Plan validation using DES and agent-based simulation

Abstract

Military plan validation is typically a long-drawn process requiring planners to validate their plans using anticipated scenarios or through military exercises. While military exercises provide realistic simulation of the plan, it is often the most expensive way of validating a plan. On the other hand, although using anticipated scenarios is relatively cheaper, the robustness of the validated plans is dependent on the extensiveness of the scenarios that they are validated against. This thesis explores the possibility of using a multi-agent system (MAS) to generate the aggressor's air strike plans, which could be coupled with a low resolution Discrete Event Simulation (DES) based air defense simulator to augment human planners in their plan validation. A MAS-based strike plan generator was built based on the tactics described in air strike doctrines. A DES-based air defense simulator was also built to provide an agent environment by modeling the behavior of air defense assets and their interactions with the aggressor's fighters. The resulting system demonstrates the ability to validate air defense plans using MAS generated strike plans and a low resolution DES-based simulator. It also provides a platform to assist air defense planners in foreseeing the action, reaction and counteraction dynamics of their air defense plan against a range of possible strike plans.