Self-Adaptive Systems and Networks

Funding and Staffing Details

This project was funded by a PhD studentship from BT in 1997. Alex Poulovassilis was the Principal Investigator and Vagelis Nonas the PhD student funded by the project.

Project Outline and Aims

Intelligent networks automatically adapt to changes in their topology and load conditions without the need for human intervention. The aim is for the network to continue to operate at near-optimum levels. The network needs to be autonomous and to have distributed control - there is no global source of knowledge but rather knowledge is distributed accross the network.

We are using techniques from multi-agent systems , active databases and genetic algorithms to support intelligent networks. We assume that there is an agent residing on each node of the network. The knowledge of each agent is expressed as set of active rules . We have proposed a method for optimising the rule-base of each agent in the face of dynamically evolving environments using a Genetic Algorithm. Initial experiments show that this approach is well-suited to network failures, varying load conditions, and network restoration.


Optimising Self-Adaptive Networks by Evolving Rule Agents. E.Nonas and A.Poulovassilis Proc. Evolutionary Image Analysis, Signal Processing and Telecommunications (EvoIASP'99 and EuroEcTel'99), Goteborg, May 1999. Springer-Verlag LNCS 1596, pp 203-214.

Optimisation of Active Rule Agents using a Genetic Algorithm approach. E.Nonas and A.Poulovassilis. Proc. DEXA '98, Vienna, August 1998. LNCS 1460, pp 332-341.

See also A.Poulovassilis' web pages on Database Languages.