传统的业务模型大多基于Poisson模型或其改进形式,假定业务突发长度显负指数分布,近期真实网络流量分析表明很多信息源会产生在多时间尺度下具有自相似特性的信息流,该性质显著地影响宽带网的流量控制及其排队分析,本文综合地介绍了自相似的数学描述和性质,五种自相似业务流的产生方法,其中有两种给出了具体的算法,还总结了几种主要的估计Hurst参数的方法。 Most of the traditional business models are based on the Poisson model or its improved form. Assuming that the burst length of the business is exponentially distributed, the recent real network traffic analysis shows that many information sources produce information flows with self-similar characteristics on multiple time scales. This paper introduces the description and properties of self-similar math and the methods of generating five kinds of self-similar traffic streams. Two of them give concrete algorithms, and also summarize several The main method of estimating Hurst parameters.