【作者】 尹泽明；

【导师】 杨放春；

【作者基本信息】 北京邮电大学 ， 计算机应用， 2008， 博士

【摘要】 在融合网络为开放各种网络的业务能力提供可能和服务技术为软件复用问题提供很好的解决方案的背景下,电信业务提供领域更迫切地需要采用软件领域中的新科学方法和技术来改进其原有的业务提供方法与技术。随着Parlay等技术被成功地应用,基于“开放业务能力+协调框架”思想的架构被证实是融合网络环境下一种很好的电信业务提供架构。构件技术发展衍生出来的服务技术在继承了构件技术原有的解决软件复用的优点的基础上,为构建更加松耦合的系统提供了很好的支持。因此,采用服务来封装开放业务能力,以服务为元素进行组合来生成电信业务的方法被认为是一种灵活、有效的业务提供方法。而服务本身的合约性为基于服务组合方式提供满足一定程度下业务质量需求的业务提供了可能。目前关于服务组合技术已经展开了广泛的研究,本文立足于解决采用服务组合来提供电信业务的方法中有关业务质量的问题,进行了包括对业务提供系统架构、业务质量体系、服务发现、服务选择、服务组合执行中负载均衡这五个问题的研究。本文的结构安排如下,首先阐述了研究背景和意义并介绍了主要研究工作(第一章),然后在研究综述中分析了基于服务组合来提供电信业务的方式是网络融合背景下的趋势并概述了当前服务组合研究的现状并对其进行了归纳,从而指出基于服务组合、支持业务质量感知的业务提供方法的研究方向(第二章),接着提出了一种基于服务开放业务能力的业务提供系统架构,并基于服务的合约性构建了一种支持业务质量感知的增强架构(第三章),结合面向融合网络的角色模型和支持业务质量感知的服务组合架构,提出了一种面向基于服务组合提供业务的业务质量体系架构,和相应的业务质量指标体系及业务综合质量评价方法(第四章),针对该增强架构下如何有效发现可用服务实例的问题,提出一种基于语义的服务描述模型和基于此模型的服务匹配机制(第五章),针对开放网络环境下每个服务请求可能发现多个满足其功能需求的服务实例时如何选择服务实例来满足业务质量约束的问题,根据服务组合特性构建了单目标优化的服务选择模型,并通过改进的遗传算法来求解该模型(第六章),针对服务组合执行中的负载均衡问题,基于博弈论方法和相对负载测度方法提出了在服务叠加网中的有效的服务组合路由方法(第七章),最后对论文研究工作进行了总结,并提出一些进一步研究的可能方向(第八章)。论文的主要工作和创新归纳如下:(1)在现有SDP(Service Delivery Platform)模型基础上,引入SOA(Service OrientedArchitecture)的概念,构建了一个在融合网络环境下以服务来开放业务能力进而通过服务组合方式来提供电信业务的基本业务提供模型;在基本模型的基础上,通过引入服务合约模型进而构建出支持业务质量感知的增强模型。基于此增强架构,可以基于封装了来自各种网络的业务能力、符合现有服务规范要求的服务通过组合的方式来提供满足特定业务质量需求的业务。(2)为支持上述业务提供模型,结合面向下一代网络的业务提供角色模型,提出了相应的业务质量体系结构,并使用Z语言对其进行形式化规约并验证该体系结构的有效性(对层内元素及层间映射关系的验证),针对其业务层元素,提出了相应的业务质量指标体系和对业务质量进行综合评价的评价方法。(3)针对服务描述中语义信息不足的问题,通过应用领域本体实现了基于语义扩展的服务发现,在服务发现算法中进行了语义扩展,使得对服务语义的计算更加全面和准确;针对服务质量描述和匹配问题,本文在服务描述模型中加入了可扩展的服务质量描述,并给出了匹配计算的方法。(4)针对其它针对基于业务质量感知的服务选择问题研究的不足,提出了一种服务组合流程的表示方法,进而将服务选择全局优化问题建模为一个带有业务质量指标约束的组合优化问题,针对服务选择的特征来设计遗传算法的编码方法、适应度函数、交叉算子、变异算子、选择算子来执行基于组合流程信息的搜索,生成一组满足约束条件的服务组合流程集。在此遗传算法基础上,引入免疫学知识对该算法进行扩展来验证该遗传算法和给出改进方向。(5)针对分布式服务组合中服务路由和负载均衡问题,提出了一种自适应的分布式负载均衡算法。该算法基于非合作静态博弈方法和本地负载率测度来进行下一跳服务节点的选择,从而在服务叠加网中建立起一条合适的服务组合执行路径,达到负载均衡的效果以提高服务组合执行的性能。更多还原

【Abstract】 In the background that network convergence provides the possibility of opening serviceF capability of different networks and component technology provides good solution for software reuse, such software methodology and technology are urgently demanded to be imported into telecom service delivery to meet the requirements of providing rich and colorful serviceFs. With the widely use of Parlay, the "open serviceF capability plus coordination framework" based architecture is proved to be efficient for serviceF delivery in the convergence network environment. Service, derived from component, provides powerful support for constructing loose coupling system besides inheriting the component’s virtue for software reuse. Therefore, encapsulating serviceF capability with service and composing these services into serviceFs is thought of as an efficient and flexible method. The contract property of service assures that the requirement on serviceF quality could be met.Currently the research on service composition technology has been carried out widely. This dissertation aims to solve some serviceF quality problems existing in service delivery with service composition. The research includes the work on the architecture, serviceF quality, service discovery, service selection, and load balance in the execution of service composition.The structure of the dissertation is as follows. Firstly, research background and significance are introduced (Chapter 1). Secondly, the conclusion that telecom serviceF delivery with the service composition method in the network convergence background is a trend is presented, and the research points existing in service composition based, serviceF quality aware service delivery are pointed out (Chapter 2). Thirdly, a basic service delivery architecture is presented, and an enhanced architecture, which supports the serviceF quality awareness, is presented based on service’s contract property (Chapter 3). Integrating the role model of convergence network and the enhanced service delivery architecture, we present the architecture for quality factors of service delivery, together with the serviceF quality parameters and serviceF quality evaluation methods (Chapter 4). As for the problem about how to discover service instances suitable for service requests, a semantic based service description model and the corresponding service matching mechanism are presented (Chapter 5). As for the problem that how to select a service instance for a service in the service composition flow, we present a service selection model with a single optimization object, and design an genetic algorithm to solve optimization problem (Chapter 6). As for the load balance in the execution of service composition, the method based on game theory and relative metric is presented for efficient service routing (Chapter 7). Finally, the research work is summarized and some possible directions for future research are presented (Chapter 8).The main work and innovation are summarized as follows. (1) Based on the research on serviceF delivery, we import the concept of SOA to construct a basic serviceF delivery architecture based on service composition, in which the services encapsulate the concrete serivceF capabilities from different networks. Based on the basic architecture, we import contract model to get an enhanced architecture. According to the enhanced architecture, the services that observe the specification could be composed to provide the serviceFs with special quality requirements.(2) To support the service delivery architecture above, we present a layered serviceF quality architecture, referring the role model for NGN. Furthermore, we describe the architecture with formal specification and verify its efficiency (for the elements within a layer and the map between layers). For the element in serviceF layer, we present the corresponding serviceF quality metrics and evaluation methods.(3) As for the lack of semantic information in service description, service discovery is executed via domain ontology. Expanded semantic is presented for service discovery algorithm, which assures the precision of calculating the semantics of services. Based on a scalable service description method, a service matching algorithm is presented to implement the service discovery.(4) As for the disadvantages in other research for serviceF quality aware service selection, we present a denotation method for service composition flow, based on which we model the service selection problem as an optimization with serviceF quality constraints. We present a genetic algorithm whose coding mechanism, fitness function, crossover operator, aberrance operator, selection operator are all designed based on service composition flow, and the algorithm would output a set of service composition satisfying the constraints. Besides the genetic algorithms, we import the immune knowledge to expand and verify the enhanced genetic algorithm, pointing out the improved direction at the same time.(5) As for the service routing and load balance problem, we present a self-adaptive and distributed load balance algorithm. The algorithm is based on (1) non-cooperation static game theory method and (2) local load metric to select the next drop service node. Such a algorithm would assure a suitable service composition execution path would be set up, with which to elevate the performance of executing the service composition.更多还原