Clustering mechanism for content-based system in DHT-based overlay networks

Abstract: A generic pub/sub communication system (often referred to in the literature as Event Service or Notification Service) is composed of a set of nodes distributed over a communication network. Clients to the systems are divided according to their role into publishers, which act as producers of information, and subscribers, which act as consumers of information. Clients are not required to communicate directly among themselves but they are rather decoupled: the interaction takes place through the nodes of the pub/sub system. There are several architectural options and subscription models for publish-subscribe communication. We focus on content-based publish subscribe, and we explore an architectural options for realizing the broker overlay, namely peer-to-peer structured. We propose a novel adaptive content-based subscription management system, relying on a Distributed Hash Table routing infrastructure. We define a model for the event space guaranteeing the expressiveness for any application domain. Also we provide mechanism to dynamically identify groups of users with similar preferences (multicast group) based on clustering algorithms for the users preferences