A comparison of geostatistcal interpolation methods for the characterization of background radiation collected on a mobile sensor network in a cloud computing environment

There is a need to identify illicit radioactive sources in urban environments to prevent potential national security crises. These illicit materials can be the precursors to dirty bombs as well as broader nuclear proliferation. Mobile Sensor Networks deployed in major metropolitan areas have been proposed to address this security and proliferation challenge. However, these networks generate large amounts of data that cannot be processed on stand-alone computers. This volume of data requires the application of novel Big Data Analytics tools and techniques. Additionally, the presence of permanent human-made background radiation sources like buildings and monuments and natural fluctuations in background radiation can make detecting illicit radioactive sources with methods relying on simple radiation thresholds challenging. If the background radiation is well characterized, then more precise thresholds can be implemented to lower the rate of false alarms. The background radiation data is often sparse, and interpolation is needed to arrive at well characterized data. The performance of two methods of geospatial interpolation, Inverse Distance Weighting and Kriging, on background radiation in a cloud computing environment are evaluated in this thesis, and recommendations on data collection and interpolation methods for ideal background characterization are provided.U of I OnlyAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD syste