Exploring Model Engineering Techniques for the Integration of Heterogeneous Context Models

The provision of context-aware services in terms of the anytime/anywhere/anymedia paradigm requires the consideration of certain context information in terms of, e.g., time, location, device, and user’s preferences. These various context information will often be provided through a series of various sources being available in diverse formats and standards, e.g. device profiles, user databases, or Geo-Services. Nevertheless, not before being able to consider context information in combination during the development of such context-aware systems, the full potential of contextawareness cannot be effectively exploited. Whereas existing work mostly perceives the integration of different context sources at the system level, e.g., by a framework approach or dedicated middleware, we tackle the problem in this paper differently, namely at the model level. We propose the use of model-driven engineering techniques to facilitate the integration of context sources in a bottom-up way. For this, ModelCVS, a semantic infrastructure for model-based integration allows to overcome existing heterogeneities of various context sources through a metamodel-based transformation approach. In the scope of this paper, we demonstrate the applicability of ModelCVS in the course of three example scenarios pointing out potentials and limitations of such a modelbased integration approach.