A New Efficient Hierarchical Distributed P2P Clustering Algorithm

摘要

A new hierarchical distributed P2P architecture and clustering algorithm (HDP2PA) is proposed in this paper. HP2PA can address the problems of large central storage requirement in centralized data mining and heavy traffic in traditional distributed P2P network. The architecture is a multi-layer overlay network of peer units based on Super-P2P model. Clustering process is divided into two phases consisted of local clustering and global combination. Using the hierarchical architecture, the partial results produced by local clustering are aggregated to a global one. Experiments prove the method is effective.