Context management and situation reasoning for ubiquitous applications in mobile peer-to-peer environments
2017-02-09T06:01:26Z (GMT) by
Recent advances in mobile computing, networking and sensing technologies have enabled mobile devices to become a major development platform for various applications, such as disaster prevention, environmental monitoring and intelligent transportation systems. These devices can be programmed to individually sense and capture context information and develop situation awareness. Mobile devices typically have limited capacity in terms of processing and sensing capabilities. Thus, to gain a better understanding of their surrounding situations, co-located mobile devices can jointly establish a mobile peer-to-peer (MP2P) environment and share captured context information. However, available context information may be uncertain, ambiguous, obsolete or even irrelevant. These issues can influence the reliability of the produced context information used for situation reasoning and/or adaptation criteria. This dissertation proposes, develops and validates a novel distributed middleware framework for context management and situation reasoning for ubiquitous applications in MP2P environments. The first contribution of this dissertation is our theoretical model termed CoMoS (Context Mobile Spaces), which is an approach to represent context and situations for ubiquitous applications in MP2P environments. Our second contribution is the formulation, development and validation of a theoretical approach to estimate the reliability of context information captured in MP2P environments. A data fusion technique based on Dempster-Shafer theory is integrated into the proposed context and situation model as the basis for situation reasoning using the captured context information. Our third contribution is the design, implementation and validation of DiFraCSi2 (Distributed Framework for Context Management and Situation Reasoning for Ubiquitous Applications in MP2P environments) middleware. This middleware implements the proposed concepts, models and algorithms of CoMoS. To minimise processing overheads, an event-based strategy to control the related processes is also developed and validated. We evaluate the contributions of the proposed theoretical model and framework through extensive experimental studies. The high accuracy of the situation recognition rate when using the DiFraCSi2 middleware in different situation scenarios highlights the significance of this research project’s contribution in providing situation-reasoning services. Performance of the proposed middleware functionality is also reported and analysed. The experimental results highlight the feasibility, significance and usability of DiFraCSi2 middleware in managing context and reasoning services. Thus, the research work presented in this dissertation takes a step forward in enabling reliable context management and situation reasoning for ubiquitous applications in MP2P environments. The results of this research project have been published in five peer-reviewed international conference papers, one technical report and one journal article submitted for review.