Model-driven techniques have proven to yield significant benefits for context-aware systems. Specifically, semantically-rich models are used at runtime to monitor the system context and guide necessary changes. Under the closed-world assumption, adaptations are fully known at design time. Nevertheless, it is difficult to foresee all the possible situations that may arise in uncertain and complex contexts. In this paper, we present a model-based framework to support the dynamic evolution of context-aware systems to deal with unexpected context events in the open world. If model adaptations are not enough to solve uncertainty, our model-based evolution planner guides the evolution of the supporting models to preserve high-level requirements. A case study about a context-aware Web service composition, which is executed in a distributed computing infrastructure, illustrates the applicability of our framework. A realization methodology and a prototype system support our approach.
Unfortunately, ACM prohibits us from displaying non-influential references for this paper.
To see the full reference list, please visit http://dl.acm.org/citation.cfm?id=2404971.