基于FARIMA过程的网络业务预报与应用

在高速网络控制与带宽分配研究中，网络业务量预报是一个重要的问题。本文首先介绍自回归分数整合滑动平均（FARIMA，Farctal Autoregressive Integrated Moving Average)过程的概念及其具体形式，并给出了FARIMA过程的预报方法，然后基于FARIMA过程的最小均方误差预报方法，提出了一个具有补偿功能的网络自相似业务的预报方法，最后给出这种预报方式在网络控制研究中的应用。

Traffic Prediction and Its Application Using FARIMA Models

Netvrork traffic prediction is important for netvrork control and bandwidth alloca-tion. This paper first introduces how to get optimal forecasting values, and provides prediction procedure for FARIMA models. And then provides a prediction method for self-similar traffic with compensation function and also gives out an example to demonstrate how to use this method in network control.