Data Management and Coordination in Sensor Networks
Many sensor networks often follow a centralized approach, with an emphasis on simple data collection and routing. As the amount of sensor information increases and the networks expand, scalability and data throughput are a weak point, both in terms of network size and platform performance.
„... through machine learning, models of predictive climate control (MPC) are automatically adapted to changing conditions of use in buildings ...“
The Center for Distributed Systems and Sensor Networks follows a different approach, in which the data is processed in a distributed and cooperative manner. In this novel strategy, data is already pre-processed on low-power and energy-efficient nodes and thereby reduced, and then fed to local decision making in sensor clusters. Within such clusters, decisions are made and additional nodes are only contacted if the particular problem cannot be solved within the cluster. This not only allows for better scalability of systems, but also significantly simplifies operation, installation and configuration.
The application fields are in particular energy efficiency and autonomous operation for building and production automation systems as well as the traffic sector. The focus is on event-driven autonomous decision making (e.g. (software) agent systems), machine learning and synchronized data processing as well as the use of digital twins and electronic horizons.