网络流量模型的非线性特征量的提取及分析  

Analysis on Network Flow Time Sequences and Extraction of Nonlinear Characteristic Quantities

作　　者：刘东林[1] 帅典勋[2] 

机构地区：[1]华东理工大学计算机科学系,上海200237 [2]清华大学智能技术与系统国家重点实验室,北京100084

出　　处：《电子学报》2003年第12期1866-1869,共4页Acta Electronica Sinica

基　　金：973计划国家重点基础研究发展规划项目(No.G1 9990 32 70 7);国家自然科学基金项目 (No .697730 37);国家自然科学基金项目 (No.60 0 730 0 8);清华大学智能技术与系统国家重点实验室基金项目

摘　　要：本文基于相空间重构理论 ,在高维相空间中对网络流量的宏观和微观特性进行研究分析 .首先 ,提取网络流量的宏观非线性特征量 ,如关联维数、Kolmogorov熵和最大Lyapunov指数 ,实现了网络流量时序非线性动力学特性的定量分析 .然后 ,通过对四种典型突发性流量模型的多重分形谱的计算 ,揭示了流量模型不同层次的行为特征 ,并给出了刻画突发性流量的有效微观参数 .Many efficient approaches and analysis techniques are applied to analyze the macro and micro characterizes of network flow data. The attractors are reconstructed by making use of time-delay coordinates. Then the macro nonlinear characteristic quantities of the network flow time sequences such as the fractal dimension Kolmogorov entropy and the largest Lyapunov exponents are extracted in this multi-dimension phase-space. The study on the temporal characteristics of these three parameters shows that the network flow is featured by some chaotic behaviors. The multifractal spectrums of the four difference network flow data are calculated in order to characterize and recognize the dynamic structure of network flow data more deeply, and thus can be effectively exploited for the controlling and modeling of the network behaviors.

关 键 词：相空间重构 分形维数 网络流量 多重分形谱 时序分析 

分 类 号：TP393.01[自动化与计算机技术—计算机应用技术]