Georgia Institute of Technology
This project has two broad aims. The first is to develop improved risk analysis methods for quantifying the vulnerability of a food product supply chain that may be used by an adversary to deliver a chemical or biological agent. The second aim is to create new food supply chain design and control strategies that seek to maximize system productivity while mitigating risks posed by intentional attacks. Existing approaches for analyzing vulnerabilities to attack and the associated risks rely primarily on expert scoring of potential targets and attack scenarios. Most do not attempt to explicitly model the decision interactions between adversaries attempting to exploit a supply chain and the managers and regulators working to protect it. We assume that an adversary intends to use a given food supply chain as a weapons delivery platform by contaminating food product(s) with a chemical or biological agent. We propose, develop, and implement new risk modeling methods that explicitly model the potential actions of adversaries and corresponding defensive actions over time in order to compute quantitative risk measures, such as expected consequences. The risk models that we will build explicitly consider the time dynamics of this attacker-defender game over time and the fact that both the defender and attacker may not have accurate information about each other, representing decision strategies and outcomes using a multi-objective, partially observed Markov game as a model. We will implement algorithms to analyze these models, which tend to be challenging computationally, and deploy the approaches within software tools. We will demonstrate our methods using case studies of supply chains serving the U.S. market that involve liquid eggs and other foods or food systems of interest to stakeholders.