Towards the adaptive integration of multiple context reasoners in pervasive computing environments

The pervasive computing vision consists in realizing ubiquitous technologies to support the execution of people's everyday tasks by proactively providing appropriate information and services in a natural and transparent way based on the current context. Hence, a fundamental ingredient of pervasive computing is a mechanism to recognize the current high level context of users based on lower level context data provided, for instance, by body-worn and environmental sensors. Given the variability of encountered contextual conditions, the currently available data sources are highly dynamic; hence, context reasoning should continuously adapt to the change of available sources. In this paper we propose a technique to dynamically discover sources of context data, and to modularly integrate reasoners that use those data to infer higher level context information. Our proposal is corroborated by an implementation on mobile devices and sensors, and by an experimental evaluation showing its efficiency and effectiveness.