Foundations of Content-Based Networking

University Of Colorado At Boulder, Boulder CO

Investigators

Abstract

This research project focuses on foundations for a new kind of advanced communication infrastructure referred to as a content-based network. A content-based network is intended to complement traditional uni-cast and multicast networks, providing improved support for the communication modes underlying large-scale, loosely coupled, multi-party, distributed applications such as auctioning, information sharing, fusion and dissemination, sensor grids, personalized news distribution, service discovery, and multi-player games. The main research challenge in this project is to formulate a viable technological basis to support the service model of a content-based network. In particular, the work involves six basic technical thrust areas: Network Configuration, Subnetting, Routing, Forwarding, Quality of Service and Security. For Network Configuration, research involves design of an expressive, yet tractable, data model and associated predicate language for messages. Subnetting is expected to simplify network configuration and reconfiguration, as well as to reduce the computational cost of forwarding. Based on the chosen approach to addresses and subnets, an appropriate routing algorithm must be selected and adapted. Fundamental to the efficiency of content-based networking will be the availability of the compact forwarding tables and fast forwarding algorithms. The range of delivery guarantees for Quality of Service in a content-based network is conceivably rich. Finally, content-based networking introduces some challenging twists to traditional communication security, which have to be identified, thoroughly characterized and evaluated. The results in all these thrust areas are subject to rigorous evaluation. Two-pronged evaluation strategy is planned: simulation-based experimentation and prototype-based measurement. The simulations will provide a means to evaluate the robustness and scalability of the service under a variety of synthetic usage scenarios, while the prototype will allow the utility and comparative performance of the service to be demonstrated in implementing a variety of higher-level applications.