摘要
流量识别是指根据网络流以及流中数据报文的某些信息将网络上的流分成若干既定的类别.随着网络的高速化,业务应用和协议日趋复杂,传统的基于端口的流量识别方法已经不够准确,因此各种新的识别方法成为研究热点.本文在介绍流量识别的基本概念、流量识别的评价指标的基础上,对目前正在使用或研究的流量识别方法进行了分析比较,指出了现有各种方法的优点和不足,最后给出未来流量识别技术的发展趋势和研究方向.
Traffic identification refers to deviding the flow on the network into a number of the established categories making use of network flow and the certain information of packets in the flow.Due to high speed of network and complexity of new protocol,traditional port-based traffic Identifying method is becoming much more inaccurate.Therefore,a variety of new traffic Identifying methods become a research hotspot.In this paper,several basic concepts of traffic Identifying are introduced first;then the advantages and disadvantages of the traffic Identifying methods being used or researched are analyzed;finally the research directions are presented.
出处
《小型微型计算机系统》
CSCD
北大核心
2010年第8期1514-1520,共7页
Journal of Chinese Computer Systems
基金
国家"九七三"重点基础研究发展计划项目(2008CB317111)资助
教育部科学技术重点项目(208117)资助
重庆市教委科学技术项目(KJ080528)资助
关键词
流量识别
净荷特征
行为特征
机器学习
traffic Identifying
payload signature
behavior signature
machine learning
