Date of Award
8-2026
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Industrial Engineering
Committee Chair/Advisor
Dr. Thomas C. Sharkey
Committee Member
Dr. Bryan Lee Miller
Committee Member
Dr. Yongjia Song
Committee Member
Dr. Emily L. Tucker
Abstract
In this dissertation, we study novel intervention strategies and network adaptation behaviors motivated by illegal drug trafficking networks. First, we start with a new class of network interdiction problems (NIPs) considering network responses following the interdiction. This new model focuses on problems where interdictions may not be successful and where network restructuring after successful interdictions is uncertain. We propose a two-stage stochastic model (2SSM) for this problem where the attacker makes the interdiction plan in the first stage, the network stochastically restructures based on the realization of successful interdictions, and then the defender solves a maximum flow problem in the resulting network. We formulate the 2SSM problem as a bilevel mixed integer problem (MIP) and then we convert it to a single-level MIP by using restructuring properties and duality concepts. To solve this single-level MIP, we apply the sample average approximation (SAA) approach. We propose relaxation and heuristic methods that find a valid lower and upper bound. The heuristic method can solve problems that have at least four times the number of scenarios as the original SAA approach. We then demonstrate the importance of our problem by examining its applications in city-level drug trafficking networks.
After that, we further expand our idea on network response by considering its temporal aspect. We develop another unique class of NIPs where the attacker seeks to maximize the recovery time required for the defender to reestablish a certain flow level, while the defender seeks to minimize this time. In particular, the defender will need to make network restructuring decisions in the interdicted network to reestablish the flow. We demonstrate how to solve this class of problems through a binary search method that requires solving, or approximately solving, a single-period network interdiction problem with restructuring at each step of the search. For certain realistic assumptions on the restructuring decisions, we use reformulation techniques to convert this problem into a single-level reformulation and computationally demonstrate the superiority of our proposed binary search method compared to solving this single-level reformulation. We apply our binary search method to two practically important problems related to disrupting illegal drug trafficking networks, including how to significantly improve the scale of the problems solved through new types of cuts in a column and constraint generation algorithm that solves an interdiction problem with restructuring.
Lastly, we address a key limitation in existing research on applying network analysis approaches to drug trafficking networks. In particular, there is a lack of data on modern drug trafficking networks, such as those that move fentanyl, and there is a limited understanding of network adaptation following an interdiction. We address these challenges by studying local level fentanyl trafficking networks, generated from a network generator based on law enforcement interviews conducted throughout South Carolina in 2024–2025. With these networks, we aim to better understand the properties of fentanyl trafficking in South Carolina and the dynamic behavior of these local networks following an interdiction. We begin by applying the max flow model to determine the maximum flow in these networks and then use min cut analysis to identify their bottlenecks. In most cases, the user level emerges as the bottleneck of the network, and we compute the minimum percent reduction necessary at the wholesaler level to impact on the local-level trafficking networks. We then shift our focus on understanding how local fentanyl networks adapt after disruptions. We augment the network generator data with additional restructuring data that captures how individuals in the network respond following an interdiction, and we validate these adaptations with qualitative researchers. Our restructuring data helps to understand how resilient local fentanyl trafficking networks are to disruptions to their dealers.
Recommended Citation
Janan, Farhatul, "Network Interdiction with Restructuring: Novel Models to Disrupt Drug Trafficking" (2026). All Dissertations. 4367.
https://open.clemson.edu/all_dissertations/4367
Author ORCID Identifier
0000-0001-6342-3711