Quantitative Decision Tools for the Management and Analysis of the Risk from Terrorist Attacks

Abstract

A natural epidemic is a disease that suddenly affects many individuals in a short time period, spreading from person to person in a locality where thedisease is not usually prevalent. The sudden outbreak of an epidemic is usually modeled as a random variable because it cannot be anticipated. Epidemics introduced by bioterrorists are planned events by intelligent adversaries, who might also introduce other terrorists’ activities that dependon the responses of the defenders. Since these events are not random, models maybe helpful for anticipating terrorist attacks. Since defendingagainst such attacks does not fit into the classical modeling paradigmbecause there is a scarcity of data, the defender must respond quickly, the attacker can also adapt new strategies in response to the actions of thedefender, new modeling strategies are required to improve the strategies of the defender. In this article, a Stackelberg model combined with fault trees is proposed for determining sequential optimal defense strategies for thedefender by identifying minimal cut sets of events that would most likely lead to a successful terrorist attack. Further, if the model can be formulated as a sequence of Markovian state changes based on default trees, a dynamic programming problem with the Bellman equation reduces the solution from evaluating a complex model to evaluating a sequence of simple problems