A Study Of Context-Aware Service Adaptation In Pervasive Computing

Recently, with the rapid development and popularization of wireless networksand portable devices, a mobile pervasive computing platform, consisting of variousnetworks, portable devices, sensors and servers, is gradually emerging andpenetrating into people’s daily lives. Pervasive computing enables people to accessto the variety of information services with their portable devices at any time andany place. However, compared with the traditional distributed computing withresource-rich and relatively stable environments, pervasive computing reveals itsmobile, resource-constrained, and changeable weakness. Accordingly, theapplications of pervasive computing need the support of context-aware serviceadaptation so as to provide satisfying services to the mobile users.This dissertation focuses on the context-awareness techniques and the adaptiveservice management in pervasive computing. The related study applies first-orderpredicate logic, fuzzy logic and hierarchical evaluation, and fuzzy logic control, etc.to the deep investigation of context model, service evaluation, selection, as well asQuality of Service (QoS) management. To meet the users’ requirements on servicesand to improve the utilization ratio of limited resource, this dissertation carries outa systematical research on the key techniques of service adaptation, such ascontext-aware framework, service provisioning model, and service configurationapproach.The major points and contributions of this dissertation are presented asfollows:1. Based on the previous research on the framework of pervasive computing applications, this dissertation has structured a contextinformation service framework which includes context acquisition,classification, inference, as well as delivery mechanism, to providecontext service for the development and implementation ofcontext-aware service adaptation. Furthermore, context information hasbeen modeled as an extended predicate model with time tags and ametrics type, which can be applied as a unified context presentation tofacilitate context sharing and reuse.2. According to the attributes of pervasive computing, the thesis propsed“Degree of Service Satisfaction (DSS)” model, focues on the mobileusers and the portable devices. Considering the fuzzy features of humanperceptive and complex context in pervasive computing, a hierarchicalQoS evaluation aplgorithm has been developed in this study, includingfirst-order predicate logic and fuzzy logic inference. Additionally, wehave probed into the relationships between the number of service/mobilepeers and the system overload, to instruct the improvement of systemperformance. The related experimental results have demonstrated thesound practicability of the algorithm with relatively low system overload(less than1.35seconds) in response to a large number of serviceprovisioning requests.3. In light of the fluctuations of critical QoS caused by the mobility andresource-limitedness, the management process of critical QoS has beenmodeled as a closed-loop control system. Then, based on thefuzzy-control strategy, a dynamic service configuration approach hasbeen proposed. Moreover, two kinds of fuzzy logic controllers and aconfiguration algorithm have been presented and implemented in thisstudy. They can adaptively drive service chain reconfiguration and QoSparameters readjustment in service components, in tune with the variations of the critical QoS and the constrained resource. The results ofthe experiments have showed that, compared with other methods, theproposed approach can obtain higher average value of critical QoS andlower value of critical QoS vibration.4. By utilizing Context Toolkit, the Jess, and the open source package offuzzy control engine, a prototype system for service adaptation has beendeveloped and implemented in the pervasive computing environments inthis study, including service evaluation, selection as well as algorithm,the configuring algorithm, and the fuzzy logic controllers for the serviceconfiguration. The server-end part of the prototype system isimplemented by using Eclipse Platform with Java, while the mobile-endpart employing EclipseME along with the SDKs for J2ME provided bythe various smart phone manufacturers.This dissertation takes the service adaptation in pervasive computingenvironments as the research objective, and focuses on theories and techniquesassociated with the service provisioning and service configuration in the adaptationprocess. The results of the simulation experiments and the implementation of theprototype system have demonstrated the validity, scalability and practicability ofthe proposed models and algorithms.