User Location Modeling Based on Heterogeneous Data Sources

Over the past decade, interest in home automation systems constantly grew. Especially in the daily life with the connection of intelligent everyday devices through the Internet of Things. To allow automatic actions on these devices, user localization systems have become a major input modality for smart home systems. The location of a user (or rather a subject) can be determined by different localization techniques, such as sensitive floor systems, discrete activity sensors like light switches or RSSI-based WLAN/Bluetooth beacons (e.g. smartphones). These heterogeneous data sources provide various means of user location certainty, the ability to identify a user or the ability to recognize multiple subjects in the same location. In order to achieve a higher grade of accuracy, multiple data sources can be combined by location fusioning algorithms. However, to allow the integration of such algorithms on hardware independent basis, a common user location model is needed, which can represent all important aspects of these localization techniques. Furthermore, the overall system needs to be easily extendable with new methods of localization. This Bachelor thesis will investigate the concepts of existing user localization systems as well as develop a new model to represent the location of subjects based on already existing location models. A reference implementation will be provided based on Eclipse SmartHome, an open source building automation framework. The implementation integrates multiple data sources within the Living Lab of Fraunhofer IGD, in particular a capacitive localization system, intelligent light switches and WLAN beacons. Furthermore, a simple location fusioning algorithm will be presented based on the implemented user location model as a proof of concept