Abstract (English)
Advancements and wide availability of cheap sensors are fuelling the development of Internet of Things (IoT) applications. More sensors mean more data, and more relationships between data points need to be tracked in order to effectively understand, manage and control systems. Having access to either real-time presence data or patterns extracted from historical data is particularly valuable when dealing with facility design and management due to their direct correlation with energy consumption and indoor comfort. We propose a straightforward, cost effective and privacy-preserving method to extract the occupancy information. By aggregating semantic knowledge, motion sensor data and data from dwelling entrance doors, a robust virtual occupancy sensor has been developed; it is underpinned by an ontology that was developed on top of the set of standard ontologies like Building topology ontology (BOT) and Smart appliance reference ontology (Saref) that allowed describing all relevant datapoint and demo site metadata and enabled automated processing of collected data. The method is replicable to all built environment described in a similar way where motion information is collected and where there are clear boundaries of monitored space, and the occupancy information can be useful in different application cases. While predictive occupancy models or expensive sensing alternatives have been already exploited for similar purposes, our solution is simple, inexpensive, replicable and easy to implement in existing buildings.