Improving Efficiency and Coordination in Self-Organizing Emergent Systems

Diplomarbeit, September 2008

The Efficiency and Coordination Advisor (ECA) has the ability to improve efficiency and coordination in self-organizing emergent systems by analysing their runtime behaviour and deducing simple rules that allow the agents of the advised system to solve a problem that includes recurring tasks in a way that is very close to the optimum. The proposed approach retains the self-organizing property and the autonomous nature of the advised system as it is not working as a control mechanism and does not interfere with the actual execution of the system. An abstract definition of the mechanism is given as well as an instantiation of the concepts for instances of the Pickup and Delivery Problem solved with an implementation of Digital Semiochemical Coordination, an approach to design self-organizing emergent systems. An evaluation of different scenarios of the PDP shows substantial improvements for scenarios with known sub-optimal behaviour as well as for random scenarios.