业务自相似性分析方法研究

Analytical Approaches to Traffic Self-similarity

自相似性是真实网络业务数据中的一个主要特性,其强度大小对网络性能、流量控制及排队分析结果有重要影响,因此,准确计算表征自相似性强弱的参数Hurst值是一项很有意义的研究课题.从时域与频率域两个方面出发,综述了各种计算方法的基本原理,介绍了各自统计量定义方法,并对各种方法的结果进行横向比较,分析了各自的影响因素,最后对下一步研究方向进行了预测.

It is significant to calculate accurately Hurst Parameter value representing the degree of self-similarity because self-similarity is the primary characteristic of traffic running in the actual networks and its intensity has an important influence on network performance, traffic control and queuing result. This paper reviews the basic principle of various published methods to estimate Hurst value from the perspectives of time domain and frequency domain, defines or presents in detail the statistics used by each method. Based on the comparison made among the methods, we analyze the impact factors and provide some guidance and forecast for future studies.