摘要
P2P流量逐渐成为了互联网流量的重要组成部分,在对Internet起巨大推动作用的同时,也带来了因资源过度占用而引起的网络拥塞以及安全隐患等问题,妨碍了正常的网络业务的开展。文中提出了基于机器学习的P2P流量识别方案,并运用FCBF(Fast Correlation-Based Filter)特征选择算法形成了流量特征子集,构建了机器学习P2P流量识别模型并对比了几种常见的机器学习算法在流量识别方面的性能。测试实验结果表明,C4.5算法和贝叶斯网络算法都适合于P2P流量检测,其个别模型达到了90%以上的识别率。
P2P traffic has taken great portions in the network traffic.While having a significant impact on the Internet,it brings serious problems such as network congestion and traffic hindrance caused by the excessive occupation in the bandwidth.Proposes a P2P traffic identification based on machine learning.Firstly the FCBF（Fast Correlation-Based Filter）feature selection algorithm is used to select the attribute features subset,then P2P flows identification model is built and several machine learning algorithms are compared.The result showed that in P2P traffic identification based on machine learning algorithms,C4.5 and Bayesian network was feasible and the identification accuracy of some models can reach above 90 percent.
出处
《计算机技术与发展》
2010年第11期133-136,共4页
Computer Technology and Development
基金
国家自然科学基金(60973139
60773041)
江苏省自然科学基金(BK2008451)
江苏省级现代服务业发展专项资金
江苏高校科技创新计划项目(CX09B-153Z
CX08B-086Z)
南京邮电大学青蓝工程项目(NY206034
NY208011)
江苏省六大高峰人才项目(2008118)
关键词
对等网络
流量识别
机器学习算法
特征选择
P2P
identification of traffic
machine learning algorithm
feature selection
