Research On Multi-index Quality Management Framework And Inconsistency Elimination Algorithm For Context-aware Systems

Ubiquitous Computing(UbiComp)aims to provide users with anytime,anywhere and transparent services.As one of its main technical applications,context-awareness system can automatically perceive useful context and actively provide users with the required services after information processing.With the progress of various sensing devices and the improvement of computing power,context-awareness systems have been applied to many fields such as smart home,smart healthcare and smart agriculture,and the degree of intelligence of them is constantly improving.However,all kinds of context-awareness systems still lack a unified quality assessment and management framework at present,and further research is needed in establishing the interaction between devices and improving the adaptive optimization ability of the system.On the other hand,obtaining context is the basic step in the context-awareness systems.The quality of context(QoC)will affect the accuracy of the system’s subsequent decisions and service recommendations.In scenarios with multiple context sources,the redundancy of context is usually used to exchange for certain reliability.However,at the same time,due to the multi-source and heterogeneity of equipment,network equipment or service anomalies and other reasons,the context is often inconsistent,and thus the quality of service(QoS)and quality of experience(QoE)of the system are reduced.In view of the above problems,this thesis has carried out the following research based on the improvement of adaptive ability and QoC of context-aware system:(1)For the quality evaluation management and adaptive optimization of context-awareness systems,a multi-attribute quality management framework for context-awareness systems is proposed.The performance of context-awareness systems is comprehensively evaluated by studying the effects of multiple quality indicators such as quality of device(QoD),QoC,QoS,quality of actuation(QoA),and QoE on the evaluation of system quality.At the same time,the quality of machine experience(QoME)is added to the QoE index,and combined with the QoA index to enhance the interaction between devices in the system,reduce the disturbance to users,and improve the adaptive adjustment ability of the system.(2)In order to improve the measurement accuracy of QoC indicators,and take full account of the correlation between context and its own uncertainty,a new QoC measurement indicator,comprehensive measure of correctness(CMoC),is proposed.The CMoC indicator combines information from multiple dimensions of belief entropy,evidence distance and correctness(one of the basic parameters of QoC)to comprehensively evaluate the correctness of context,enhances the reliability of QoC level evaluation,and provides a new dimension for the improvement of context inconsistency elimination algorithm.(3)Aiming at the problem of context inconsistency in context-awareness systems,combined with the proposed CMoC index,an algorithm to eliminate the context inconsistency based on CMoC and two-dimensional mass function(TDMF)is proposed.The CMoC is used to comprehensively measure the correctness of context,which is used as the benchmark for eliminating the context inconsistency,and the TDMF is introduced as the uncertainty processing method to further deal with inconsistent context.The method of constructing TDMF using CMoC,membership degree and some traditional QoC parameters is studied to make up for the problems of evidence conflict in traditional Dempster-Shafer(D-S)evidence theory.Taking the humidity monitoring in smart home as the application background,multi-dimensional simulation comparisons are carried out to verify the effectiveness and stability of the algorithm.Through the above research,an efficient and active quality assessment and management framework of context-awareness system is constructed,which provides an effective solution for the unified quality assessment and management of different systems.On the other hand,the research on the algorithm for context inconsistency elimination improves the accuracy of the decision for eliminating the inconsistency,provides a new idea for the combination of the uncertainty processing method and the context inconsistency.