【作者】 张跃；

【导师】 陈廷伟；

【作者基本信息】 辽宁大学 ， 计算机应用技术， 2011， 硕士

【摘要】 由于Web服务处于动态的网络环境当中,受网络的开放性以及各服务的自治性等因素的影响,容易发生服务失效和组合服务无法执行的问题。当一个组合服务中出现原子服务失效的状况时,需要及时发现失效的服务并寻找服务予以替换,从而保证组合服务系统的执行。这就要求在选择替换服务时选择一个能够尽可能长时间不会发生失效的服务,而一个服务是否容易发生失效可以通过这个服务的可信度(包括可靠性、信誉度和执行效率等相关Qos属性)反映。为了能够在维护组合服务运行时选择出不易发生服务失效的服务进行替换以维护组合服务的执行和长期有效性,本文基于服务可信度展开对组合服务的动态自适应维护研究。本文主要工作如下:首先,为了实现基于可信度的组合服务自适应维护,在相关基础上提出了Web服务是否容易发生服务失效状况的度量标准。由于单纯的服务可靠性或者服务信誉度无法完备地对服务是否容易发生失效进行刻画,本文采用服务的可靠度、信誉度和执行效率三个属性共同来刻画服务的可信度,并利用不确定因素修正后再叠加的计算方法得到客观的服务可信度,为选择替换服务提供依据,在维护组合服务运行过程中有效防止“二次失效”现象的发生。其次,在反馈控制思想基础上设计了基于可信度的服务自适应维护体系结构,实现服务的监测、预测和发现功能,为可信度的计算和组合服务的维护提供必要的Web服务信息,并支持动态网络环境下组合服务的动态自适应维护。然后,将基于可信度的组合服务自适应维护过程转化为马尔可夫决策(MDP)过程,在MDP框架下设计了服务替换算法和自适应维护算法,实现组合服务的自适应维护,组合服务系统可有效降低服务响应时间和降低系统维护频率。最后,通过仿真实验,验证了本文方法的可行性和有效性。更多还原

【Abstract】 As the web services are in the dynamic network environment, so they are not absolutely identified, Given the impact of network’s openness, platform’s heterogeneity, communication asynchronous and component‘s autonomy, it’s very easy to occur service failure problem. When some atom service failure in a composite services, it needs perceive the failed service timely and look for new service to replace it, thus ensuring the execution of composite services. This requires to selecting a substitute service that won’t have the failure problem, and a service whether prone to failure can reflect by the service creditability (Including reliability/credibility and executing efficiency and other related Qos attributes). In order to choose a service that not prone to failure in the process of composition services maintenance to maintain the composition services can execute and have long-term effectiveness. This paper will do research on dynamic adaptive maintenance of the composite services maintenance based on the service creditability. The main work is as below:In order to realize the adaptive maintenance of composite services based on service creditability. The paper defines clearly the service creditability that it is measure show the degree the Web service prone to failure at first. As simple service reliability or credibility cannot complete measure and describe the failure tendency, so the paper adopts service reliability, credibility and processing efficiency together to describe service creditability and calculate creditability using the approach of the superposition of the above uncertain factor to get the objective service creditability value, and provide proof for choosing replacement service. Also, it can avoid service“failed again”phenomena in the process of replacement.Next, this paper designs a service creditability adaptive maintenance structure based on feedback control idea and service creditability,to realize service monitor, predict and found function, and provide requisite Web service message for the maintenance of the composite service,and support the dynamic maintenance of composite services. After that, take the dynamic maintenance of composite services based on service creditability as a MDP. Under the MDP construct, we design the service replacement algorithm and adaptive maintenance algorithm, to realize the feedback controller’s function, and finally realize the maintenance of composite services. The algorithm can effective lower the service response time and the system maintenance frequency.Finally, through the simulation experiments, proved the feasibility and effectiveness of the method in this paper.更多还原