SensorFeed: An architecture for model-based sensor network data enrichment

Perceiving the environment, a sensor network collects huge volume of data to be used in different application domains. The gathered data is often used and analyzed by e-research scientists other than the original investigator. Therefore, the sensed data needs to be captured, processed and stored in a form that will allow someone to use the data with confidence long after the original investigators have left the scene. To address this need, in this paper we present SensorFeed, which integrates metadata repositories and sensor data management systems. Using SensorFeed, scientists can annotate sensor readings automatically as they are streamed, through direct use of statistical modeling frameworks. These annotations enrich sensor readings, thus making datasets generated from sensor network deployments usable by external scientists. A real-world use case of SensorFeed for agriculture engineering is presented to show the applicability of our approach to an e-research application.