RTCR: a soft real-time context reasoner

Abstract

Context-aware applications in pervasive computing environments consume a large variety of context information. Some of the contexts are highly dynamic and applications put real-time requirements on their provision. This kind of requirement applies to any component that provide such contexts, including context reasoners, which synthesize high level contexts from low level ones. This paper presents the architecture of real-time context reasoner (RTCR), a context reasoner that is designed to satisfy soft real-time requirements on the reasoning process. It manages to meet timing constraints by scheduling executions of reasoning jobs with respect to demanded deadlines on result contexts. The reasoner follows a first order logic model and uses rule based reasoning. A heuristic temporary rules mechanism is introduced in its design to reduce repeated reasoning at run time. Simulation results show that the reasoner provides all demanded contexts on time in situations where an identical reasoner without real-time scheduling misses many deadlines. Another experiment illustrates the potential of the temporary rules mechanism in reducing workload for the reasoner.