Simulation-based dynamic optimization: planning United States coast guard law enforcement patrols

A primary mission for the United States Coast Guard (USCG) operating in coastal United States waters is to interdict contraband. The USCG schedules a fleet of cutters to meet this mission and seeks a way to determine the operational efficiency of a particular schedule. This paper develops a methodology based on generating a sequence of finite horizon dynamic programs (DPs), where each DP differs only in the way the smuggling vessels and the cutters interact. The DP takes the point of view of the smuggler who wishes to develop the smuggling strategy which maximizes some characteristic (e.g., the mean) of the profit attained. The DP explicitly accounts for a smuggler who must combine his short-run profit goals with his need to gain future information about the configuration of the cutters. We develop a Monte Carlo sampling procedure to generate estimates of the random variables used in the DP.