Efficient Content-Based Publish/Subscribe Systems over Peer-to-Peer Networks

Abstract

Content-based publish/subscribe provides a convenient platform for users to specify their interested events as subscribers and ensure publications to be efficiently delivered to those users who are interested in the contents of them. Significantly, the current research trend is towards involving peer-to-peer network technology into this kind of system for scalability and fault tolerance. The dominant factor of system performance is to hash subscriptions and events into peer nodes for having time and space efficiency. In this paper, we propose two approaches with and without user clustering, respectively. Moreover, we introduce data replication and partition techniques to further distribute system workload. Our proposed system is more practical and flexible than existing ones. The performance is evaluated through a discrete event based simulator.