应用科学学报 ›› 2015, Vol. 33 ›› Issue (2): 155-166.doi: 10.3969/j.issn.0255-8297.2015.02.005

• 通信工程 • 上一篇    下一篇

一种高精度低负载的可用带宽测量机制

周逸秋1, 陈兵1, 钱红燕1, 吕宗磊2   

  1. 1. 南京航空航天大学计算机科学与技术学院,南京210016
    2. 中国民航大学中国民航信息技术科研基地,天津300300
  • 收稿日期:2014-10-16 修回日期:2014-12-03 出版日期:2015-03-30 发布日期:2014-12-03
  • 作者简介:陈兵,教授,博导,研究方向:计算机网络,E-mail:cb_china@qq.com
  • 基金资助:

    中国民航信息技术科研基地开放课题基金(No. CAAC-ITRB-201301)资助

Accurate and Low Overhead Mechanism for Measuring Available Bandwidth

ZHOU Yi-qiu1, CHEN Bing1, QIAN Hong-yan1, LÜ Zong-lei2   

  1. 1. College of Computer Science and Technology, Nanjing University of Aeronautics and
    Astronautics, Nanjing 210016, China
    2. Information Technology Research Base of Civil Aviation Administration of China,
    Civil Aviation University of China, Tianjin 300300, China
  • Received:2014-10-16 Revised:2014-12-03 Online:2015-03-30 Published:2014-12-03

摘要: 端到端的可用带宽是网络测量的重要性能指标之一. 针对现有的大多数测量工具及
方法在估算可用带宽前需要大量时间分析和测量等问题,在全面分析了网络中背景流特征的
基础上,提出一种高精度低负载的可用带宽测量机制. 该机制考虑了高速网络环境中丢包的情
况,建立网络利用率与探测速率间的关系模型,结合扩展卡尔曼滤波实时获得最新的端到端
的可用带宽. 该机制不需要事先知道瓶颈链路容量,探测速率可低于可用带宽,降低了突发背
景流对测量的影响. 数值模拟表明,所提出的机制能快速响应突发背景流,准确高效地测得端
到端的可用带宽.

关键词: 可用带宽, 突发背景流, 扩展卡尔曼滤波

Abstract: The available end-to-end bandwidth is an important specification in network
measurement. Most tools need to spend much time in measuring and analyzing before
calculating the available bandwidth. This paper proposes a low overhead mechanism for
accurate and fast measurement of the available bandwidth by analyzing the cross traffic
effect in the Internet. A model is established to reflect the relationship between utilization
and the sending rate of probe packets, which is combined with an extended Kalman filter to
obtain the new available bandwidth. This mechanism does not require any prior knowledge
of the bottleneck link capacity and can reduce the effect of accuracy. Besides, the sending
rate of probing packets can be much lower than the available bandwidth. Performance
of the mechanism is verified numerically, showing fast response to burst cross traffic, low
estimation error, and short convergence time.

Key words: available bandwidth, burst cross traffic, extend Kalman filter

中图分类号: