An approach for decisionmaking and control in geographically distributed systems

Many large, complex systems can be represented by multiple agents with loose coupling. Examples include air traffic control, military command and control, and manufacturing applications. The level of autonomy granted to agents in such systems has a substantial impact on problem solving quality and speed. This paper proposes a distributed agent-based approach that supports two levels of autonomy, local and command-driven. The architecture includes a simple agent model with the ability to conduct limited reasoning about its level of autonomy with level switching when the situation warrants. Application of the approach to naval radar frequency assignment and management is discussed