摘要： 网络业务具有自相似性。为了有效地对自相似数据业务流进行控制、管理与疏导，必须快速获得业务流的Hurst参数。该文通过分析自相似业务流的特性，在研究小波分解、经验模态分解和R/S估值过程的基础上，提出一种快速(数据量每减少50%，估值时间减少85.7%)、高精度的Hurst指数估值方法，可直接用于网络业务的实时监测、调度和路由仲裁。

关键词: 自相似, 小波变换, 经验模态分解, Hurst指数

Abstract: The data traffic in the Internet is self-similar. In order to realize the control, management and grooming of the data traffic efficiently, the traffic self-similarity should be gained as fast as possible. Base on the analysis of the self-similar data traffic and the characteristics of the methods including wavelet, EMD and R/S, a fast computation process with high precision is put forward for all kinds of online monitoring, flow control and routing arbitration.

Key words: self-similar, Wavelet transform, EMD, Hurst value

中图分类号: 

• TP301.6