Detector deployment under a small vessel attack threat to port security

The 9/11 events have highlighted awareness about threats presented by terrorism. Ports and waterways play a crucial role in the global trade and homeland security. Because of its strategic importance and characteristics, a port is easily exposed to terrorist attacks. Providing an effective and efficient port security system has become one of the primary tasks for homeland security. There are various types of terrorist attack threats in a port. In our research, we focus on scenarios where terrorists use small boats to attack maritime targets (e.g., military vessels, cruise ships, and oil tankers). We consider how to protect these targets by deploying detectors which can sense whether a passing small vessel carries radiological, chemical, biological, or other suspicious materials. Detectors are not perfectly reliable and the probability of detection depends on both the detector's characteristics and how long the small vessel would stay in the detector's effective detection area. At the first step, we study how to place multiple types of fixed-position detectors to protect maritime targets from such a small vessel attack. The probability of detection depends on the type of detectors. The resulting detector placement problem is formulated as a nonlinear binary integer program such that the expected damage cost caused by the small vessel attack is minimized on the condition that the cost of detectors is within a budget limit. Two exact algorithms (standard branch and bound and linearized branch and bound algorithms) and a greedy adding heuristic are proposed. We conduct a detailed computational study and find that the linearized branch and bound algorithm is preferred to the standard one and the greedy adding heuristic performs well. Besides, our study addresses a more challenging problem in which the strategic behavior of a government and a terrorist group is taken into account. More specifically, we study a defender-attacker Stackelberg game with two players: a government (a leader) and a terrorist group (a follower). The government first decides the detector deployment strategy to minimize the maximum expected damage while keeping the cost of detectors within a budget limit. The terrorist group, after obtaining the full information of detector locations via intelligence, decides the small boat attacking strategy (determines attack routes) to maximize the expected damage. A path-based model is established and the resulting problem is formulated as a min-max nonlinear integer program. Two exact cutting-plane algorithms and a heuristic are proposed and tested. Computational results show the plan suggested by our defender-attacker model for the port of New York and New Jersey and the performance of the proposed algorithms. In our above models there are three basic assumptions (detectors work independently, detectors are stationary, and the terrorist group obtains full information) which make the models simple but not reasonable for some situations. We will relax these assumptions one by one in future research.DOCTOR OF PHILOSOPHY (MAE