基于改进R/S估计算法的网络流量长相关性分析  

Analysis of Network Traffic Long Correlation based on Improved R/S Estimation Algorithm

作　　者：荣红佳 盛虎[1] 闫秋婷 RONG Hongjia;SHENG Hu;YAN Qiuting(School of Electrical and Information Engineering,Dalian Jiaotong University,Dalian 116028,China)

机构地区：[1]大连交通大学电气信息工程学院,辽宁大连116028

出　　处：《大连交通大学学报》2021年第2期114-119,共6页Journal of Dalian Jiaotong University

基　　金：辽宁省博士启动基金资助项目(20170520215);辽宁省教育厅自然科学基金资助项目(JDL2019014)。

摘　　要：R/S估计算法被广泛应用于随机信号的长相关性分析,但存在准确性和计算效率不高的缺点,很难应用于信号的长相关性的实时分析.为了提高传统R/S估计算法的准确度和计算效率,基于R/S估计算法中重新标度的方法进行了优化,并采用分数阶高斯噪声序列和真实的网络流量数据对算法进行验证.结果表明改进R/S估计算法在准确度和计算效率上都有明显提升,可以广泛应用于网络流量数据和水纹数据等具有长相关特性的随机信号实时处理.Aiming at the disadvantages of low accuracy and computational efficiency and the difficulty to realtime analysis of long correlation of signals,the accuracy and computational efficiency of the traditional R/S estimation algorithm is improved based on the rescaling method.The rescaling method of the R/S estimation algorithm is optimized,and the algorithm is verified by fractional Gaussian noise sequence and real network traffic data.The results show that the improved R/S estimation algorithm has obvious improvement in accuracy and computational efficiency,and it can be widely applied to real-time processing of random signals with long range dependence such as network traffic data and water wave data.

关 键 词：R/S估计算法 长相关 网络流量 

分 类 号：TP393.06[自动化与计算机技术—计算机应用技术]