I delivered this talk at the EURO Conference in Bonn, July 2009.

We place nuclear detectors on a transportation network so that we minimize the probability of a smuggling event succeeding. In a smuggling event, a nuclear smuggler maximizes the probability that he smuggles material from a nuclear facility, through the transportation network, to a destination. We model the smuggler's movements via a Markov Decision Process (MDP), where the smuggler is probabilistically operating the direction in which he moves. The optimal detector placement decision can be viewed as hedging against the worst of a set of probability distributions, where the smuggler selects the worst probability distribution on paths out of those allowed by the MDP. Another way to view the detector placement decision is as an adversarial MDP design, where we attempt to maximally disrupt the MDP operated by the smuggler.

See the related papers on Nuclear Smuggler Detection and .

- Models of Smuggler Movement (89%)
- Interdiction Models (56%)
- MDP Design (56%)
- Nuclear Smuggler Detection (56%)
- Vector-Borne Disease Control (44%)

## Add A Comment