Mobile radiation sensor networks for source detection in a fluctuating background using geo-tagged count rate data

Mobile radiation sensor networks integrated with geographic information system provide an attractive option for the real-time anomalous radiation source detection. In order to obtain an accurate alarm of the presence of an anomalous radiation source, continuous measurements of the temporal and positional background radiation distribution are needed. The fluctuations of background radiation can be caused by several reasons, such as the variation in soil composition, building materials, and weather patterns. In this thesis, a radiation sensor network is deployed, and a maximum likelihood estimation-based algorithm is developed to evaluate measurements from the sensor network and estimate the experimental area's radiation distribution and fluctuation. Using the reconstructed background radiation distribution and fluctuation, the probability that each individual measurement includes an anomalous source is calculated. This thesis presents the work of using statistical inference to adjudicate gamma-ray count-rate data from a sensor network based on measurements of sources in an urban environment. Results show that the maximum likelihood estimation-based algorithm enhances the sensor network's anomaly detection accuracy over traditional approaches where the background radiation is measured only periodically and considered static in geoposition across large geographic regions