A Distributed Agent-based Approach to Stabilization of Global Resource Utilization

Distributed management of complex, distributed systems is the focus of this paper. Adaptation through local deliberation by software agents within a hierarchical virtual organization is the approach taken. Global stabilization of resource utilization is the goal. Electricity networks are used to illustrate the potential of two fitness functions on the basis of which local choices for resource utilization are made: minimizing oscillations is the first function considered, reversing oscillations the second. Results reveal considerable increase in the stabilization of resource utilization compared to a system that utilizes resources in a greedy manner.