一种基于正反馈的对等网络拓扑获取方法

精确有效的对等网络测量方法是解决其建模和网络设计优化难题的重要基础.对等网络是Internet上的一层覆盖网络,网络协议多样,节点及节点间的关系变化迅速,获得精确完整的对等网络拓扑数据面临很大困难.研究对等网络协议特点、分析特定的对等网络结构实体成为认识对等网络拓扑特性的一种可选研究方案.以Gnutella网络为测量对象,构造了正反馈结构的分布式Gnutella拓扑测量系统D-crawler;分析了系统实现中的主要算法;定义了拓扑数据准确性和完整性评价指标;实验验证了测量系统的性能.实验结果表明,D-crawler系统具有较好的节点信息获取速度,能够得到反映Gnutella网络特征的拓扑数据,数据准确.

A P2P Topology Measuring Method Based on the Positive Feedback Strategy

Measuring and analyzing the topological properties of P2P networks will provide guidelines for their further optimization and development.However,it seems infeasible to capture a complete and precise snapshot of all the P2P networks due to the variety of their protocols and the dynamic characteristics of the networks.An alternative way is to measure the specific P2P network instance by studying its protocol details and constructing a high-speed topology crawling system.In this paper,two most difficult problems confronted by the crawling systems are discussed,a P2P network topology measuring framework using the positive feedback strategy and its key algorithms are presented,based on which the measured Gnutella network is basically taken as an example,the distributed topology crawling system(called D-crawler)is implemented,and the evaluation methods of the framework are defined.The performances of various topology crawling systems are compared,and the accuracy and completeness of the D-crawler system are analyzed in detail.The results show that the D-crawler system can collect more accurate,complete,and stable topology data of the Gnutella network with relatively shorter crawling period and fewer hardware requirements.The measuring framework and algorithms can be applied to other P2P networks with slight modification of the D-crawler system.