Radioactive source detection by sensor networks

Detection limits of sensor networks for moving radioactive sources are characterized, using Bayesian methods in conjunction with computer simulation. These studies involve point sources moving at constant velocity, emulating vehicular conveyance on a straight road. For networks involving ten nodes, respective Bayesian methods are implementable in real time. We probe the increased computational requirements incurred by larger numbers of nodes and source trajectory parameters. The complexity appears quadratic in the number of nodes and, also, numerous trajectory parameters may be used. We investigate the consequences of different levels of background radiation. Simulations are shown to be useful for ranking candidate node layouts. We study the detection capabilities of individual sensors and the scalability of detection with sensor density; near the detection limit, increasing the number of sensors can accrue subproportional network sensitivity.