结构化对等网测量方法研究

Research on Network Measurement of Structured P2P Network

网络测量是深入开展结构化对等网研究的基础,结构化对等网络协议设计、共享内容检索、态势感知乃至安全性的研究都需要以网络测量为前提.在节点分布对等、实时变化显著、未知瞬发扰动频繁的结构化对等网络中,获得其准确、完整的网络信息更是十分困难的.通过形式化分析结构化对等网节点搜索过程,研究节点信息在全网分布情况与查询返回率之间的关系,将历史测量数据与具体对等网特征信息相结合挖掘节点搜索优化策略,提出了一种网络资源占用显著降低、搜索速度较快、信息完备率较高的搜索测量优化方法.KAD网络是目前得到大规模部署运行的为数不多的结构化对等网络之一,以KAD网络为主要研究对象开发了KadCrawler对等网搜索系统,进行了大量测量和分析,验证了搜索优化方法的可行性和有效性;同时,对当前KAD网络拓扑结构特征、节点重名等现象进行了初步分析,发现KAD网络近年来发生了显著的变化.

Network measurement is the foundation for the in-depth research on P2P network. It’s a prerequisite for the P2P protocol design, shared content searching, situational awareness as well as research on the security of P2P network. In structured P2P network with decentralized peer to peer relationship, high dynamics and unpredictable instantaneous disturbance, achieving highly accurate and near-complete information retrieval is much more difficult. This paper formalizes the search （or crawl） process of structured P2P network, studies the relationship between the node’s route spreadness in the whole network and the query response rate in the midst of crawling, derives some improved search strategies from the knowledge of historic measurements and characteristics of specific P2P network, and proposes a improved search method with much lower bandwidth consumption, fast crawling speed as well as relatively high coverage of nodes in structured peer-to-peer network. KAD network is among the few of structured P2P networks that are extensively deployed. This research mainly concentrates on KAD network and develops a search tool called KadCrawler, upon which large amounts of measurements and analysis are conducted. The result shows that the proposed method is both feasible and effective. Lastly, an analysis on the topology and phenomenon of ID repetition reveals that KAD network has changed significantly over the years.