作者: A. Agrawal , H. Casanova
DOI: 10.1109/CCGRID.2003.1199389
关键词: Host (network) 、 Web application 、 Computer network 、 The Internet 、 Computer science 、 Distributed computing 、 Internet topology 、 Overlay network 、 Internet hosting service 、 Hierarchical clustering 、 Cluster analysis
摘要: Being able to identify clusters of nearby hosts among Internet clients provides very useful information for a number internet and p2p applications. Examples such applications include web applications, request routing in peer-to-peer overlay network, distributed computing In this paper, we present formulate the host clustering problem. Leveraging previous work on distance measurement, propose two hierarchical techniques solve The first technique is marker based partitioning approach. second well known K-means algorithm. We evaluated these approaches simulation using representative topology generated with GT ITM generator over 1,000 hosts. Our results demonstrate that our algorithmic effectively arbitrary diameters. conclusion by leveraging estimation, it possible cluster benefit various requirements.