SENSOR OPTIMIZATION FOR RADIATION PORTAL MONITORING

Radiation portal monitoring is tasked to detect illicit nuclear material in a difficult measurement environment. The challenge is to operate in a varying background and rapidly screen cargo (that often includes naturally occurring radioactive material such as ceramics and cat litter) to detect threats (shielded or not). As always, no system can achieve a zero false fail and false pass rate, so decisions regarding acceptable risks are made. Usually such decisions are implicit, mostly because of the subjective nature of assigning costs to the false pass errors. This paper indicates how a cost-benefit analysis could be applied to candidate or existing sensor systems. Models of real data and decisions generated from staged sensors inform how decision rules or thresholds can be optimally set at each stage. The approach can be applied to either networked sensors or non-networked sensors; however, presently it is not known whether networked sensors that share complete data rather than nonnetworked sensors that share only “pass-fail” data will be cost effective.