Project Leader
George Roussos
g.roussos@birkbeck.ac.uk

Alex Poulovassilis
Niki Trigoni
Michael Zoumboulakis

Project Details
Partially funded by CTES
Start: January 2004.
Duration: 36 months.

Project Web Site http://www.dcs.bbk.ac.uk/~gr

Keywords
Ubiquitous computing, sensor networks, reactive behaviour

Programming Sensor Networks
Application development for sensor and actuator networks presents unique challenges since it has to address the complexities of distributed and often decentralized operation, the highly resource constrained nature of network nodes, and the highly transient nature of network topology. Moreover, applications must operate unattended for prolonged periods of time and still maintain their integrity and quality of service.

Event-Condition-Action Rules
Event-condition-action rules (also referred to as gactive rulesh) offer a good fit for sensor network applications that require reactive behaviour e.g. security monitoring and rapid forest fire response. In such applications, the system needs to provide a timely response to certain events. Although in principle this would be possible using a sensor network query processor, its deployment would unnecessarily consume the limited resources of the network nodes by periodically checking for events that may not have occurred.

In our work we are exploring Event-Condition-Action rules as a computational model for reactive systems, with particular reference to sensor and actuator networks. In particular, we are developing the Asene system for Active SEnsor NEtworks.

Asene Architecture
Asene addresses the novel requirements of sensor and actuator networks by implementing event channels, which are also viewed as data object primitives. Asene adapts to changing communication topologies, routing in the presence of network segmentation, and incorporates mechanisms for in-network data reduction and storage.

Crossbow Mica 2 Smart Dust nodes used for habitat monitoring.