Year
2006
Abstract
Cost effective decisions to allocate resources to prevent successful acts of terrorism require an evaluation of the process by which the adversary selects targets and attack scenarios. To date, a Design Basis Threat (DBT) has been used to drive the design of physical protection systems for high-consequence targets, such as nuclear weapons and nuclear materials facilities. The DBT has been assumed to occur. As the DBT increases, the cost of physical security increases. A more comprehensive evaluation should consider the impact of real-world factors such as increased intelligence efforts to detect an attack in the formulation stage, and the fact that the adversary has a choice as to which targets to attack and performs an evaluation of the potential consequence versus the resources required for success in deciding what to attack and how to carry out the attack. An adversary scenario is defined to include the target, the attack plan, and the resources used by the adversary. A terrorist attack is not a random event, but to us as a defender there is significant uncertainty as to what scenario the adversary will chose. A traditional probabilistic approach is not appropriate for evaluating attack scenarios; probability is best suited for problems where the uncertainty is aleatory (random) such as earthquakes. For problems dominated by epistemic (state of knowledge) uncertainty, such as evaluation of adversary scenarios, a methodology is required that appropriately considers that type of uncertainty. The belief measure from the Dempster/Shafer Theory of Evidence is a measure of likelihood that was developed to deal with epistemic as well as aleatory uncertainty; probability is a special case of belief. The adversary process for selecting scenario(s) is complicated, but employs simple measures for selection. Since the adversary has a choice, unless all factors involved in the decision are “good” from the perspective of the adversary, the adversary will discard the scenario and develop others for consideration. An appropriate model for the adversary scenario selection process is approximate reasoning using fuzzy sets for linguistic factors. Approximate reasoning provides a rule base to combine the factors based on how the adversary decides to select scenarios. Approximate reasoning with fuzzy sets is the model for the scenario selection. To evaluate the model with the significant epistemic uncertainty that is present, degrees of evidence are applied to the inputs and belief/plausibility intervals are propagated up the rule base. We are implementing this technique in software and applying it to problems involving the selection of scenarios.