Dr. Hill is exploring the extension of the combat “intangibles” that can be represented with agent-based modeling to the analysis of combat “tangibles” in traditional models like Brawler (an Air Force air-to-air simulation).

He is also exploring a classic search scenario represented planes flying out of England in 1942 - 1943 to search for submarines in The Bay of Biscay. Part of his effort is to get agent behavior in an agent-based model to comport with the historical record. Though his data “looked” like it corresponded, he also wanted to prove that it statistically matched.

The simulation represented a two-person game of day versus night searching. The U-boats didn’t want to surface during the day; however, if they only surfaced at night the Allies would concentrate only on nighttime searches by air. Both sides had imperfect information, so neither side knew the other’s strategy. The Allies used two pure search strategies – day and night searches – with a goal, of course, of maximizing detections. The Germans used two pure surfacing strategies – again, day and night – with the goal to minimize detection. In his study of this two-sided game, Hill showed that there existed an equilibrium point between both strategies.

He then allowed both players to deviate from the equilibrium during a 6-month period, based on their guesses about the adversary’s strategy each month. He looked at the convergence of different design points and used a hill-climbing model to apply “what-ifs” to determine a range of outcomes.

In a related simulation optimization study, Hill’s team examined the process of finding the best-input variable from all possibilities without explicitly examining each possibility. WPSU Graduate student Subhashini Ganapathy worked on the simulation optimization study and focused on bringing the human into the algorithmic loop of the simulation, a constrained optimization problem. The process of finding the best input variable from all possibilities without explicitly examining each possibility often involves the use of a search heuristic in the simulation.

Future applications of this work include deep water surface searches and air searches in rugged or urban environments. Also involved are Unmanned Air Vehicles (UAVs) in search, rescue and reconnaissance, and missions using Unmanned Combat Air Vehicles (UCAVs). This research could be expanded into game theory and the focus expanded more in search theory. Future plans also include work on agent optimization and real-time planning.