Overview

Wireless sensor networks are more and more seen as a solution to large-scale tracking and monitoring applications. The deployment and management of these networks, however, is handled by a central controlling entity and the sensor network is often dedicated to a single application. We argue that this is due to the fact that we do not yet have the means to deal with a secure multi-purpose federated sensor network, running different applications in parallel and able to reconfigure dynamically to run others.

The communication paradigms which have been usually devised for small and single owner sensor networks simply do not have the right scalability, security, reconfigurability characteristics required for this environment.

The Federated Secure Sensor Network Laboratory (FRESNEL) aims to build a large scale federated sensor network framework with multiple applications sharing the same resources. We want to guarantee a reliable intra-application communication as well as a scalable and distributed management infrastructure. Orthogonally, privacy and application security should also be maintained.

We evaluate our proposal though a large scale federation of sensor networks over the Cambridge campus. The sensors monitor different aspects (temperature, pollution, movement, etc) and the network will be running various applications belonging to different authorities in the city.

Research Objectives

The overall aim of the project is to provide a federated sensor network platform that could be used by multiple applications in a seamless and secure manner. To achieve this aim, we have set the following technical objectives:

• Develop a language for service level description and agreement for this kind of sensor network sharing. The language will be the basis of our resource and application management and monitoring framework.
• Develop techniques to enable dynamic resource allocation in a decentralized fashion. These techniques will be designed to take into account application needs and network resources. They will be adaptive to varying application demands, scalable in terms of network resources, and robust to resource failures.
• Identify mechnanisms for dynamic partitioning of the sensor network into application-specific virtual sensor networks. These mechanisms should protect nodes belonging to different partitions (i.e. applications) against each other, especially during the transitive phase of dynamic repartitioning. Protection is in the form of privacy preservation and observation of resource allocation boundaries.
• Devise distributed and scalable communication protocols that will allow the components of a single sensor application (that may span multiple physical networks with different owners) to communicate with one another in a reliable and efficient way. The underlying collection of nodes, which belongs to the same virtual sensor network, will be able to configure itself dynamically to include new resources and heal from network failures.
• Develop distributed algorithms for processing multiple queries with very different quality-of-service requirements for accuracy and delay. In the presence of high query loads, these algorithms should gracefully degrade the quality of query answers giving priority to addressing the needs of critical applications. Privacy policies may also require that the specificity of certain query answers be limited.
• Deploy a prototype federated sensor network across the campus of the University of Cambridge, also partly integrating existing sensor networks. This will be a limited version of the envisioned CityNet application. It will span different colleges and will cover heterogeneious monitoring needs ranging from room usage and college security (using cameras), to air quality (using bike-mounted bio-sensors) and traffic monitoring (using vehicle-generated GPS trajectories forwarded to fixed nodes installed in colleges).

Partners

This project is a collaboration between the Computer Laboratory (University of Cambridge), the Oxford Internet Institute (OII), and the Oxford University Computing Laboratory (ComLab).

Project website