Optimising self adaptive networks by evolving rule-based agents

Abstract

The need for networks that adapt autonomously to dynamic environments is apparent. In this paper we describe how self adaptive networks can be optimised by means of agents residing on the nodes of the network. The knowledge of these agents is a set of active rules. A genetic algorithm dynamically prioritises these rules in the face of dynamically evolving conditions. To our knowledge, this is the first time that GAs have been used for this purpose. We demonstrate the applicability of our method by presenting several experiments and results.