Querying sensor data for environmental monitoring

Abstract

Data sources are increasing in number and distribution on the territory: sensor networks and individual devices are not any more atomic devices, but they are transforming data providers and consumers, paving the way to new personalised services. In this context, the paper presents a new software infrastructure for the integration of heterogeneous data (web services and sensors' API for streaming data, databases and files for historical data) and any type of data analysis and inference algorithms (e.g. forecast, statistical analysis, data cleaning, anomaly detection and optimisation). This infrastructure, named SUSHI (Supporting Unified access for Streaming and Historical data), provides explicit representation of domain objects, queries over heterogeneous data sources, online access to different sources, inclusion of processing components such as forecasting and simulation modules. The paper describes the SUSHI concept and architecture, a case study in environmental management and possible future developments for new services.