【作者】 冯名正；

【导师】 顾冠群；

【作者基本信息】 东南大学 ， 计算机应用技术， 2006， 博士

【摘要】 随着计算机网络技术和应用的迅猛发展,人们对互联网的需求也进一步提高,希望互联网能够有效地提供他们所需的各种服务。计算机网络及其应用的研究重心呈现出向应用层延伸的特征。未来的下一代网络将是一个面向用户的、以服务为中心、具有QoS使能的网络,可以提供丰富多样的服务,具备服务定制和和管管理的功能。在应应用用领域,经济竞争的加剧需要企业能够对市场和用户需求的变化快速反应,这要求企业应用能够利用已有资源,在新的Internet环境下根据用户的新新需需求,整合出符合要求的、有效的新服务,这已经成为一个迫切需要得得到到解解决决的问题。服务组合问题的提出是应应用用需求求驱驱动动的的结果。当前正在迅速发展的是基于Web标准的服务组合,该方法被广泛用来改进企业软件系统的敏捷性、灵活性和可用性。在新的面向服务的应用环境中,在业务需求驱动的促使下,以Web服务方式进行的、有服务质量确保的Web服务组合已经成为Web服务领域研究的重点。在Web服务组合研究领域,相关研究已经展开,但是在对Web服务协作流程中每个节点上的服务选择如何确保全局服务质量问题方面还研究甚少,同时在对服务组合方案的评价优化问题方面尚无相关研究,这两方面的薄弱影响了Web服务组合的广泛应用。如何较为圆满的解解决决以上问题还是一个崭新的研究课题。论文首先综述了服务组合相关研究领域的研究现状,对相关基本概念、术语进行了统一和定义。随后,集中研究Web服务组合的更为通用和全面的描述,提出了一个开放的、具有可扩展性和可伸缩性的Web服务组合模型:包括了一个通用和统一的抽象框架模型,以此作为开展服务组合其他部分研究工作的基础;以及一个灵活可扩展的QoS组合模型,提供对服务质量确保机制的支持。基于上述在服务组合模型方面的研究成果,论文提出了一个具有QoS保证的服务选择算法SS-IP,将服务选择问题抽象成整数规划问题,利用多目标规划技术得到相关的质量约束,最终得得到到结合多目标规划和整数规划的服务选择算法。然后,基于层次分析法和网络计划方法,论文研究了对多个组合方案进行评估优化的最优化算法PlanEO,提供在多维QoS参数空间下对多个可行方案进行评价和优化的通用方法。另外,论文还设计实现了服务组合原型系统,提供了对以上模型和算法的验证和检验,同时系统也可以投入到制造业、服务业等应应用用领域,解解决决实际的服务组合问题。最后,总结全文研究工作,进行了研究展望。通过有机的综合应用相关多学科的理论和方法,论文对Web服务组合问题进行了合理的分析建模,建立了更为通用和灵活的服务组合框架模型和QoS组合模型。更进一步的,论文对服务选择问题和方案评价优化问题开展了深入研究,对这两个问题进行了有效地描述和求解,提出了服务选择算法SS-IP和评估优化算法PlanEO。同时,论文设计和实现了支持以上模型和算法的服务组合原型系统。论文对Web服务组合问题及其关键技术的研究很好的顺应了企业应用需求和计算机网络技术两方面的发展趋势,研究工作和成果有机的综合应用了多学科领域知识,完善了已有研究并特别弥补了在QoS方面的不足,具有较高的理论指导意义,能够对当前的Web服务组合研究提供有益的推进作用;同时,论文提出的Web服务组合系统提供了服务选择算法和方案评价优化算法,具备较高的实际应应用用价值,拥拥有有广阔的应应用用前景。更多还原

【Abstract】 With the mushroom of the network technology and its application, user’s requirements were en-hanced. Thus, various services which were able to satisfy all requirements effectively were expected.The center of gravity for the Network research has been extended to the application layer in Network.The next generation network in the future would be a user-oriented, QoS-assured network which focuseson service. It would provide a lot of various services and has the ability to customize and manage theseservices. In order to enhance competitive preponderance and adaptability, enterprises should enhancetheir agility and ?exibility in response to customer’s and market’s needs.The business software system should have the ability to compose congruous and available new ser-vices, based on the new requirements in the new Internet environment, utilizing those legacy resources.This has become a challenging problem needing to be resolved imminently. The problem is the result ofdemand-driven.Web Services Composition (WSC) which is growing rapidly could be an excellent approach toachieve all these goals. It supports business agility, ?exibility, and availability–the main features in E-Business software. In the new service-oriented application environment, under the boost of requirements-driven, WSC with QoS-assured has turned into a primary research field in Web Services.In the past few years, many issues have been addressed. In each node of WSC plan (process /flow);there exist lots of Web Services with the same function. How to select a right one exactly in QoS view?Related research had some disadvantages in this problem named Service Selection. On the other hands,with the global Web Services market, there exist a number of WSC plans satisfying functions require-ments. Those plans provide different QoS. How to valuate or optimize them in QoS view? Heretofore,there have not some works on plan valuation/optimization. Weakness in there two aspects hamper thewide deployment of WSC. How to resolve these issues ideally is an open problem.First of all, this paper defines the basic notion, summarizes the initial research and the actualityin related domains. Secondly, a common abstract framework model and a QoS composition model arediscussed in detail. With the production in model aspect, based on Integer Programming (IP) and Multi-objective optimization, a Service Selection algorithm named SS-IP is presented. Subsequently, PlanEvaluation/Optimization is considered and resolved by PlanEO algorithm, based on Analytic HierarchyProcess (AHP) and Net Plan. Furthermore, a prototype system is designed and implemented to validateabove models and algorithms. This prototype system also could be deployed into some real applicationfields. Finally, a conclusion is presented, associating with some expectable trends.Based on utilizing some theories and methods in related fields, we researched on the analysis andmodeling of WSC and presented two models (abstract framework model and QoS composition model)which are more general and ?exible. Then we took an in-depth study in services selection and planvaluation/optimization, and contributed two algorithms (SS-IP and PlanEO). Further more, a prototype更多还原