文献类型：期刊文献

中文题名：自相似数据流下的网络相变与临界现象

英文题名：Network phase transitions and critical behaviors under self-similar traffic model

作者：于强[1];帅典勋[1];韩卫丽[1]

机构：[1]华东理工大学计算机系,上海200237

年份：2003

卷号：18

期号：6

起止页码：535

中文期刊名：系统工程学报

外文期刊名：Journal of Systems Engineering

收录：CSTPCD;;CSCD:【CSCD2011_2012】；

基金：国家重点基础研究发展规划973资助项目(G1999032707);国家自然科学基金重点资助项目(60135010);国家自然科学基金资助项目(60073008);清华大学智能技术和系统国家重点实验室开放课题基金资助项目.

语种：中文

中文关键词：计算机网络;网络行为;自相似数据流;网络相变;临界现象;网络仿真

外文关键词：network behavior; network simulation; self-similar; phase transition

摘要：随着网络规模的不断扩大和网络用户的不断增加,目前的网络已经发展成为一个海量的非线性复杂系统.对网络非线性动力学的研究,已经引起了计算机和统计物理学等相关领域研究者的浓厚兴趣.论文提出了一种网络仿真模型,并在自相似数据流下进行了各种网络仿真实验.通过实验分析,发现了在自相似数据流下序参数d在整个网络从非拥塞进入间歇性拥塞,再到持续性拥塞的过程中起了重要的标示作用.通过对队列长度的功率谱分析,揭示了序参数d与1/f,1/f2分布之间的关系.
With the ever-increasing growth of the Internet, networks have become a giant_complex system. Research on the computer network dynamics has attracted great attention in the fields of computer science and statistical physics. In this paper, a simulator for computer network dynamics is put forward, and then the simulation results on network behavior under self-similar fractal traffic model are shown. It has been found that the order parameter d plays a very important role in the phase-transition of the whole network performance, which is referred to from non-congestion to intermittent congestion and further to durative congestion. Through the power spectrum analysis for buffer queuing length, the relations between the order parameter d and 1/f,1/f^(2) distribution are revealed.