A context consistency management technique using on-time context elimination rules for ubiquitous computing systems

Abstract

Many context-aware systems have been built without context consistency management. The wrong/invalid contexts break down the consistency of the knowledge base and, therefore, cause abnormal actions in context-aware services. In this paper, we propose a personalized on-time context-elimination scheme for improving context consistency that allows context manager to delete inconsistent contexts timely and automatically. The proposed scheme detects the contexts that are either wrong or invalid for the present situations of each user using ``context-elimination rules`` by analyzing context log. The context manager registers the generated rules and continually checks the consistency of the whole knowledge base using those rules. We have designed the scheme and implemented into our context-aware system called Active Surroundings. Evaluation results show that the ratio of context inconsistency considerably decreases as the system proceeds with the proposed scheme in runtime. Therefore, the proposed scheme helps KB keep the optimized number of contexts and reduce inference time for the system to provide application services timely.