Screening Strategies in the Face of Strategic Applicants with Network Queuing

In security check systems, tighter screening processes increase the security level, but also cause more congestion for normal applicants, possibly delaying their application. This may in turn decrease the approver's payoffs. Adapting to the screening policy, adversary and normal applicants choose whether to apply to the system. Security managers could use a screening policy to deter adversary applicants but may also lose the benefits from admitting normal applications, therefore generating a tradeoff. The objective of this research is to advance the study of security screening strategies in series and parallel queuing networks and provide security screening insights to decision makers such as authority, manager, and screener. First, we analyze the optimal screening policies in an imperfect and multi-stage screening system, balancing security and congestion in the face of strategic normal and adversary applicants. We use game theory to rst study the best application strategies for both normal and adversary applicants, and then provide analytical solutions for the optimal screening strategies for the approver. We compare between the two-stage and one-stage screening systems and between the discriminatory and non-discriminatory screening policies and determine the conditions under which system/policy is better than the other. We also study the effects of the adversary applicants' risk preferences, the normal applicants' abandonment behavior, partially non-strategic normal applicants, and an N-stage imperfect screening system. Second, we study models on parallel queuing systems, where the approved participants have chances to be assigned to a faster screening channel with imperfect screening processes. The number of carry-on items is considered as an endogenous decision variable, which could affect the screening error probability, service rate, and the normal applicants' convenience payoffs. A multi-port selection model is also studied. Finally, we use Arena to simulate series and parallel imperfect screening queuing networks under discriminatory screening policies to validate and verify the results from the theoretical models. We compare the approver's payoffs and strategies between the theoretically optimal screening policies with the simulated results. We also consider some general probability distributions for service and arrival processes. A queuing network simulation based on visa Mantis process is built and analyzed.