DIVIDE : adaptive context-aware query derivation for IoT data streams


In the Internet of Things, it is a challenging task to inte-grate & analyze high velocity sensor data with domain knowledge &context information in real-time. Semantic IoT platforms typically con-sist of stream processing components that use Semantic Web technologiesto run a set of fixed queries processing the IoT data streams. Configur-ing these queries is still a manual task. To deal with changes in contextinformation, which happen regularly in IoT domains, queries typicallyrequire reasoning on all sensor data in real-time to derive relevant sen-sors & events. This can be an issue in real-time, as expressive reasoningis required to deal with the complexity of many IoT domains. To solvethese issues, this paper presents DIVIDE. DIVIDE automatically derivesqueries for stream processing components in an adaptive, context-awareway. When the context changes, it derives through reasoning which sen-sors & observations to filter, given the context & a use case goal, withoutrequiring any more reasoning in real-time. This paper presents the detailsof DIVIDE, and performs evaluations on a healthcare example showinghow it can reduce real-time processing times, scale better when there aremore sensors & observations, and can run efficiently on low-end devices.

Joint Proceedings of the International Workshops on Sensors and Actuators on the Web, and Semantic Statistics co-located with 18th International Semantic Web Conference (ISWC 2019)