Content-Based Publish/Subscribe in P2P Using Distributed R-Trees
Intervenant(s) : Silvia Bianchi (Université de Neuchatel)
We propose a stabilizing overlay for P2P networks optimized to reduce the false positives in publish/subscribe systems. Our overlay is based on the extension of R-trees, well known indexing structures specially designed to support spatial database queries. We propose a distributed self-stabilizing implementation of an R-tree-like overlay that copes with nodes dynamicity (frequent joins and leaves) and memory and counter program corruptions. The maintenance of this structure is local and no additional memory cost is needed for guaranteeing its stabilization. Additionally, we propose an application of the designed structure to Complex Content-based filtering in Publish/Subscribe systems. These systems provide a useful platform for delivering data (events) from publishers to subscribers in a decoupled fashion in distributed networks. Publish/Subscribe systems have many applications, including web services, stock quotes, alerts monitoring, free riding monitoring, and Internet games. Developing efficient publish/subscribe schemes for complex subscriptions (subscriptions spanning multi-dimensional intervals) in dynamic distributed systems is challenging nowadays. The use of dynamic self-stabilizing R-trees in this context offers an efficient balanced structure and provides guaranties in terms of fault-tolerance, response time and storage space.