计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (35): 75-79.

• 网络、通信、安全 • 上一篇    下一篇

动态网络增强自适应虚拟队列管理新算法

闻小帆   

  1. 南京工业职业技术学院,南京 210016
  • 出版日期:2012-12-11 发布日期:2012-12-21

Enhanced adaptive virtual queue algorithm for active queue management in dynamic network

WEN Xiaofan   

  1. Nanjing Institute of Industry Technology, Nanjing 210016, China
  • Online:2012-12-11 Published:2012-12-21

摘要: 针对动态网络主动队列管理算法中控制参数设置难题,提出了一种增强自适应虚拟队列管理新算法(EPAVQ)。结合TCP流体流模型运用经典控制理论分析自适应虚拟队列管理(AVQ)算法稳定性基础上,利用极点配置技术给出了AVQ算法控制参数的调整规则。同时,针对网络参数时变问题,通过对网络参数的在线估计,利用网络参数与控制器参数的确定关系实时调节控制参数,使得控制器能够适应网络参数的变化。通过ns-2仿真实验与现有的AVQ算法比较,EPAVQ算法具有更快的动态响应,更好的网络性能。

关键词: 拥塞控制, 主动队列管理, 自适应虚拟队列, 极点配置, 自调整

Abstract: A novel Enhanced Adaptive Virtual Queue(EPAVQ) algorithm based on pole assignment is proposed to cope with the control parameters of Active Queue Management(AQM) hard to tuning problem in dynamic network. Based on the stability analysis of Adaptive Virtual Queue(AVQ) algorithm using classical control theory combined with fluid flow model of TCP, the tuning rule of AVQ’s control parameter is described exploiting pole placement technique. The control parameters of EPAVQ algorithm are automatically tuned in response to on-line estimation of link capacity and traffic load so that EPAVQ algorithm can perform well for a wide-range of network conditions. Compared with AVQ, the ns2 simulation results show the EPAVQ algorithm exhibits good transient response and better performance when network load changes.

Key words: congestion control, active queue management, adaptive virtual queue, pole assignment, self-tuning