Agent-Directed Simulation (ADS'05) 

Part of the 2005 Spring Simulation Multiconference (SpringSim'05)

 

LEARNING IN A COMPLEX ADAPTIVE SYSTEM FOR  

ISR RESOURCE MANAGEMENT

 

 

by

 

Steve Hall

Lockheed Martin


Abstract

The US DoD is committed to improving the effectiveness of the collective set of ISR assets in generating mission relevant information. Their vision is of a self-synchronizing, horizontally integrated “swarm” of sensors, processing elements, and communication assets that autonomously organize to meet the dynamically evolving ISR needs of future missions/campaigns. This talk describes an analysis workstation that is being constructed to support the evaluation of alternative designs for such a system. Within the workstation a complex adaptive system models the dynamic formation of collection teams while evolutionary algorithms, operating at the agent level, attempt to optimize the global performance of the ISR family of systems by modifying the various agent’s responsiveness to attributes of the “support requests.” Joint MEASURE, a DEVS based Monte Carlo mission effectiveness simulator developed by Lockheed Martin, provides the infrastructure for this workstation enabling the unpredictability of scenario evolution to be an integral element of the optimized solution. An overview of the architecture is provided along with sample results and lessons learned.