长程突发通信量的分数自回归预测

Fractional Autoregressive Prediction for Long Range Bursty Traffic

以数据包传输的通信量在不同网络条件下均表现出自相似性,自相似通信量在各时间尺度上均具有长程突发特性,其是以泊松过程为模型所描述的短程相关通信量所无法描述的。近来对自相似通信量的高精度测量与研究证实:网络中广泛存在的重尾特性是通信量自相似产生的原因。同时充分提取通信量的自相似性与重尾特性相关信息,是准确预测长程突发通信量的关键。在一种α-基于平稳过程的自相似通信量模型基础上,提出两种独立的自回归预测方法:FAR(Fractional Auto Regressive)预测、FNAR(Fractional Nonlienar Auto Regressive)预测。对这两种预测值进行不同方案的混合预测得到最终预测结果,进一步提高预测精度。

The traffic with data packet transmission in various network conditions exhibits convincingly self-similarity causing the long range burstiness which cannot be captured by traditional telecommunication traffic models based on Poisson process or Markov process. The updated explicit high-resolution measurement and researches for the traffic reveal that the heavy tallness existing extensively in the network brings about the self-similarity of the traffic. The information extraction from the self-similarity and long range dependence is the key fact for the exact prediction of the long range bursty traffic. Two distinctive AutoRegressive predictors based on a-stable self-similar traffic model were presented. The predictors including FAR（Fractional AutoRegressive）, FNAR（Fractional Nonlinear AutoRegressive） can minimize the dispersion according to the criteria with infinite variance. The final predicted values with the different schemes were obtained by combining the previous two individual predicted values for the higher predicted precision.