WED NOON Poster - Military Applications

Full Report:

The Commander’s Model Integration & Simulation Toolkit (CMIST), by James Melhuish

The particular model presented was designed to assist air strategy planners in providing commanders with campaign plans prior to and during operations. This was accomplished using a unique combination of system dynamics, Agent-Based, and Bayesian Network models. Over a two year period, several models of varying scale and scope were developed using this construct to examine the reactive vs. proactive nature of commanders acting within an air and/or ground-based combat operation to contain insurgent growth and violence.

There are three basic modes in this model for a commander to counter-act insurgent growth: Reactive Attack, Proactive Attack, and Psyops (psychological operations) Campaign. These three modes correspond to negative loops that attempt to stabilize or dominate a positive loop of insurgent growth.  Different mechanisms trigger each. An observed threshold of violence triggers the first, Reactive Attack. In this case, the commander may be too late to adequately contain growth.  The second, Proactive Attack, gives the commander the ability to forecast the future state of the conflict by using embedded cause-effect (or belief) systems. This allows the commander to act sooner to more effectively contain escalating violence. The third, Psyops Campaign, may be combined with either of the other two to slow growth so that a swift attack is less critical.

The one question I had for James was why they had used a Bayesian Network to represent the Commander’s belief system when it might have also been modeled using a CLD. His answer was simple: the Air Force already had a model to represent this (OAT). I think that this feature clearly represented the flexibility of the CMIST environment to incorporate outside models and data to shorten the development cycle.

Lou Mauro, BAE Systems / M.S. Student at WPI