Towards a proactive system based on activity recognition

Recent progress in sensing technology has led to ever more inexpensive and smaller sensors. This has opened the door for novel applications. A particularly promising one is what is called “proactive computing” [1], which suggests an anticipatory aspect to ubiquitous computing. An anticipatory system can be defined as “a system containing a predictive model of itself and/or of its environment that allows it to change state at an instant in accord with the model’s predictions pertaining to a later instant” [2]. In order to generate and maintain such predictive models, a system needs context-awareness, that is, the ability to determine what to observe in its environment and how to interpret it [3]. The predictive models allow the system to pro-actively adjust to changes in the real-world. In this work, we address underlying principles of preventive systems for proactive applications in closed domains