A Study Of Service-Oriented Dynamic Maintenance And Resource Configuration In The Internet Of Things

The internet of things(IOT) is considered to be the next step in the revolution ofinternet.People are expecting to be able to quickly connect between objects existed allthe time no matter where.This requirement not only propels rapid development of IOTtechnology and wide acceptance of RFID devices,but also deviates the traditional com-puting style to smart IOT computing.Compared to hardware and network support in per-vasive computing,the reliability of sensors software for IOT,which in return,limited theapplication and deployment of IOT to some extents. Unlike the traditional distributedcomputing software running in stable environments with rich resources,sensor nodes inIOT runs in unforeseen wireless network environments,limited resources and continuous-ly altered context.These put a great deal of demands on the dynamic software,resourceconfiguration and energy maintenance technology.From perspectives of service, includ-ing resource service,data transmission service and component service,this thesis mainlyconcentrates on the dynamic maintenance and resource configuration issues at runtimein IOT infrastructure environments.In order to minimize the resource requirements andpoint-to-point and end-to-end delay in a dynamic IOT environment,multiple technolo-gies,such agent,ECA rules mechanism,game theory,graph theory,lagrangian optimizationand fuzzy control are further explored and used to study problems in device service net-works dynamic maintenance,dynamic software updating maintenance,dynamic resourceadjusting and configuration optimization and rechargeable RFID sensor networks dy-namic energy maintenance.The major contributions of this dissertation are:(1) Aiming at the failure of components and link problem in device service net-works of internet of things,reliability of components and link is analyzed theoretical-ly.This dissertation proposes four service maintenance strategies.Firstly,idle service com-ponent joining in service network.Secondly,component-based migration maintenance s-trategy,which analyzes degree of components migration and link probability and presentscomponents migration three-tier framework based on CM-ECA rules.Thirdly,link-basedmaintenance strategy,which is local maintenance method,in case of failure of the com-ponent migration,checking its one-hop neighbors in order to quickly recover ser-vice.Finally,path reconfiguration strategy,the failure of the link is estimated and reconfig-uration of the target is analyzed theoretically.The dissertation establishes the path proba-bility model for the reconfiguration analysis in order to minimize service recovery time.(2) A key challenge of IOT is to assure high reliability software in a scalable anddynamic computing environment for data dissemination and task cooperation amongsensor nodes.To support higher reliability,sensor software is often required to provideupdating service.This dissertation presents a service pack update framework for ser-vice of sensors software management in the internet of things context environment.Itconsists of three methods.The method of One-Sink-To-One-Sensor(OSTOS) solves keysensor node software update or upgrade.However,with the growing of the number of sensor nodes,to solve the problem of updating multiple sensors software,One-Sink-To-K-Coverage-Sensor(OSTKS) and Multiple-Sink-To-All-Coverage-Sensor(MSTAS) canbe used in multi-sink area sensors networks. The results show that the proposed methodscompared with flooding and unicast update reduce the traffic for service pack deliveryand keep high delivery ability for update or upgrade operations.(3) Aiming at the defect that the rate between tag and reader is not match and theirresource-constrained problem in RFID systems,the collaboration communication contex-t is analyzed theoretically.Service-oriented dynamic resource configuration optimizationscheme in RFID systems is proposed which uses the component QoS information of tagand reader to establish closed-loop control framework and develops three control strate-gies to adjust QoS parameter according to the state and type of its resources.The schemewas analyzed and validated for performance through experiments.The numerical exper-iments results show that the optimization mechanism can make throughput improve,useresources efficiently and reduce the energy consumption and interference of reader-to-reader in RFID systems.(4) Aiming at the proliferation of networks lifetime for adaptive and dynamic IOTenvironment,this dissertation presents a lightweight overview of service-oriented RFIDsensor networks architecture (SORA) and develops an algorithm of discovery resourcepath coupled with the use of agent technologies.Each node actively publishes its QoSresource state so that resource coordinator(RC) can act on behalf of users or objects toobtain the best service resource in order to reduce the point-to-point and end-to-end delayin RFID sensor service networks and discover paths of energy maintenance.Then,the dis-sertation proposes deeper analysis on the costs of recharge for low energy resource pathby combining point and path energy maintenance in order to minimize energy problemfor a rechargeable RFID sensor service networks.To improve throughput of one-hop multiple active RFID tags identification sys-tem,this dissertation proposes a share node energy recharging strategy for energy-limitedmultiple RFID tag identification system.Firstly,it can dynamically adapt the variable en-ergy resources due to the cognitive radio technique.Secondly,it combines the energy con-trol with tags groups in wait time T,through classifying the tag into different groups ac-cording to its distances.Thirdly,it introduces the optimization theoretical analysis energyfor multiple RFID tags identification system,so as to minimize the time and energy ofit takes to send tags data to reader.We compare and analyze effiect of tag number on thepower consumption,identification time and system throughput of different groups withnon-share energy in RFID systems.Experimental results demonstrate proposed strategyreduces both total power consumption of all tags and total identification time at the sametime,improve RFID system throughput.(5) A dynamic maintenance prototype system in IOT environments has been imple-mented in this study,including HIO communication pattern application,dynamic softwareupdate coordinator,dynamic resource configuration coordinator,dynamic energy data col- lection coordinator. The prototype system is implemented by using different programlanguage,because prototype system involves in different devices platform.This dissertation takes the service networks maintenance in the internet of thingscomputing environments as the research objective,and focuses on theories and techniquesassociated with service-oriented dynamic maintenance and resource configuration inIOT infrastructure environments.Simulation experiments and the prototype system havedemonstrated the validity and practicability of the proposed theories and algorithms.