【作者】 李源惠；

【导师】 赵德鹏；

【作者基本信息】 大连海事大学 ， 交通信息工程及控制， 2011， 博士

【摘要】 以网络化信息技术为支撑的航运信息服务平台是现代航运服务模式的重要标志之一。航运信息服务平台以实现众多组织参与的、广域分布的航运信息资源交互与共享、系统集成为目标、以实现船舶、航道、货物等航运要素的动态信息适时交换为重点,建设一个面向航运全行业及社会公众的、集行业管理、运输组织及运输服务于一体的航运信息服务平台。其构建必须充分考虑已有系统和资源的有效利用,需要整合不同时期、不同类型和跨行业边界的异构系统；消除信息孤岛、实现系统间协同工作以及信息共享与交互；能够快速响应航运动态变化的业务需求和实现自由扩充；系统的扩展性和开放性、异构系统松耦合交互和节点动态性等问题。网格(Grid)技术提供了研究解决上述问题的有效途径。它将高速互联网、高性能计算机、大型数据库,传感器,远程设备以及各种航运资源融为一体,实现全面共享与互操作。而服务网格将Web Service体系结构与网格相结合,提出了“网格服务”的概念,用于解决服务发现、动态服务创建、服务生命周期管理等与临时服务有关的问题,从而保证了服务网格能够及时地响应与处理各种航运信息的服务请求,为存储于网格上的各种航运资源提供一种安全、高效的共享机制,使得各种异构和孤立的资源能协同工作,以一个整体服务于不同的应用,实现资源在跨组织之间应用的共享与集成。鉴于此,本文研究的重点是探讨将服务网格应用到航运信息系统,搭建灵活、高效、可靠和可扩展的航运信息服务平台。在构建了面向航运的信息服务网格体系结构的基础上对其中的关键技术,服务网格的任务调度和资源分配进行了分析与研究,为航运信息化的发展提供了新的思路和方向。为服务网格在航运信息系统进行进一步的应用研究奠定了坚实的基础。首先,本文分析了航运信息系统具有海量数据以及异构系统之间缺乏信息共享机制和协调机制等现实情况,并对现有的服务网格系统进行了探讨和研究。在OGSA和WSRF基础之上提出并构建了面向航运的信息服务网格体系结构。将服务模型融入到网格之中,网格中的节点基于服务的概念,应用程序被划分成更小的组件,大大提高了网格各个节点之间的松散耦合性能。同时由于每个节点的组件是基于服务的,可以方便的在节点与协调者之外建立节点与节点之间的通信,从而实现更加灵活的分布式、自由通信的结构。该体系结构充分考虑了网格的服务发现机制、策略机制、安全机制、资源管理机制和调度机制等多项功能模块的配置,使得网格用户能够快速的匹配到需要的服务,并在与系统内其它用户协调的基础上,在授权的范围内,能够优化服务,利用最优资源,实现自己的用户需求。满足了航运系统中各个异构系统和平台的共享和协作问题。文中对体系结构中的网格服务层结合航运信息系统进行了展开描述,通过模拟案例描述了为航运系统用户提供服务的流程。其次,本文分析了服务网格任务调度的目标和对当前各种任务调度算法进行了分析研究,考虑了面向航运的信息服务网格中支持复杂多目标的服务质量(QoS)控制和用户任务的关联性的特点,参考Buyya提出的调度策略,在已有的基于计算经济调度基础上,提出一种基于关联任务的多个QoS属性效用函数调度算法。把用户的多QoS的属性描述为效用函数,考虑到多个QoS属性很难同时满足,而且有些服务目标甚至相互冲突的问题,本文采取将多个效用函数进行折中考虑得出一个综合的总效用函数,并将其作为任务调度优化的目标函数来驱动动态资源调度,寻找效用最大化的任务调度。本文通过GridSim对该算法进行模拟,并与现有经典的DBC-T算法、DBC-C算法和DBC-CT算法进行了比较,结果说明,该算法可以权衡多QoS属性,而不是单独优化一个目标,反映了用户的最佳效用,并且相对于现有的任务调度算法具有更高的执行效率。最后,对于面向航运的信息服务网格中的资源分配,本文在引入微观经济方法的基础上运用博弈论对服务网格中的资源分配进行了进一步的研究。根据对现有的服务网格资源分配方法的研究,结合航运信息服务网格中所获取的信息具有动态可变的特点以及各个服务请求之间对资源的占有和使用也存在相互依存和相互竞争的关系,设计了一种双模式的网格资源分配模型,运用竞价机制,在网格资源丰富时,采用动态的序贯博弈模式,基于Nash均衡最优确定网格最优资源分配,同时为提供者确定最优售价；在网格资源稀缺时,采用竞价模式基于效用函数的效用最大化为网格服务请求代理分配资源。保证了服务网格资源在多个服务请求之间的有效分配。文中还对这个模型中的几个重要问题,例如,如何使得网格资源利用的最大化等进行了进一步的探讨。为服务网格中的资源有效分配提供了一个良好的应用解决方案和进一步研究的基础。将服务网格引入到航运信息系统构建面向航运的信息服务网格是航运信息系统研究领域的一个新思路和新方向,具有很好的发展前景。结合服务网格以及其它的一些智能化方法,可以很好的解决复杂航运信息系统中的海量数据、异构平台和信息共享等问题。但本文的研究只是对航运信息系统与服务网格相结合的初步探讨,限于时间和精力,主要针对面向航运信息服务网格的体系结构、任务调度和资源分配进行了较为深入地研究。还有一些更深入和具体的工作需要进一步的研究和探讨。在本文的最后,给出了服务网格在航运信息系统的应用所面临的问题和未来研究的方向。文中的内容、观点不免有疏漏、偏颇之处,敬请不吝指正更多还原

【Abstract】 Shipping information service platform is one of main sign of modern shipping service mode, based on the network information technology. It takes shipping information interaction, sharing and system integration as goal, stresses on achieving dynamic information exchange for ship, cargo and sea-route etc, and finally build an information service platform facing to management, organization and service. The architecture of shipping information service platform must consider the effective utilization of existing system and resource, need to integrate heteroid systems for different periods, type and fields, eliminate isolated information resources and achieve cooperation, interaction and sharing among systems, react rapidly to dynamic business requirements, improve system expansibility and openness and make loose coupling system, etc. Grid provides the effective approach to research and resolve the issues above. Grid integrates high-speed Internet, high performance computers, large database, sensors, remote equipments and other shipping resources and achieves the sharing and interoperability within the system. Service Grid, combining Web Service architecture and Grid, proposes the concept of "Grid Service" and apply to service discovery, dynamic service creation and service life-cycle management etc. which are related to temporary services. This ensures the Service Grid to respond to and handle various kinds of requirements for shipping information, provide a safe and effective sharing mechanism for shipping resources, make various heteroid and isolate resources work cooperatively, in order to provide service to different applications as a whole and achieving resource sharing and integration among organizations.The emphasis of this paper is to discuss how to apply Service Grid in shipping information system, and establish flexible, effective, reliable and extendible shipping information service platform. Based on this Shipping Information Service Grid architecture, the paper analyzes and researches the key technologies including task scheduling and resource allocation of Service Grid and provides new direction and solid foundation for further research of applying Service Grid in shipping information system.Firstly, the paper analyzes characters of current shipping information system such as mass data and missing information sharing and cooperation among heteroid system, etc, investigates current Service Grid and propose to build Shipping Information Service Grid architecture based on OGSA and WSRF. This architecture integrates service model into Grid and make nodes in Grid based on service, so that applications are divided into more granular module and improve loose-couple capability among Grid nodes. Since modules on each node are based on service, it is easy to establish communication among nodes and cooperators and achieves flexible, distributed and unrestricted communicated architecture. This architecture completely considers service discovery, strategy, security, resource management and scheduling mechanism of Grid, to ensure Grid users discover required services quickly, cooperate with other users within Grid, optimize services according to his authority and utilize the best matched service to satisfy the user’s requirement. This resolves the sharing and cooperation issues among heteroid systems in shipping system. Also the paper investigates deeply on how to integrate Grid Service layer with shipping information system in this architecture and to describe whole process of how shipping system to provide services in this architecture according to one of cases.Secondly, the paper analyses goal of task scheduling of Service Grid, investigates and compares existing task scheduling algorithms. Based on this, the paper considers Shipping Information Service Grid to need to support QoS of complex and multiple objectives and the character of tasks related, then to propose a integrated utility of multiple QoS attributions scheduling algorithm based on correlative tasks, referring to Buyya’s scheduling strategy and computing economic scheduling. This algorithm describes multiple QoS attributions of users as utility function. Since it cannot satisfy multiple QoS attributes at the same time, even some services objectives have conflicts, the paper trades off among multiple utility functions and get an integrated general utility function, take this general utility function as optimized objective function for task scheduling to drive dynamic resource scheduling, finally find the task scheduling with maximum utility. This paper utilizes GridSim to simulate the algorithm, compares with classic DBC-T algorithm, DBC-C algorithm and DBC-CT algorithm. The result shows that this algorithm can balance multiple QoS attributes but not only a single optimized objective. This can reflect best requirements of users and has higher execution performance, comparing with current task scheduling. Finally, the paper investigates the resource allocation of Shipping Information Service Grid. This paper takes further research for resource allocation of Service Grid by utilizing game theory and microeconomic method. Based on the research of existing Service Grid resource allocation methods, according to the dynamic character of information of Shipping Information Service Grid and the correlative dependent and compete each other characters of different service requests for the resource occupation and utilization, the paper designs a bi-modes resource allocation model. This model applies compete price mechanism, when there are plenty of Grid resources, it applies proportional sharing mechanism pattern based on sequential game, make optimalizing resource allocation based on Nash equilibrium and provide optimal selling price for resource provider; when Grid resources are absent, it allocate resources by utilizing compete price pattern and based on the maximum utility function, which ensure Service Grid resources are allocated among multiple service requests. The paper also takes further discussion about some key points of this model such as maximumly utilizing Grid resources. This provides a good solution for how to effective resource allocation and a solid foundation for further research.The paper provides a new thought and a new direction of research of shipping information system, which is to introduce Service Grid and build a Shipping Information Service Grid. Integrating Service Grid and some other intelligent methods, this model can resolve key issues in shipping information system, such as handling complex mass data, heteriod system and information sharing, etc. But since the limited time and energy, this paper only did preliminary research on integration of shipping information system and Service Grid, it mainly focuses on research on Shipping Information Service Grid architecture, task scheduling and resource allocation. There are still more detail works need to be further researched and discussed in the future that are indicated in the last of paper.Please point them out if there are deficiencies in this paper.更多还原