RBMA: Replication Based on Multilevel-Agent for P2P Systems

In peer-to-peer (P2P) systems, one important technique to improve the effectiveness of blind search is to proactively replicate file, which can lighten peers´ load as well. But for autonomy and selfishness of peers in P2P systems, some peers are not willing to contribute their storage to save replications, which cannot be created successfully in this case. This paper presents a replication method based on multilevel-agent, namely RBMA, recurring to information economics about principal-agent relationship and incentive concept. One is that using incentive can encourage peers to provide their storage and strengthen cooperation between peers. The other is that using multilevel-agent can make the best of idle storage resources and balance load better than single agent. The experimental results demonstrate that the proposed replication method can weaken negative effect brought by noncooperation between peers as far as possible to create replications successfully, and indeed improve search performance to a certain extent. Compared with other replication methods, this new method can reduce search response time, promote search success rate, and balance peers´ load.