A Hybrid Approach to Context Modeling in Large-Scale Pervasive Computing Environments

Abstract

Pervasive computing aims to unobtrusively embed computer systems into everyday life environments to enrich the user experience without demanding the user‟s explicit attention. For pervasive computing to be minimally invasive, computer systems must be conscious of and ultimately, be able to act according to the context of the user and her intelligent environment. However, such context is often heterogeneous as it is derived from myriads of independent systems and sensors, can be incomplete, and may even be erroneous. For independent systems to manage, share, correlate, and reason over context, contextual information must be modelled in a homogenous fashion. This paper proposes a hybrid approach to modelling contextual information that incorporates the management and communication benefits of traditional object-oriented context models, while also taking advantage of the semantic and inference benefits of ontology-based context models. Unlike other approaches, our hybrid model has been designed to support a specific large-scale pervasive domain, namely the transportation domain, and promotes exploiting primary context as the key to accessing and correlating distributed knowledge. The Primary-Context Model and the Primary-Context Ontology are part of a pervasive middleware architecture for integrating independent Intelligent Transport Systems and pervasive transportation services and has been applied to a prototypical realisation of such system and service integration.