Processing Multi-Attribute Queries in Peer-to-Peer Systems

To process multi-attribute queries in peer-to-peer data management systems flexibly and efficiently, a scheme based on a trie-structured overlay is proposed. Putting concretely, first a mapping algorithm is designed based on a virtual partition tree, which is then used to map the multi-attribute values to their binary keys of different length according to the data distribution in the multi-attribute space without destroying their original order. Afterward, the keys are indexed with a trie-structured overlay named P-Grid. Furthermore, the algorithms processing one- dimensional data in P-Grid are extended to the multi- attribute situation. Qualitative analysis and comparison show that this scheme is closer to the goal of multi-attribute query processing in the peer-to-peer systems than the related work.