云计算环境下服务组合技术研究

发布时间：2018-04-25 21:12

本文选题：云计算 + 可信服务　；参考：《南京航空航天大学》2016年博士论文

【摘要】：面向服务的计算模式能够将功能单一的服务进行组合,形成新的增值服务来满足用户的复杂需求。随着云计算的普及,服务组合也面临新的问题。开放动态的云环境给服务的评估带来了更多的不确定性,而海量的云服务也对现有服务组合技术的有效性和实时性提出了新的挑战。随着云服务的多样化和规模化,如何从众多的候选服务中快速选择出满足用户要求的服务进行组合是目前服务计算领域的研究热点之一。针对上述问题,本文在分析现有服务评估和组合技术的基础上,从云服务的可信评估、提高服务组合效率、以及复杂需求下的服务组合等方面开展了系统而深入的研究。本文的主要创新工作概括如下:(1)提出了一种基于一致性强度的云服务可信评估方法。在云服务评估框架下为基础设施服务和应用服务分别设计了可信评估指标;并提出了基于一致性强度的模糊评估方法,通过引入语义折扣因子和一致性强度,可从模糊的服务评价信息中合理地分析出确定的评估值,从而解决了不确定环境下云服务的可信评估问题,并在仿真实验平台NetLogo上验证了所提出方法的实用性和有效性。(2)提出了一种基于人工蜂群算法的可信服务组合方法。在服务组合模型中引入时间衰减函数,提高了近期评分值的时间权重,从而使服务质量更符合当前时刻服务的特征;将服务组合问题进行非线性整数规划建模,提出了全局指导人工蜂群(DGABC)算法来求解该模型,通过蜂群对食物源的探索来实现对最佳服务组合方案的搜索。实验结果表明在大量服务信息下,本文提出的DGABC算法可以在保证服务质量的同时提高服务组合的效率。(3)提出了一种基于成本效益优化的多目标服务组合方法。根据用户对服务组合的多样化需求,以最大化服务质量、最小化成本为优化目标,将服务组合问题建模为多目标整数规划模型;提出了基于精英指导的多目标人工蜂群算法(EMOABC),在原始人工蜂群算法中加入快速非支配排序、种群选择、精英指导离散解生成、以及多目标适应度计算等多目标优化策略,实现对该模型的求解。相较于同类算法,EMOABC算法无论在运行效率还是求解质量上均具有明显优势,从而验证了本文提出的方法可较好地解决复杂需求下的云服务组合问题。(4)提出了一种基于skyline计算的非线性服务组合方法。通过skyline计算筛选掉每个服务群中的冗余服务,可以减少搜索空间,有效提高服务组合的效率;进而采用建模语言AMPL将服务组合问题建模为0-1非线性规划问题,并利用求解器Bonmin对所建立的模型进行求解。实验结果表明本文提出的基于skyline计算的服务组合方法在保证服务质量的同时可显著提高服务组合的效率。
[Abstract]:The service-oriented computing model can combine the services with a single function to form a new value-added service to meet the complex needs of users. With the popularity of cloud computing, service composition also faces new problems. The open and dynamic cloud environment brings more uncertainty to the evaluation of services, and the massive cloud services also pose a new challenge to the effectiveness and real-time performance of the existing service composition technology. With the diversification and scale of cloud services, it is one of the research hotspots in the field of service computing that how to quickly select the services that meet the needs of users from many candidate services. In order to solve the above problems, based on the analysis of the existing service evaluation and composition techniques, this paper has carried out a systematic and in-depth study on the trusted evaluation of cloud services, improving the efficiency of service composition, and service composition under complex requirements. The main innovation work of this paper is summarized as follows: (1) A cloud service trust evaluation method based on consistency strength is proposed. In the framework of cloud service evaluation, the trusted evaluation indexes for infrastructure services and application services are designed, and a fuzzy evaluation method based on consistency strength is proposed, which introduces semantic discount factor and consistency strength. We can reasonably analyze the definite evaluation value from the fuzzy service evaluation information, and solve the problem of cloud service credible evaluation in uncertain environment. The practicability and validity of the proposed method are verified on the simulation platform NetLogo) and a trusted service composition method based on artificial bee colony algorithm is proposed. The time attenuation function is introduced into the service composition model, which improves the time weight of the recent scoring value, thus making the quality of service more consistent with the characteristics of the service at present time, and the service composition problem is modeled by nonlinear integer programming. A global directed artificial bee colony (DGABC) algorithm is proposed to solve the model, and the best service composition scheme is searched by the colony searching for food source. The experimental results show that the proposed DGABC algorithm can improve the efficiency of service composition while ensuring the quality of service under a large amount of service information. A multi-objective service composition method based on cost-benefit optimization is proposed. In order to maximize the quality of service and minimize the cost, the service composition problem is modeled as a multi-objective integer programming model. This paper presents a multi-objective artificial bee colony algorithm based on elitist guidance, which includes fast undominated sorting, population selection, elitist directed discrete solution generation and multi-objective fitness calculation. The solution of the model is realized. Compared with the similar algorithm, EMOABC algorithm has obvious advantages in terms of running efficiency and solving quality. It is verified that the proposed method can solve the cloud service composition problem with complex requirements. (4) A nonlinear service composition method based on skyline computation is proposed. The redundant services in each service group can be filtered by skyline calculation, the search space can be reduced and the efficiency of service composition can be improved, and then the service composition problem can be modeled as a 0-1 nonlinear programming problem by using the modeling language AMPL. The model is solved by the solver Bonmin. The experimental results show that the proposed service composition method based on skyline computing can significantly improve the efficiency of service composition while ensuring the quality of service.
【学位授予单位】：南京航空航天大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP393.09

【相似文献】