Airborne Plume Tracking With Sensor Networks

This paper presents a framework and demonstrates results from a process detection based approach to tracking an airborne plume in sensor networks. Data integration and pattern detection in large sensor networks measuring gas and radiation plumes suffer from low resolution observations, missed detections, and numerous false positive reports. Large numbers of nodes and the hypothesis management concept of a Process Query System (PQS) can compensate for lower data quality. A result of the process detection based approach to this problem is models that can be implemented in many different scenarios. Plume predictor models are illustrated which allow data association between sensor nodes in typical outdoor wind conditions. We demonstrate a simulation of a mobile plume source in a sensor network designed for use in the same PQS. A kinematic model is developed for a vehicle carrying a plume source. Inverse models for this mobile plume source will work in conjunction with the existing software systems, thus allowing PQS to rapidly be adapted to a new problem domain with minimal modifications. This scenario of a mobile airborne plume source approximates a moving container emitting a detectable substance in a transportation network, where the container movement is restricted by existing vehicle corridors.