【摘要】 针对现有基于传输层特征的 P2P 识别方法的不足,分析了局域网环境下主流P2P 应用的客户端口连接数、失败连接数、TCP/UDP 服务端口连接数、所连接的远端 IP 数与远端端口数等传输层连接特征,根据 P2P 应用与普通 Internet 应用在这些连接特征上的显著差异,提出了一种适用于局域网环境的实时P2P主机识别算法,并以该算法为核心,设计和实现了实时 P2P 应用监测系统 traffMon。在实际网络环境中对 traffMon 系统进行了测试,结果表明,该系统能够对运行了主流 P2P 应用的计算机进行实时识别,并能够统计出对应主机的各类连接数、平均传输速率、P2P 应用的服务端口号等信息,以作为局域网网络管理的依据。更多还原

【Abstract】 In consideration of the limitations of the existing transport lager characteristics-based methods for identification of peer-to-peer (P2P) applications in LAN,a real-time algorithm for P2P host identification in LAN was proposed based on the analyses of transport layer characteristics of modem P2P applications such as numbers of client-port connections, failed connections,TCP/UDP server-port connections,remote IPs and Ports connections,and the differences of those characteristics between P2P applications and common Internet applications,and by using it as the kernel,a real-time P2P traffic monitoring system,named traffMon,was designed and implemented.The test in real network environments indicated that the traffMon could identify hosts on which modern P2P applications are running in real-time.In addition, the traffMon can give statistics such as varieties of connections,average data transfer rates,server port numbers of P2P applications,which are very useful for network administration in LAN.更多还原