基于时变Hurst指数的网络流量分析
|
摘要
现在网络业务流量大并且流量数据分析越来越复杂,数据主要表现出长相关特性。好的流量模型必须能够准确描述网络实际流量的特征,才能准确预测流量状况。因此本文在对网络流量进行研究分析时,以已知Hurst指数的分形高斯噪声(FGN)序列仿真信号为主要研究对象,在原有Hurst指数估计方法的基础上加入滑动窗,对仿真序列进行时变Hurst指数分析。实验结果显示,与传统算法的估计结果相比,时变Hurst指数估计方法能更好地反映网络流量的局部长相关特性。
Now the network traffic is large and the analysis of traffic data is more and more complex.Besides,the data mainly shows long correlation characteristics.A good traffic model must be able to accurately describe the characteristics of the actual network traffic,and then it can accurately predict the flow state.So when we make the research and analysis of network traffic,taking the Fractal Gauss Noise(FGN) sequence simulation signal with known Hurst index as the main research object,adding the sliding window to the original Hurst index estimation method to analysis time-varying Hurst index about the simulation sequence.The experimental results show that the time-varying Hurst index estimation method can better reflect the local correlation characteristics of network traffic compared with the traditional algorithm.
Now the network traffic is large and the analysis of traffic data is more and more complex.Besides,the data mainly shows long correlation characteristics.A good traffic model must be able to accurately describe the characteristics of the actual network traffic,and then it can accurately predict the flow state.So when we make the research and analysis of network traffic,taking the Fractal Gauss Noise(FGN) sequence simulation signal with known Hurst index as the main research object,adding the sliding window to the original Hurst index estimation method to analysis time-varying Hurst index about the simulation sequence.The experimental results show that the time-varying Hurst index estimation method can better reflect the local correlation characteristics of network traffic compared with the traditional algorithm.
引文
[1]张宾,杨家海,吴建平,Intemet流量模型分析与评述[J].软件学报,2011,01:115-131.
[2]魏进武,邬江兴.网络流量长相关特性估计算法性能评估[J].计算机工程,2007,02:22-24+27.
[3]王晖,季振洲,朱素霞.自相似网络流量模型研究[J].智能计算机与应用,2013,02:34-41.
[4]魏进武,张进,邬江兴.网络流量长相关特性的滑窗时变估计算法[J].计算机研究与发展,2008,03:436-442.
[5]韩忠玲.分形高斯噪声Hurst参数估计的实验评价[D].华东师范大学,2008.
[6]朱婉莹,安晓光,王小卫.基于自相似网络流量模型的分析[J].网络安全技术与应用,2008,(12):11-12.
[7]Ivan Osorio,Mark G.Frei.Hurst parameter estimation for epileptic seizure detection[J].Communications in Information and Systems,2007,7(2):167-176.
[2]魏进武,邬江兴.网络流量长相关特性估计算法性能评估[J].计算机工程,2007,02:22-24+27.
[3]王晖,季振洲,朱素霞.自相似网络流量模型研究[J].智能计算机与应用,2013,02:34-41.
[4]魏进武,张进,邬江兴.网络流量长相关特性的滑窗时变估计算法[J].计算机研究与发展,2008,03:436-442.
[5]韩忠玲.分形高斯噪声Hurst参数估计的实验评价[D].华东师范大学,2008.
[6]朱婉莹,安晓光,王小卫.基于自相似网络流量模型的分析[J].网络安全技术与应用,2008,(12):11-12.
[7]Ivan Osorio,Mark G.Frei.Hurst parameter estimation for epileptic seizure detection[J].Communications in Information and Systems,2007,7(2):167-176.