摘 要
计算机网络拥塞控制是确保网络高效稳定运行的关键技术，随着互联网规模的不断扩大和应用类型的日益多样化，传统拥塞控制算法面临诸多挑战。为此，本文深入研究计算机网络中的拥塞控制技术，旨在探索适应现代网络环境的有效方法。研究基于对现有拥塞控制机制的全面分析，包括TCP的传统拥塞控制算法如慢启动、拥塞避免等，并结合当前网络流量特征，提出一种融合机器学习的新型拥塞控制算法。该算法通过构建流量预测模型，实时感知网络状态并动态调整拥塞窗口大小，在保证数据传输效率的同时降低丢包率。实验结果表明，与传统算法相比，新算法在网络负载均衡、响应速度等方面具有显著优势，特别是在高带宽延迟乘积环境下表现更为突出。本研究创新性地将机器学习引入拥塞控制领域，为解决复杂网络环境下的拥塞问题提供了新思路，不仅丰富了拥塞控制理论体系，也为实际网络优化提供了有效方案，对提升网络服务质量具有重要价值。
关键词：计算机网络拥塞控制 机器学习 流量预测模型
Abstract
Congestion control in computer networks is a critical technology for ensuring efficient and stable network operation. As the scale of the Internet continues to expand and application types become increasingly diverse, traditional congestion control algorithms face numerous challenges. This study delves into congestion control technologies in computer networks, aiming to explore effective methods suitable for modern network environments. Based on a comprehensive analysis of existing congestion control mechanisms, including traditional TCP congestion control algorithms such as slow start and congestion avoidance, this research proposes a novel congestion control algorithm that integrates machine learning by considering current network traffic characteristics. The algorithm constructs a traffic prediction model to perceive network states in real-time and dynamically adjust the congestion window size, thereby enhancing data transmission efficiency while reducing packet loss rates. Experimental results demonstrate that compared with traditional algorithms, the new algorithm exhibits significant advantages in network load balancing and response speed, particularly in high bandwidth-delay product environments. Innovatively introducing machine learning into the field of congestion control, this study provides new approaches to addressing congestion issues in complex network environments, enriching the theoretical fr amework of congestion control and offering practical solutions for network optimization, which is of great significance for improving network service quality.
Keyword:Computer Network Congestion Control Machine Learning Traffic Prediction Model
目 录
1绪论 1
1.1计算机网络拥塞控制背景与意义 1
1.2拥塞控制技术研究现状综述 1
1.3本文研究方法与创新点 1
2拥塞控制基本原理 2
2.1拥塞产生的原因分析 2
2.2常见拥塞控制机制解析 3
2.3拥塞控制性能评价指标 3
3主流拥塞控制算法 4
3.2显式拥塞通知机制研究 4
3.3新兴拥塞控制算法进展 4
4拥塞控制应用实践 5
4.1数据中心网络拥塞管理 5
4.3物联网环境下的拥塞应对 6
结论 7
参考文献 9
致谢 10
