摘要
为了解决基于统计特征的P2P协议识别中,因特征选择不当而引起的识别准确率低的问题,采用免疫粒子群算法(Immune-PSO)选取最优特征子集,选择出最能区分P2P协议的特征子集。实验结果表明,该算法较标准粒子群算法具有更高的全局搜索能力,能更准确地找出最优特征子集,该方法能有效地提高协议的识别率,对常见的P2P协议如BitTorrent、eMule等有高达90%的识别率。
To solve the problem that the low rate of recognition accuracy about P2P traffic identification which caused by feature selection, an Immune-PSO hased method is proposed to select subset features. The Immune-PSO has hetter capahility of glohal search than standard PSO, so it can accurately identify the appropriate set of features better. Experimental results show that this method can effectively improve the recognition rate of common P2P protocols such as BitTorrent, eMule, etc with a precision about 90%.
出处
《计算机工程与设计》
CSCD
北大核心
2011年第10期3301-3304,共4页
Computer Engineering and Design
基金
国家973重点基础研究发展计划基金项目(2007CB311106)
关键词
免疫粒子群
点到点
协议识别
特征统计
lmmune-PSO
P2P
protocol identification
statistical features
