Network Self-similarity Of Operations Research

Many measurements from the work packet networks have been collected and analyzed in the past several years. These studies prove that traditional models neglect an importance characteristic--self-similarity. In fact, the traffic models which have long-range dependence are more suitable for describing the really traffic process, such as the self-similar or fractal model. Moreover, we need a kind of traffic models that can be controled freely in the reasearch. We not only can increase the efficiency, but also can get away from the passive indetermination.In the paper, self-similar traffic models are studied deeply. Then we simulate the models with computer. Moreover, the traffic models are used to analyze the aggregation and decomposition of the network traffics, etc.. The last, we propose a standpoint on the physical background of the self-similar phenomena.In chapter 2 of the paper, we give the mathmatic definition of self-similar process, give out its some characteristics on the mathematic with physics and introduce how to examine the process if it is the self-similar process. The last, we compare some examination methods concisely.It has an important means for aggregation and decomposition of self-similar process. We resolve these problems in a certain degree in chapter 4. Moreover, we make use of the self-similar traffic models to allot network bandwidth and forecast network traffic.Some of the traffics have been proved to have the property of self-similarity. But in the running network, the mainly payload can be described by the short-range dependent traffic model successfully. Then why many of the short-range dependent traffic in the network can show self-similarity in the node? What's the physical background of the phenomena? We propose a standpoint in chapter 5.