网络流量分类与应用识别的研究*

首先介绍了网络流量分析的不同层次及机器学习领域的相关知识,分析了采用端口号映射及有效负载分析的方法进行流量分类与应用识别存在的问题;然后从网络流量的统计特征出发,重点介绍了机器学习中聚类和分类的方法在流量分类的应用和问题;最后基于聚类和分类在流量分类中的效用,指出了未来的研究趋势。

Study of network traffic classification and application identification

This paper introduced the different levels of traffic analysis and relevant knowledge of machine learning,and analyzed some problems in traffic classification technologies,including port number analysis and payload-based analysis.Considering that different applications generate different traffic characters,how to apply clustering and classify method of machine learning to implement traffic classification and discussed application identification.Finally,based on the efficiency of machine learning-based method,pointed out the possible research directions.