基于形式化方法的网络业务流自相似性研究

在当今世界,计算机网络发挥了非常重要的作用,伴随着科学技术日新月异的发展,网络研究人员逐渐对网络的发展发现了一种新的现象—网络业务流突发与团结。网络业务流呈现出自相似的特性,传统的业务流建模方法不再适用。近年来,许多网络研究人员测量与分析现实实际运行中的网络,发现了一种具有统计学原理上的自相似性质也就是长相关性存在于真实的网络业务流。传统的网络业务流模型在描述实际网络业务流时,忽略了这个重要的特性,不能真实地刻画网络业务流的实际情况,网络业务流呈现出的自相似性对网络数据传输过程中的吞吐量和利用率具有极其重大的影响,最终结果可能将产生自相似网络业务流与传统模型在数据传输性能上产生很大的差异性。要想将网络数据传输的真实特性得以表达就需要对模型重新建立,以此对网络业务流的自相似性进行描述。该文所提出的需要建立的模型能够将不同的网络流量性质特质进行很好的拟合,以至于对网络特性的分析及网络模型的构建和对评估网络的特性提供了很好的理论依据基础。

Nowadays,computer network plays a greatly important role in our lives,in terms of development of new techniques,research workers discovered that traditional traffic model could not explain the burst and clot phenomenon of network traffic which contributed to the self-similarity characteristics of network traffic.In past few years,a number of network research workers measured and analyzed some real networks,the research results demonstrated that real network traffic had statistical self-similarity and,that is to say,long range-dependent.These results prove that traditional network traffic models neglect this important characteristic,so they are not suitable for describing the behavior of real network traffic.The self-similarity of network traffic has severe effect on throughput and utilization of network data transfer process.It leads to significant differences of results between the tradltional model and self-similar network traffic model in network performance.It is quite necessary to establish new model to reflect the features of self-similar phenomenon.The model brought forward in this thesis can characterize the statistical behavior of network traffic to some extent and it provides a solid foundation for construction,analysis and evaluation of the network.