Key words
Low-density parity-check codes/high-order modulation/correlated noise/deep learning/residual shrinkage building unit分类
信息技术与安全科学引用本文复制引用
杨恩鑫,袁磊,郭毅,岳新东..相关噪声下基于深度学习的LDPC码联合降噪译码算法设计[J].移动通信,2024,48(5):83-88,6.基金项目
甘肃省科技计划资助项目"智能反射面辅助的超可靠低延时通信研究"(22JR5RA490) （22JR5RA490）
