Within the PROACTIVE project, we research how to realize a generic framework that allows context-aware applications to adapt proactively, i.e. by preparing for, or counteracting, future context events.
Pervasive computing applications can adjust their behavior to a multitude of information deemed to be relevant for their situation, their so-called context. Thus far, however, adaptation in such context-aware systems is reactive and limited to the application itself. These restrictions inevitably delay adjustments to events. They cause frequent reconfigurations, and may result in inferior overall system configurations. Within the PROACTIVE project, we research how to realize a generic framework that allows context-aware applications to adapt proactively, i.e. by preparing for, or counteracting, future context events, in order to remedy these shortcomings. The framework consists of (i) a context management component with prediction capabilities, (ii) an application model for calculating adaptation alternatives, and (iii) a pool of adaptation strategies for decision-making.
The figure above shows our approach to proactive adaptation, which is based on a trisection of the research question. The context prediction management (CPM) is responsible for providing suitable access to context information as well as predicting context, whereas the application model (AM) is responsible for finding all possible configurations based on the provided context and the set of context requirements posed by the application. Finally, the adaptation strategies (AS) can determine the best chain of adaptations by applying their predefined policy.