【作者】 黄润才；

【导师】 曹奇英；

【作者基本信息】 东华大学 ， 控制理论与控制工程， 2010， 博士

【摘要】 基于任务驱动的普适计算模式可以充分调度和集成系统中的软硬件资源并将它们作为可利用的服务,此模式以高效完成用户任务为首要目的。由于系统中可利用的服务具有异质性、资源限制性和移动性,因而普适计算环境具有显著的动态特性,这给用户任务的完整执行带来了诸多的不确定因素。因此,任务驱动的动态服务集成机制便成为了普适计算研究领域的一个重要方法和途径。本研究从用户任务的驱动出发,针对普适计算环境的动态特点进行服务集成机制的研究,主要从任务驱动的层次结构、动态服务的模型分析、任务驱动的服务集成以及服务集成的中间件机制等几个方面展开研究,通过本研究,可以为普适计算的动态环境提供服务集成的有效机制,为用户任务在普适计算环境下的执行提供高效的途径。(1)从任务驱动的概念出发,在阐述了任务的属性及层次关系的基础上,详细分析了上下文依赖的主动任务模型；设计了任务驱动的层次结构,并给出了各层的详细说明；最后就任务驱动的实现机制进行了研究,从任务的主动特性分析、任务的属性关联以及主动任务的执行方面给出了相应的研究成果。(2)对动态服务的模型进行建模和分析。先后从动态服务的体系结构、动态服务的建模、集中式动态服务、分布式动态服务以及动态服务的综合分析等五个方面详细探讨和研究了动态服务的基本思想,有针对性地为普适计算环境下动态服务的聚合和任务的分解提出了相应的解决方案。(3)综合分析了任务驱动的服务集成机制,提出了一系列有效的解决方案及其实现算法。首先从任务驱动的服务转换和基于会话的存取控制的研究出发,定义了基于会话的服务访问机制；然后在研究上下文建模、上下文本体设计、上下文分类等方面的基础上,设计了一个任务驱动的服务管理体系结构,并对该结构进行了详细的分析和讨论,提出了相应的上下文感知服务管理机制；接下来针对任务驱动的特点,从服务发现的框架、服务发现的查询以及服务发现的过程等三个方面,对任务驱动的服务发现机制进行了详细研究,设计了相应的解决方案；最后分别利用粗糙集和遗传算法的有关思想,设计了相应的算法,实现了动态服务的匹配和多任务多服务的分配问题,实验表明这些算法是可行有效的。(4)在上述研究成果的基础上,分析了普适计算环境下的中间件需求,讨论了中间件应用的关键问题,设计了基于任务驱动的普适计算中间件模型及体系结构,并阐述了该体系结构的各组件功能,给出了任务驱动应用中服务集成的策略方法,讨论了该体系结构的有效性。更多还原

【Abstract】 In task-driven pervasive computing, software and hardware resources can be fully scheduled and integrated to form available services, and effective implementation of user task is given the first priority. As the services in pervasive system are featured by heterogeneity, resource-limitation and mobility, pervasive computing environment is characterized by dynamism, which brings uncertainties to the fullfillment of user tasks. Consequently, task-driven dynamic service integration mechanism has become an indispensible aspect of pervasive computing.The paper mainly focuses on the task-driven user task and researches on service integration mechanism with aspects to dynamism in pervasive computing environment. Task-driven hierarchy, model analysis of dynamic service, service integration of task-driven and middleware mechanism of service integration are discussed to present a effective mechanism for service integration and a useful method for user task execution in dynamic pervasive computing environment.(1) From the perspective of task driving, the attributes and hierarchy of task are analyzed to give a detailed description of context-aware active task model. A task-driven hierarchy is designed with a clear explaination of each layer. The paper researches on task-driven integration mechanism and specifies on activenesss anaylsis, attribute association, active execuation of task.(2) In the process of modeling and analyzing dynamic service, architecture of dynamic service system, modeling of dynamic system, centralized dynamic service, distributed dynamic service and comprehensive analysis of dynamic service are discussed to demonstrate the basic concepts in dynamic services. A solution for the aggregation and decomposition of dynamic services in pervasive computing environment is put forward.(3) On the basis of task-driven service integration mechanism, effective solutions and implementations are presented. A session based service access mechanism is put forward with task-driven service transformation and session based access control mechanism. Task-driven service management architecture is designed on the basis of context modeling, ontology design and classification. And then a context-aware service management mechanism is presented. Furthermore, with aspect to the features of task-driven, service discovery framework, service discovery query and service discovery process are discussed to specify the task-driven service discovery mechanism and the relevant solution is designed. Finally, rough set theory and genetic algorithm are applied to realize dynamic service matchmaking and multi-task multi-service allocation. The experiment result shows that the algorithm is feasible and reasonable.(4) Above all, middleware requirements in pervasive computing environment are analyzed and some key issues in middleware applications are discussed. A task-driven pervasive computing middleware model and architecture are presented. Moreover, the functionalities of each component are clarified to explain the task-driven service integration strategy and the validity of the architecture is discussed.更多还原