基于非参数估计的应用层网络流协议识别
pplication layer protocol identification network flow based on Non-parametric estimation
本文通过对网络应用层中各种协议的流量特点进行分析,提出将网络流协议特征作为统计值的基本元素,利用基于非参数估计的Parzen窗分类方法,设计网络应用层协议的分类器。在确定了最优特征之后用网络流的自然数据来构建并改进分类模型并进行大规模的网络流协议的分类,得到分类结果并与参数估计模型分类结果进行比较。实验结果表明,基于非参数估计的Parzen窗估计的方法对于应用层网流流协议的识别的结果要高于参数估计方法并得到了很好的结果。
On the basis of the research of non-parametric estimation Parzen Window method and analysis for the various protocols flow in network application layer, the paper designs a classifier of network application layer protocols which uses the network flow protocol feature as basic elements of statistical value.After having determined the optimal feature the network protocols classification model is constructed and improved by using the natural data,and the classification result can be got. The rmethod is compared with the parametric estimation rmethod. The experiments results show the Non-parametric identification method based on Parzen Window estimation is better than that of parameter estimation .
孙广路、高明、李兰英、陈志峰
通信无线通信计算技术、计算机技术
模式识别非参数估计网络流分类
Pattern recognitionNon-parametric EstimationNetwork flow classification
孙广路,高明,李兰英,陈志峰.基于非参数估计的应用层网络流协议识别[EB/OL].(2010-12-10)[2025-08-16].http://www.paper.edu.cn/releasepaper/content/201012-348.点此复制
评论