Towards distributed reasoning for behavioral optimization

Abstract

We propose an architecture which supports the behavioral self-optimization of complex systems. In this architecture we bring together specification-based reasoning and the framework of ant colony optimization (ACO). By this we provide a foundation for distributed reasoning about different properties of the solution space represented by different viewpoint specifications. As a side-effect of reasoning we propagate the information about promising areas in the solution space to the current state. Consequently the system’s decisions can be improved by considering the long term values of certain behavioral trajectories (given a certain situational horizon). We consider this feature to be a contribution to autonomic computing