Extending context spaces theory by proactive adaptation

Context awareness is one of the core features of pervasive computing systems. Pervasive systems can also be improved by smart application of context prediction. This paper addresses subsequent challenge of how to act according to predicted context in order to strengthen the system. Novel reinforcement learning based architecture is proposed to overcome the drawbacks of existing approaches to proactive adaptation. Context spaces theory is used as an example of how existing context awareness systems can be enhanced to achieve proactive adaptation. This recently developed theory addresses problems related to sensors uncertainty and high-level situation reasoning and it can be enhanced to achieve efficient proactive adaptation as well. This article also discusses implementation options and possible testbed to evaluate the solutions