基于深度学习的信道译码算法综述  

The Research Overview of Channel Decoding Based on Deep Learning

作　　者：李慧 柳铮 易卫明 LI Hui;LIU Zheng;YI Wei-ming(Troop 75775,Kunming 650000,China;Fourth Research institute of Telecommunications Technology Co.,Ltd.,Xi'an 710061,China)

机构地区：[1]75775部队,昆明650000 [2]电信科学技术第四研究所有限公司,西安710061

出　　处：《无线通信技术》2022年第2期18-22,共5页Wireless Communication Technology

摘　　要：近年来,深度学习方法的发展优化线性码的译码提供了使用人工神经网络拟合的新思路,本文介绍了深度学习的发展历史,并且深入研究了近些年LDPC码和Polar码译码算法,详细分析了算法的优缺点,最后指出了译码算法未来研究的方向。Recently, The development of deep learning methods optimizing the decoding of linear codes provides a new idea of using artificial neural network fitting. This paper introduces the development history of deep learning, and deeply studies the decoding algorithms of LDPC codes and Polar codes in recent years, and analyzes them in detail.The advantages and disadvantages of the algorithm are discussed.the future research direction of the decoding algorithm is pointed out.

关 键 词：信道译码 深度学习 LDPC码 Polar码 

分 类 号：TN92[电子电信—通信与信息系统]