作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2024, Vol. 50 ›› Issue (5): 306-312. doi: 10.19678/j.issn.1000-3428.0066870

• 开发研究与工程应用 • 上一篇    下一篇

基于流量自相似性的网络队列管理算法

魏德宾1,2, 杨力1, 潘成胜1, 沈婷2   

  1. 1. 南京理工大学自动化学院, 江苏 南京 210094;
    2. 大连大学信息工程学院, 辽宁 大连 116622
  • 收稿日期:2023-02-06 修回日期:2023-07-18 发布日期:2023-10-12
  • 通讯作者: 魏德宾,E-mail:weidebin@163.com E-mail:weidebin@163.com
  • 基金资助:
    国家自然科学基金(U21B2003, 61931004)。

Queue Management Algorithm for Network Based on Traffic Self-similarity

WEI Debin1,2, YANG Li1, PAN Chengsheng1, SHEN Ting2   

  1. 1. School of Automation, Nanjing University of Technology, Nanjing 210094, Jiangsu, China;
    2. College of Information Engineering, Dalian University, Dalian 116622, Liaoning, China
  • Received:2023-02-06 Revised:2023-07-18 Published:2023-10-12
  • Contact: 魏德宾,E-mail:weidebin@163.com E-mail:weidebin@163.com

摘要: 网络流量的自相似性会导致数据突发状态持续,传统队列管理算法无法对网络流量突发状态进行预测,从而影响网络端到端时延、丢包率和吞吐性能。针对该问题,提出一种基于网络流量预测的主动队列管理算法P-ARED。基于网络流量的均值和方差给出网络流量等级的概念,讨论网络流量等级转移概率与Hurst参数之间的关系,提出基于贝叶斯估计思想的网络流量等级预测方法。在此基础上,在对自相似网络流量环境下的平均队列长度、缓存队列长度最小阈值等参数优化设置的基础上,基于Hurst参数和自相似流量等级预测结果,重新设计ARED算法中分组丢弃概率的计算方法,以提高缓存队列长度的稳定性。仿真结果表明,P-ARED算法与对比的主动队列管理算法相比,降低了网络端到端时延和丢包率,提高了端到端吞吐性能,其中平均吞吐量最高提升7.63%,平均时延最多降低17.52%。

关键词: 网络流量, 自相似性, 主动队列管理, 随机早期检测, 流量等级

Abstract: The self-similarity of network traffic result in a continuous burst state of data in the network. In this context, conventional queue management algorithms cannot predict the network traffic burst state in advance, thus resulting in end-to-end delays, high packet loss rates, and deteriorated network throughputs. To solve these problems, an Active Queue Management(AQM) algorithm based on prediction of the network traffic, P-ARED, is proposed. Based on the mean value and variance of network traffic, the concept of network traffic level is proposed. The relationship between the probability of network traffic level transition and Hurst parameters is discussed, and a network traffic level prediction method based on Bayesian estimation is proposed. On this basis, based on the optimization of parameters such as the average queue length and the minimum threshold for cache queue length in self-similar network environments, and based on the Hurst parameters and self-similar traffic level prediction results, the calculation method for group dropout probability in the ARED algorithm is redesigned to improve the stability of cache queue length. Simulation results show that, compared with the AQM algorithms in the existing literatures, the P-ARED algorithm reduces the end-to-end delay and packet loss rate, improves the end-to-end throughput performance, and increases the average throughput by 7.63% at the maximum. Additionally, it reduces the average packet loss rate by 17.52% at the maximum.

Key words: network traffic, self-similarity, Active Queue Management(AQM), Random Early Detection(RED), traffic level

中图分类号: