考虑费用优化的Web服务组合可靠性分配研究

发布时间：2018-11-04 11:46

【摘要】：随着Web服务技术的发展,其服务质量(Quality of Service, QoS)也备受关注。Web服务的QoS指服务的响应时间、价格、可用性、可靠性、信誉等非功能属性。而可靠性和价格作为服务质量的重要属性,且可靠性是服务消费者选择Web服务的一个重要依据,价格则是服务提供者商需要考虑的首要因素。目前国内外学者在这方面做了一定的研究,但仍然存在一些不足之处,如只提高Web服务的可靠性而对费用不加以控制。针对这种现状,本文研究了Web服务可靠性和费用的一些关键问题,包括在Web服务组合中可靠性的量化、可靠性指标约束分配给各组件服务、费用的最小化等方法。研究的主要内容是将Web服务组合的可靠性指标合理分配给各组件服务,并将费用控制到最低,以提供一个高可靠性且低费用的Web服务。本文提出了两种方法来实现,方法一是从失效率的角度出发,首先在基于结构化流程的基础上总结了Web服务的四种组合模式,即顺序(Sequence)、循环(Loop)、选择(XOR)、并行(AND)四种组合模式,接着分析了四种组合模式的失效模式,结合组合结构的失效模式特征与故障树模型语义,给出了将四种组合模式映射为故障树单元结构的方法,然后将整个Web服务组合流程映射成为一个完整的故障树模型,再利用故障树分析法,建立了Web服务组合的失效模型,基于组件服务的失效率和费用之间的关系函数和组合服务失效模型,建立了费用最小化的可靠性优化分配模型,最后利用非线性规划对模型进行求解。方法二是从可靠性的角度出发,也建立了可靠性分配优化模型,再利用遗传算法对模型求解。两种方法都能够在Web服务组合可靠性约束得到保证的前提下,将Web服务组合的可靠性指标合理分配给组件服务,并把成本控制到最低。最后,都通过对比实验证明了方法的有效性和实用性。
[Abstract]:With the development of Web service technology, the quality of service (Quality of Service, QoS) has attracted much attention. The QoS of Web services refers to the non-functional attributes such as response time, price, availability, reliability, reputation and so on. Reliability and price are important attributes of service quality, and reliability is an important basis for service consumers to choose Web services. Price is the primary factor that service providers need to consider. At present, scholars at home and abroad have done some research in this area, but there are still some shortcomings, such as improving the reliability of Web services and not controlling the cost. In view of this situation, this paper studies some key issues of Web service reliability and cost, including quantization of reliability in Web service composition, allocation of reliability index constraints to component services, cost minimization and so on. The main content of the research is to distribute the reliability index of Web service composition to each component service reasonably, and to control the cost to the lowest, in order to provide a Web service with high reliability and low cost. In this paper, two methods are put forward to implement. First, from the point of view of failure rate, four kinds of Web service composition modes are summarized based on structured process, that is, sequential (Sequence), cyclic (Loop), selects (XOR),. After analyzing the failure modes of the four combined patterns, combining the failure mode characteristics of the composite structure and the semantics of the fault tree model, the method of mapping the four combined patterns to the fault tree unit structure is given. Then the whole Web service composition process is mapped into a complete fault tree model, and then the failure model of Web service composition is established by using the fault tree analysis method. Based on the relationship function between failure rate and cost of component services and the failure model of composite services, the reliability optimal allocation model of cost minimization is established. Finally, nonlinear programming is used to solve the model. Secondly, from the point of view of reliability, the reliability allocation optimization model is also established, and then the genetic algorithm is used to solve the model. Both methods can reasonably assign the reliability index of Web service composition to component services and keep the cost to the minimum on the premise that the Web service composition reliability constraints are guaranteed. Finally, the effectiveness and practicability of the method are proved by comparative experiments.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】