基于半监督的网络流精细化识别方法研究  

Research on network traffic fine recognition based on semi-supervision

作　　者：陈绣瑶[1] 

机构地区：[1]广东女子职业技术学院应用设计系,广东广州511450

出　　处：《广州大学学报（自然科学版）》2017年第5期67-71,共5页Journal of Guangzhou University:Natural Science Edition

基　　金：2017年度广州市哲学社会科学发展"十三五"规划资助课题(2017GZGJ10)

摘　　要：为了实现复杂网络环境下网络应用的精细化识别管理,文章在半监督的机器算法基础上,集合HMM和VSM的优点,提出基于HMM和VSM混合模型的识别算法,利用网络流统计特征属性实现网络流精细化识别.实验结果显示HMM和VSM混合算法能够有效识别加密协议及复杂网络协议的应用,并且达到较高的识别率,其分类的精度明显优于传统识别算法.In order to realize the fine identification management of the network application in the complex network environment,this paper combines the advantages of HMM and VSM on the basis of semi supervised machine algorithm,and proposes a hybrid algorithm based on HMM and VSM,using the network flow statistical characteristics of flow properties of fine recognition. The experimental results show that the HMM and VSM hybrid algorithm can effectively identify the encryption protocol and the application of complex network protocols,and achieve high recognition rate,and the classification accuracy is superior to the traditional recognition algorithm.

关 键 词：网络流 精细化识别 半监督 HMM模型 VSM模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]