Towards Seamless Binding of Context-aware Services to Ubiquitous Information Sources

The area of context-aware computing has received much attention in the last decade. However, many systems to determine a user's context are focused on integrating a confined set of sensors into their decision-making algorithms, and lack extensibility for new and novel sources of context information. This limitation however has often resulted in monolithic software systems, reasoning on the user's context from a static set of pre- defined rules. Therefore, most of these systems can not adapt to the user during runtime, and have not found wide adoption in reality. Unlike the existing approaches, we present a concept for a context-aware system circumventing these limitations by providing generic abstraction layers among its components. We show a concept for an extensible context-aware system with autonomous adaptation to changes in the user's preferences, and provide arguments for our design decisions. After presenting the constraints for all participating entities, including sensors, middleware, and actuation and/or application frontends, we describe ContextFramework. KOM, an example implementation of the proposed concept.