摘要
P2P应用的快速增长,带来网络拥塞等诸多问题,而传统的基于端口与有效载荷的P2P流量分类方法存在着很多缺陷。以抽取独立于端口、协议和有效载荷的P2P流的信息作为特征,用提出的基于ReliefF-CFS的方法选择流的特征子集,研究使用机器学习算法对P2P流量进行分类的方法,也研究了利用流的前向N个报文的统计信息作为特征,分类P2P流量的方法。实验结果显示提出的方法取得了较好的分类准确率。
As the rapid increase of P2P application, many network problems occur, while the traditional P2P traffic classification methods based on port, protocol and payload have many objections. Extract the attributes irrelevant to port, protocol and payload and use the methods of feature selection based on ReliefF-CFS to choose feature subset, this paper could classify P2P application flows with machine learning algorithms. At the same time, researched the P2P traffic classification with the statistical features of its N-forward information. The experiment shows that the method has good classification accuracy.
出处
《计算机应用研究》
CSCD
北大核心
2009年第9期3468-3471,共4页
Application Research of Computers
基金
中国博士后科学基金资助项目(20070410299)
广东省自然科学基金博士科研启动基金资助项目(7300450)
关键词
对等网
流量分类
特征选择
机器学习
P2P(peer-to-peer)
traffic classification
feature selection
machine learning