自相似模型下的交换式网络仿真

Simulation of the switched network based on self-similar model

传统模式下的网络仿真,报文到达均服从的是一种具有短相关特性的泊松分布。而经过大量业务流量监测表明,网络流量实际呈现出的确是一种具有长相关特性的自相似分布,这种特性对网络流量建模、性能分析、接纳控制等产生了重要影响。在对自相似特性深入分析的基础上,利用分型布朗运动模型的RMD算法产生自相似序列来模拟网络业务,并对该业务流特性下的交换式以太网进行了仿真实验。结果表明业务量的自相似性对交换式网络的各项性能影响很大,这与传统流量模型形成鲜明对比。

Under traditional network simulation,this paper normally uses Poisson distribution to simulate which is a kind of short-range dependence process.As going deep into network research,large number of data and experimental results show that the traffic of network exhibits a self-similar property with long-range dependence characteristic.This characteristic has significant influence on network performance and call admission control.Based on Fractional Brownian Motion model,a self-similar traffic sequence generation method of RMD algorithm is proposed.Furthermore,simulation and modeling for switched network is implemented based on this traffic model.After analyzing each network performance indicator,which has demonstrated that self-similarity has great influence on network performance,and has great contrast to traditional flow models.