Controlling evolutionary network dynamics

Description

Advances in control technology and communication technology have enabled the evolution of multi-agent networks into large-scale systems that could complete cooperative tasks which are more complicated than ever before. However, the requirements of efficiency, robustness and adaptability for multi-agent networks, when they are coordinated to execute these tasks, have increased dramatically in industrial and social fields in recent years. In this context, traditional control methodologies present some serious limitations. For example, many unexpected factors like information loss, time delays and perturbations can easily cause decision-making failure for strategic interactions in multi-agent systems, especially in some complex and hash environments.

This proposed research aims at proposing new distributed control methodologies and algorithms to enable complex decision-making multi-agent systems to execute complex cooperative tasks rapidly, accurately and robustly. Simple local information passing, storage and processing are critical to accomplish this goal for large-scale multi-agent networks. Using the tool of evolutionary game theory, smart information exchange architectures within multiagent systems are proposed, which are more likely to lead to the emergence of novel cooperative control strategies that could cope with difficult situations.