Networked Sensors for Cargo Screening

Sensors to monitor cargo are currently operated in standalone mode because of cost considerations, operational simplicity, and because potential benefits to having sensors share information do not appear to have been seriously considered in many cases. To identify situations where threat detection benefits offset networking costs (and where they do not), we describe a numerical algorithm for approximating the multidimensional integrals defining the false positive rates and detection probabilities for networked systems. The algorithm is sufficiently fast computationally to embed in optimization routines that search over sensor space (e.g., how to best augment an existing sensor network), as well as over threshold space (i.e., how to best determine alarm thresholds for the sensors) for both standalone and networked sensors. Results are applied to detection of elevated gamma and neutron counts. Circumstances under which networking is useful are characterized.