智能TCP拥塞控制算法研究进展  

Research Progress of Intelligent TCP Congestion Control Algorithm

作　　者：李英华[1] 杨琳 LI Yinghua;YANG Lin(State Radio Monitoring Center,Beijing 100037,China;Beijing OET Spectrum Institute Corporation,Beijing 100041,China)

机构地区：[1]国家无线电监测中心,北京100037 [2]北京东方波泰无线电频谱技术研究所有限公司,北京100041

出　　处：《数字通信世界》2022年第5期10-12,共3页Digital Communication World

摘　　要：拥塞控制一直是网络通信领域的重要研究课题,它可以避免因网络拥塞而导致的传输质量严重下降,同时确保网络带宽的高效利用。TCP在互联网的发展过程中,扮演着不可或缺的重要的角色,是实现网络拥塞控制的重要途径。近年来,随着网络环境日趋复杂,传统的拥塞控制算法越来越无法适应复杂多变的网络环境,但是机器学习、深度学习和强化学习的兴起,给拥塞控制提供了新的研究思路。文章分析了目前基于机器学习的几种算法原理,分析其优缺点,并提出未来的研究方向。Congestion control has always been an important research topic in the field of network communication.It can avoid the serious reduction in transmission quality caused by network congestion and ensure efficient bandwidth utilization.TCP protocol plays an indispensable role in the development of the Internet.It is also an important way to realize network congestion control.Recent years,with the increasing complexity of the network environment,traditional congestion control algorithms cannot adapt to the complex and changeable network environment,while the rise of machine learning,deep learning,and reinforcement learning provides new research ideas for congestion control.This paper analyzes operating principles and advantages/disadvantages of several machine learning algorithms and proposes future research directions.

关 键 词：计算机网络 TCP协议 拥塞控制 基于机器学习 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]