摘要
针对结构化P2P网络因忽略逻辑拓扑与物理拓扑差异而导致的路由绕路问题,首先提出一种聚类算法,将物理位置紧邻的节点自组织为若干分布式聚类;然后以聚类为单位设计了关联节点物理拓扑相邻性和逻辑拓扑相邻性的节点标识方法,并且利用该方法改进了传统Kademlia模型;最后对改进模型的性能进行了理论分析和实验测试.结果表明,改进模型通过增强每一跳路由在物理拓扑中的合理性,达到了避免路由绕路的目的,路由算法性能较传统模型提升近15%.
This paper focuses on the detouring problem caused by ignorance of difference between logical topology and physical topology in structured P2P networks.Firstly,a distributed clustering algorithm is proposed to classify the nodes into several self-organized clusters.And then,a technique relating the physical vicinity and the logical vicinity is designed to improve the traditional Kademlia routing model.Finally,the performance of the improved model is analyzed theoretically and experimentally.The results verify that the improved model solve the detouring problem by improving the rationality of every hop in physical topology and the efficiency of routing algorithm is improved more than 15 percent.
出处
《小型微型计算机系统》
CSCD
北大核心
2010年第8期1549-1553,共5页
Journal of Chinese Computer Systems
基金
军队国防科技研究项目资助