Pub-Eye: The delay constrained Pub/Sub for large scale wireless video surveillance

摘要

This paper proposes a novel design of Publish/Subscribe (Pub/Sub) based system for event-driven video surveillance over wireless networks. Different from the previous Pub/Sub works, the proposed system considers the real-time requirement together with the system scalability. We study the problem of selecting an optimal subset of brokers for subscription forwarding in a large scale delay constrained Pub/Sub system on top of a wireless mesh/ad-hoc network. We prove that the problem can be reduced from a minimum connected K-dominating set problem, which has been proved to be NP-complete. A heuristic algorithm is then proposed to solve the problem and its performance is evaluated by comprehensive simulation. We also share our experience of real-world implementation of the Pub/Sub based video surveillance system.