Cyberwar XXI Quantifying the Unquantifiable Adaptive AI for Next Generation Conflict Simulations

摘要

The era of the "Revolution in Military Affairs," "4th Generation Warfare" and "Asymmetric War" requires novel approaches to modeling warfare at the operational and strategic level of modern conflict. For example, "What if, in response to our planned actions, the adversary reacts in such-and-such a manner? What will our response be? What are the possible unintended consequences?" Next generation conflict simulation tools are required to help create and test novel courses of action (COA's) in support of real-world operations. Conflict simulations allow non-lethal and cost-effective exploration of the "what-if" of COA development. The challenge has been to develop an automated decision-support software tool which allows competing COA's to be compared in simulated dynamic environments. Principal Investigator Joseph Miranda's research is based on modeling an integrated military, economic, social, infrastructure and information (PMESII) environment. The main effort was to develop an adaptive AI engine which models agents operating within an operational-strategic conflict environment. This was implemented in Cyberwar XXI - a simulation which models COA selection in a PMESII environment. Within this framework, agents simulate decision-making processes and provide predictive capability of the potential behavior of Command Entities. The 2003 Iraq is the first scenario ready for V&V testing.