Complex radiation sensor network analysis with big data analytics

Radiation detection has long been of fundamental interest in a wide range of areas such as nuclear forensics and the environmental awareness of radioactive materials. For example, the Fukushima nuclear accident stimulated citizen scientists to collect and share radiation data across the world. However, it is non-trivial to estimate exact radiation levels using volunteered geographic information (VGI) data due to the spatial and temporal granularity of measurements as well as unprecedented levels of data volume. In addition, the accurate background measurements are unavailable in all areas. This research provides an alternative to understand radiation level changes using graph comparison. Previous work has used sensor networks to detect and track radiation. While this approach uses static sensor networks, mobile sensor networks have obvious benefit to track illicit radioactive materials. However, all previous approaches use the predefined structure of the sensor networks. They might also know or calculate the background radiation levels. The aim of this paper is to understand radiation level changes without having such details. We assume that the region of high dose rates in an environment like Fukushima continues over time, irrespective of the background. Thus the structural similarity of radiation levels based on radiation interpolation maps will reveal the changes of radiation levels.