Possibilities and limitations of context extraction in mobile devices: Experiments with a multi-sensory personal device

This paper describes a context extraction framework for a mobile device that equips a variety of sensors. The framework captures context about a user and her surrounding environment; the information is used to adapt the behavior of applications running on the mobile device. Our framework adopts the blackboard architecture to execute multiple analysis modules that analyze signals from respective sensors. Respective modules implement different algorithms to complement each other's results to retrieve accurate high-level context. Based on experiments with a sample application, we evaluate the feasibility of the framework and point out the possibilities and limitations of context extraction in mobile devices.