面向用户的Internet链路状态探测与预报

User-Oriented Probing and Forecast of Characteristics of Internet Link

构造了一个面向Internet一般用户的探测链路当前状态与预报未来状态的框架.利用pathchar和packetpair等探测方法与工具来测量网络路径延迟,得到网络链路的带宽等状态参数,并采用TCPdump对用户所在局域网的网络业务进行收集与建模.用户根据探测的链路带宽和本地网络的业务状态,对未来状态进行预测,并主动控制自己的网络行为.利用EVT,提高了链路带宽估计的精度,使用FARIMA模型,对网络业务和链路状态进行预报.实际测量与预报数据表明本方法是正确有效的.

This paper proposes a frame for Internet users to characterize Internet link and forecasts the link status and network traffic in the near future.The methods of pathchar and packet pair are used to measure the latency of hops along a link and to calculate the bandwidth. TCPdump is introduced to collect the traffic of local LAN and build models for the traffic.An Internet user could predict the future status based on the link bandwidth and the local traffic, and then control his own network behavior.To improve the accuracy of the estimation of bandwidth, the extreme value theory (EVT)is deployed.The fractional autoregressive integrated moving average (FARIMA)models are used to predict network traffic and future link status.The measurements and the prediction data show that the proposed frame is correct and feasible.