基于多协议多阶段的P2P内容分析技术研究  

Research on Multi-stage and Multi-protocol-based P2P Content Analysis Technology

作　　者：肖智鹏 张宏莉 叶建伟 XIAO Zhipeng, ZHANG Hongli, YE Jianwei （School of Computer Science ＆ Technology, Harhin Institute of Technology, Harbin 150001, China ）

机构地区：哈尔滨工业大学计算机科学与技术学院,哈尔滨150001

出　　处：《智能计算机与应用》2011年第1X期43-46,共4页Intelligent Computer and Applications

摘　　要：当前网络环境中，P2P下载软件流量占用了宽带接入的大量带宽，这造成了网络带宽的巨大消耗，因此，对P2P流量的检测和控制是十分必要的。提出了一种基于多协议多阶段的深度数据包检测的P2P流量识别的方法，该方法较已有方法具有识别准确度高、执行效率高、扩展性强的特点。基于这种方法，实现了一个多协议可扩展的P2P流量识别系统，系统以分层化可扩展插件的形式设计，可以对各种不同协议的P2P流量业务，进行高效、准确的识别。实验表明，系统能够以99．90％以上的准确度完成对P2P流量业务的识别。In the present network environment, P2P download software flow takes too much bandwidth with broadband access, which creates a huge consumption of network bandwidth. Therefore, it is necessary to detect and control P2P download flow. This paper puts forward a method of multi-stage and multi-protocol-based P2P flow recognition which has the ability of deep packet inspection. This method has many good qualities such as high recognition accuracy, high efficiency and high extensibility. Based on this method, an extensible and multi-protocol P2P traffic flow recognition sys- tem is designed in this paper. This system is designed in the form of layered and extensible plug-ins, and can identify P2P traffic flow based on differ- ent ki- nds of protocols efficiently and accurately. Experiments show that this system can identify P2P traffic flow with the accuracy rate of more than 99.90%.

关 键 词：BITTORRENT EMULE 协议分析 流量识别 对等网络 

分 类 号：TP393[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]