Implementation of a multi-agent simulation for the NPSNET-V virtual environment research project

Abstract

Traditional networked military simulation systems are technologically frozen the moment they are completed, thus limiting the participants that can interact in the simulation. When training for urgent missions in emerging conflict areas, assimilating new models, threat behaviors, and new terrain environments into the simulators requires lengthy integration, is prohibitively costly, and is non-distributable electronically at runtime. Threat behaviors are pre-scripted, lack organization, and do not accurately portray doctrine or rules-of-engagement. NPSNET-V is a novel architecture for networked simulations that supports scalable virtual worlds with built-in dynamic entity loading. These advances address each of the above concerns: scalability, entity and environment distribution, and dynamic technology loading. By combining this architecture with a system for creating autonomous, adaptable agents, threat forces canbe accurately simulated. This architecture is useful for proposing designs for strategies, tactics, or force packages during the conduct of experiments. The result of this thesis is a proof-of-concept application demonstrating the utility of these architectural advances. In this application, numerous autonomous agents form complex, dynamic, and adaptable interactions with resident and remote heterogeneous entities. These results define the course for future military models and simulations.