PNNL

Abstract

Instrumented maritime cargo containers based on small, rugged radiation sensors could enable the detection of illicit nuclear and radiological materials during transit, so that identification of suspect containers occurs well away from seaport facilities. Among the possible decision-making algorithms for the radiation sensor data is the Sequential Probability Ratio Test (SPRT). We describe the suitability of the SPRT for this long-term monitoring application and provide an overview of asymptotic results for the SPRT on continuous-time processes studied in this work. Comparisons of decision-making performance, measured in time-to-detection, are presented for the range of signal to background ratios likely to be encountered in the monitoring of gamma-ray signatures from containerized cargo. The analysis demonstrates that for these simulated long-term monitoring scenarios, alarm decisions are made significantly faster using the SPRT methodology, when compared to a commonly used algorithm based on Poisson probability distribution function integration. Subject Areas: Nuclear Measurements and Monitoring Techniques, and Sensor Networks and Homeland Security