My research interests primararily lie in the following areas.
Energy Data Analytics
Smart meter installations have opened up unprecedented opportunities to collect energy consumption data. Likewise, plug-level meters find their ways into more and more homes, often as part of home automation solutions. Energy management systems can analyze the data collected from these devices and realize novel use cases, e.g., detecting unexpected consumption behavior or predicting future energy demand.
Smart Spaces and Intelligent Building Automation
Based on the collection and consolidation of contextual sensor data, a correlation between the environmental parameters and the corresponding situation can be established. Through learning previously encountered behavior, an intelligent building automation system can adapt the physical environment to the user’s preferences. This area is also tightly coupled with my work on energy data analytics.
Heterogeneous Wireless Sensor Networks
Wireless Sensor Networks are commonly assumed to be composed of identical hardware devices with fixed sets of sensors. In the envisioned Internet of Things, where every physical device will be connected, many heterogeneous devices will be present in the same network. Seamless communication between these heterogeneous devices can be seen as a prerequisite for cooperative behavior, which lays the groundwork for the intelligent automation of buildings.
Data Processing on Embedded Systems
Energy is a scarce resource on embedded sensing systems, which form the infrastructure of Wireless Sensor Networks. The wireless communication interface in particular is often the most energy-hungry component. Thus, local data processing to reduce wireless traffic, e.g., by means of data compression, is a viable means to extend a node’s lifetime and thus contributes to the overall operability of the entire network.
Sensor Network Testbeds
Analytical and simulative means are well-suited prerequisites for the analysis of algorithms designed for sensor networks. The unpredictable nature of real-world environments, however, necessitates the analysis of these algorithms in situ. Instead of deploying the nodes into a real-world setup without any means for debugging and control, testbeds serve the purpose of combining real-world characteristics with improved control capabilities.