An ant-based approach to cluster peers in P2P database systems

摘要

Peer-to-Peer (P2P) systems with distinct advantages such as scalability, robustness and low cost have developed quickly in recent years. The resource search problem, that is to find out a peer having a desired resource, is a fundamental problem to be initially solved in most P2P applications. P2P systems can also be categorized under database systems. One of the most important problems of P2P database systems is the query processing algorithms. Even a rather small improvement over these algorithms can have a significant impact on the performance of a P2P database system. One of the most effective methods to reduce query processing cost is peer clustering. Hence, clustering peers with similar contents makes the queries directed to the appropriate cluster and thus to be efficiently answered. In this paper, we present a new peer clustering algorithm for P2P database systems that uses ant-based clustering algorithms. We have adjusted ant-based clustering algorithm to peer clustering problem. Performance evaluation of the algorithm shows its strength in clustering peers in P2P database systems to be of a higher quality and accuracy compared to the existing systems, to have a relatively short running time and to quickly converge to an optimum. Moreover, our algorithm achieves a high scalability to the increasing problem size.