基于分组抽样的P2P超级节点推测  

Inferring the P2P’s Super Peers with Packet Sampling

作　　者：韦安明[1,2] 王洪波[1] 程时端[1] 

机构地区：[1]北京邮电大学网络与交换技术国家重点实验室,北京100876 [2]广播电视规划院,北京100866

出　　处：《电子与信息学报》2009年第6期1513-1516,共4页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(90604019,60502037);高等学校博士学科点专项科研基金(200800131019);国家863计划项目(2006AA01Z235,2007AA01Z206);新世纪优秀人才支持计划(NECT-07-0109)资助课题

摘　　要：基于小世界规律和幂率特征,该文提出了P2P超级节点检测方法IPS。IPS通过分组抽样检出P2P种子节点,然后根据P2P节点间的流量关联特征检出其余的P2P节点,并将连接度大的P2P节点作为超级节点输出。基于互联网数据的实验验证了该法的正确性。Based on the small-world and power-law characters of P2P, the IPS is proposed to identify the P2P super nodes. IPS adopts packets sampling to identify seed nodes. Then according to the traffic relationship between the P2P nodes, IPS identifies all the other P2P nodes originating from the seed nodes. Finally, IPS regards the nodes with large connection degree as super nodes and outputs them. The validity of proposed method is verified by the experiments using Internet traces.

关 键 词：P2P网络 网络流量 网络测量 流量控制 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]