节点文献
基于QoS的Web服务发现与组合方法研究
Study on Methods of Web Services Discovery and Composition Based on QoS Technology
【作者】 龚小勇;
【导师】 朱庆生;
【作者基本信息】 重庆大学 , 计算机应用技术, 2008, 博士
【摘要】 电子商务应用的不断深入发展要求分布在不同地理位置、具有多种多样的硬件平台、操作系统、编程语言、软件技术和数据库管理等技术因素的企业应用系统可以在Internet环境下进行企业之间的应用互操作和应用集成,要求能够根据不同的应用需求,快速、灵活地发现合适的业务应用程序,或者在单个业务应用程序无法满足需求的情况下能自动化地组合各种己有的业务应用系统,并使得它们可以有机地协同工作来为用户提供服务。Web服务概念以及web服务体系结构的提出使得服务提供者可以将自己的软件以服务的形式提供给用户。同时,Web服务通过将不同层次的应用和资源抽象成统一的形式并提供标准的服务使用方法,为分布环境下资源共享与协同工作提供了很好的支持。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(QoS)确保的Web服务发现与组合方法己经成为Web服务领域研究的重点。目前,Web服务发现方法对QoS的支持不够,匹配算法缺乏灵活性;在Web服务功能流程中每个节点的服务选择如何确保全局服务质量和支持动态重计划方面还研究很少;在自动构建Web服务功能流程方面尚无相关研究。这几面的薄弱影响了Web服务的广泛应用,如何较为圆满的解决以上问题是一个崭新的研究课题。本文利用本体论、图论、遗传算法等学科的相关知识和方法来描述并求解Web服务领域的QoS问题,针对基于QoS的Web服务发现与自动组合方法进行了研究。本文主要的工作和成果如下:1.提出了一个基于QoS的Web服务发现与组合框架。本文在充分参考相关研究的基础上,提出了一个全面支持QoS的通用Web服务发现与组合框架。该框架将基于QoS的服务发现与组合过程分为接收用户请求、服务发现、功能流程设计、全局QoS服务选择、服务组合执行、组合结果评估等6个阶段,能够支持服务组合中功能流程的自动构建和QoS全局最优的Web服务自动选择。该框架的提出,为基于QoS的Web服务发现与组合研究起到了提纲挈领的指导作用。2.提出了一种新的支持QoS的Web服务发现方法(WSSM-Q)。针对目前服务发现匹配算法缺乏灵活性的问题,提出了一种新的支持QoS的Web服务发现方法(WSSM-Q)。该方法定义了一组描述Web服务QoS属性的分类tModel,用QoS量化值来描述服务提供者的QoS通告,用约束型需求和偏好型需求来描述服务消费者的QoS需求。定义了三个层次的服务匹配规则,引进匹配函数来计算服务相似程度,采用四阶段发现算法来选取满足消费者需求的Web服务。该方法不需要改变WSDL和UDDI的内部结构,能够直接在现有协议上进行实现。原型系统实验结果表明,该方法具有较高的服务查准率和足够的服务匹配效率,是可行的和有效的。3.提出了基于参数推导图自动构建Web服务功能流程的方法。针对目前大多数的Web服务功能流程都需要人工设计的问题,提出一种基于参数推导图自动构建Web服务功能流程的方法。该方法给出Web服务的谓词表示形式;通过建立参数推导图中的四种关系和构造算法,实现了Web服务组合的图形建模;然后定义参数推导图的完全关联矩阵,通过对关联矩阵的行实施合并运算,自动将参数推导图转换成以Web服务为结点,以服务间依赖关系为边的Web服务功能流程图。理论分析和实验结果证明了该方法的有效性和可行性。4.设计了支持QoS全局最优和动态重计划的Web服务选择算法(GODRP)。针对如何从大量具有相同功能与不同QoS特征的执行计划中选出最优方案,本文提出了一种支持QoS全局最优和动态重计划的Web服务选择算法(GODRP)。该算法使用位置矩阵编码方式来表示服务组合的所有执行路径和重计划信息;通过计算个体间服务质量的海明距离提高了服务组合的质量;通过指定用户总时间限制和实施优良解保留策略解决了算法运行时间对服务质量的影响问题。实验结果证明了算法的可行性和有效性。与同类成果相比,该算法提供了一种在动态环境下更完整和有效的服务组合QoS解决方案。
【Abstract】 With the advent and popularization of e-commerce, it is requisite that the distributed and heterogeneous business applications can be integrated and intemperate with each other in the environment of Internet. And it is also requisite that the appropriate business applications over the Internet can be rapidly and automatically found according to the demand and that they can be composed to meet the demand if there exist no individual business application to meet the demand. Web services technology is quite promising for the integration of business applications and service composition. Web services are more coarse-grained and operate at a higher level of abstraction.In the new service-oriented application environment, under the boost of requirements-driven, web Services discovery and composition with QoS-assured has turned into a primary research field in web services. In the past few years, many issues have been addressed. Recently web services discovery algorithm is lack of flexibility. In each node of function processes, there exist lots of web services with the same function and different QoS. To select a right one exactly with the global QoS-guaranteed, related research had some disadvantages. There have not some works on automatically constructing web services function flow. Weaknesses in the aspects hamper the wide deployment of web services. How to resolve these issues ideally is an open problem.This dissertation does a research on the discovery and automatic composition based on QoS by using of knowledge and methods of ontology, graph theory, genetic algorithm etc. The main points of our work are described as follows:1. A discovery and composition framework based on QoS is presented.In the framework discovery and composition based QoS is divided into six phases of receive of use requirement, services discovery, design of function flow, selection of aware-QoS services, executing of composition, evaluation of result. The framework can support automatically construction of the function flow and global optimal selection of the web services based on QoS.2. A QoS-aware method called WSSM-Q is present for web service discovery.To resolve the problem that web services discovery algorithm is lack of flexibility a QoS-aware method called WSSM-Q is present. A set of QoS categorization tModels is defined in the method to describe the QoS attributes of a web service. QoS advertisements of service providers are represented by QoS quantification values. QoS requirements of service consumers are represented by constraint and preference values. The matching rules classified three levels are defined and match functions is introduced to compute service similarity. Web services to meet the requirements of consumers are discovered by four phase algorithm. Because of no need for changes in the inner structure of WSDL and UDDI, it is easy to implement the method through the direct extension to existing protocols. The WSSM-Q is proved to be feasible and effective by high precision and efficiency of the prototype.3. A new method of automatically constructing web services composition flow based on PDG was presented.Based on the fact that current web services composition flow was designed by human, a new method of automatically constructing web services composition flow based on PDG (parameter deduce graph) was presented. Firstly the predicate form of web services was represented. Establishment of four relations in the PDG and the constructing algorithm was conduced so as to form the graphic models of web services composition. Then by definition of complete relevance matrix in the PDG and operation of combination of rows in relevance matrix, the PDG was converted automatically into SCFG (web services composition flow graph) that contains service-vertex and edges which indicate dependence relations among web services. Theoretical analysis and experimental results demonstrates the feasibility and efficiency of this method.4. A GODRP algorithm is presented for QoS-aware web services selection.In order to select a global optimal plan in a lot of execution plans with the different QoS, a GODRP (global optimal and dynamic replanning) algorithm is presented. A position matrix coding scheme was used to express simultaneously all of composite paths and replanning information in the algorithm. The quality of the services composition was improved by means of calculating Hamming Distance of QoS among individuals. It was solved that the algorithm’s executive time impairs the QoS of the services composition by prescribing a total time limit and implementing a fine solutions reservation strategy. Experimental results indicate the feasibility and efficiency of this algorithm. Comparison with similar studies, the algorithm provides a more integrated and effective QoS scheme of services composition in dynamic environment.
【Key words】 QoS(Quality of Service); Service Discovery; Service Composition; Parameter Deduce Graph; Genetic Algorithm;