摘要
保持对等网络与底层网络的拓扑良好契合是保障对等网络性能的重要条件.以对等网络系统CAN为基础,将底层网络拓扑建模为高维欧氏空间,使用主成分分析法将底层网络拓扑映射到低维CAN逻辑空间,从而根据节点在底层网络中的位置决定其在对等网络中的标识符.实验结果表明,该算法能够有效减少对等网络链路延时,提高路由效率.
It is desirable to make the topology of P2P networks match that of underlying network. This paper models the underlying network topology as an Euclidean space and uses PCA to map this higher dimensional space to the lower dimensional logical space of CAN. So a peer's identifier is determined by its location in underlying network. The experimental results indicate that this algorithm can reduce the overlay link latency and improve P2P routing performance significantly.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2007年第1期56-59,65,共5页
Journal of Shanghai Jiaotong University
关键词
对等网络
分布式哈希表
主成分分析
邻近标识符选择
peer-to-peer, distributed hash table, principal component analysis
proximity identifier selection
作者简介
蒋俊杰(1975-),男,江苏通州人,博士生,主要研究方向为计算机网络与分布式系统.
汪为农(联系人),男,教授,博士生导师,电话(Tel.):021—62932919E-mail:wnwang@sjtu.edu.cn.