【作者】 于翠波； 王刚； 勾学荣；

【Author】 Cuibo Yu Gang Wang Xuerong Gou School of network education,Beijing University of Posts and Communications,Beijing China

【机构】 北京市西城区新外大街28号院网络教育学院；

【摘要】 随着P2P(Peer to Peer)应用的进一步推广,P2PSIP(Peer to Peer Session Initiation Protocol)、异构DHT(Distributed Hash Tabk)通信等新应用的提出,超节点(SuperNode,SN)选举问题显得越来越突出。超节点选举算法由于消息与时间复杂度与节点数目密切相关,而P2P网络本身节点数目巨大,扰动效应比较明显,如果直接将一般的选举算法用到P2PSIP网络或异构DHT网络中,会造成系统难以正常运行。本文综合各方面的研究,提出了分区域的超节点选举算法。本算法优点是可以大大降低消息复杂度和时间复杂度,缺点是选出的节点是局部(区域)最优,而不是全局最优,考虑到P2P网络的扰动特性,这种本地最优算法的性能是可以接受的。更多还原

【Abstract】 With the development of P2P(Peer to Peer),the emerging of application of P2P SIP(Peer to Peer Session Initiation Protocol)and heterogeneous DHT(Distributed Hash Table),the problem of Supernode(SN)election was brought forth.Be- cause of message complexity and time complexity were decided by the number of node,and the number of P2P network is huge, moreover,there was churn in P2P’s topology,if we use general algorithms of election in P2P SIP and heterogeneous DHT net- work,the algorithms would be too complex to run normally.In this paper,a new election algorithm based on area partition was brought forward.This algorithm could decrease the complexity of SN election,the drawback was that the elected SN was not the optima of whole network.Combining the chum and requirement of P2P network,the performance of local optima was acceptance.更多还原