基于物理规划的服务组合方法研究

发布时间：2018-07-12 22:10

本文选题：服务组合 + 物理规划　；参考：《华东理工大学》2014年博士论文

【摘要】：实际应用系统在设计和运行等各个阶段都面临着需求变更的问题,尽管可以通过对既有系统进行改进或替换来应对,但往往会耗费用户非常大的人力和财力。为了实现对既有系统的良好继承,节省用户开支,面向服务架构(Service-Oriented Architecture, SOA)的设计思想应运而出。SOA将不同的应用封装成服务,服务之间通过定义良好的接口联系,并可以按照用户需求和服务契约组合成新的综合应用。除了传统的大型企业软件应用,物联网、泛在网、云计算等新应用的推进也需要SOA技术的支持,并给SOA带来了一些新的问题和技术挑战,包括服务组合建模、服务发现、服务选择优化等。本文围绕SOA中的服务组合问题进行研究,考虑服务组合中的用户偏好问题,将一种改进的基于权重函数的物理规划用于服务组合建模；通过对移动Ad hoc网络特点进行分析,提出了一种负载均衡的服务发现协议,用于发现搭建组合服务所需的候选服务；同时,针对服务选择优化问题,提出了基于交叉熵的启发式算法。在此基础上,通过对异构终端和网关的研究开发,搭建了面向服务组合的多终端协同实验平台,验证了服务发现协议、服务组合建模、服务选择算法的可行性和有效性。主要完成的研究工作包括： 1.为使物理规划方法更加严格地遵守OVO准则,提出了一种基于权重函数的物理规划。该方法首先将待优化目标进行归一化处理,然后根据构造的权重函数为目标赋予动态权重,并采用加权和的方式构造综合目标函数。通过在求解最优解和Pareto解两类数学问题中的应用,验证了该方法的有效性及其对用户偏好的契合度。 2.针对MANET网络的动态性、资源受限性和临时群组性,结合服务组合对于多候选服务的需求,提出了一种基于局部最大独立子集缓存的服务发现协议。该协议遵循后应式的服务缓存策略,当网络中发生服务请求和服务应答信息时,节点对所转发的服务信息进行缓存。为了实现节点维护服务信息的负载均衡,基于局部最大独立子集方法选择缓存节点。通过仿真验证了服务发现协议的有效性,协议在响应时延、报文开销和缓存开销等方面取得了较好的均衡性。 3.针对服务组合建模中的用户偏好问题,采用权重函数物理规划的方法,将满意度和模糊重要度两种偏好信息融合在综合目标函数的设计中。通过该方法所得到的最优解在各目标之间更为均衡,且能够保证重要度较高的目标优先得到优化。同时,针对服务选择这一NP困难问题,提出了一种改进的启发式交叉熵算法。算法通过近似局部最优、删除并排序候选服务等策略缩小了问题规模,降低了计算复杂度,并通过仿真实验进行了验证。 4.在上述研究的基础上,搭建了面向服务组合的多终端协同示范平台,对异构网络互连互通、异构终端服务发现、基于用户偏好的服务组合等内容进行了研究和应用,为本文研究的关键技术提供了可行性验证。
[Abstract]:The actual application system is faced with the problem of demand change at all stages of design and operation. Although it can be improved or replaced by the existing system, it often consumes very large human and financial resources. In order to achieve good inheritance of the existing system, save the user expenditure, and the service oriented architecture (Service-Oriented The design idea of Architecture, SOA) should be shipped out of.SOA to encapsulate different applications into services, services are defined by good interface connections, and can be combined into new and integrated applications according to user needs and service contracts. In addition to the traditional large enterprise software applications, the advance of new applications such as wk, pan, cloud computing and other applications requires S The support of OA technology brings some new problems and technical challenges to SOA, including service composition modeling, service discovery, service selection optimization, and so on.
This paper focuses on the problem of service composition in SOA and considers the user preference in the service composition. A improved physical programming based on weight function is used in the modeling of service composition. By analyzing the characteristics of the mobile Ad hoc network, a load balancing service discovery protocol is proposed to find a combination service. On the basis of the research and development of heterogeneous terminals and gateways, a multi terminal cooperative experiment platform for service combination is built on the basis of the research and development of heterogeneous terminals and gateways, which validates the feasibility of service occurrence protocol, service composition modeling and service selection algorithm. Sex and effectiveness. The main research work includes:
1. in order to make the physical programming method more strictly observant of the OVO criterion, a physical programming based on weight function is proposed. First, the target is normalized, then the dynamic weight is given according to the weight function of the structure, and the weighted sum is used to construct the comprehensive objective function. And the application of Pareto to solve two kinds of mathematical problems, verifies the validity of the method and its fit for user preferences.
2. in view of the dynamic, resource constrained and temporary groups of MANET networks, a service discovery protocol based on the local maximum independent subset cache is proposed, which is based on the local maximum independent subset cache. The protocol follows the post response service cache strategy. When the service requests and service response information occur in the network, the node is the same The service information is cached. In order to achieve the load balance of the node maintenance service information, the cache node is selected based on the local maximum independent subset method. The effectiveness of the service discovery protocol is verified by simulation. The protocol has achieved good equilibrium in response delay, message overhead and cache overhead.
3. aiming at the problem of user preference in the modeling of service composition, the method of weight function physical programming is adopted to integrate the two kinds of preference information of satisfaction and fuzzy importance in the design of comprehensive objective function. The optimal solution obtained through this method is more balanced among the targets, and it can ensure that the target with higher importance is obtained first. At the same time, an improved heuristic cross entropy algorithm is proposed for the NP problem of service selection. The algorithm reduces the scale of the problem by approximating the local optimal, deleting and sorting candidate services, and reduces the computational complexity. The simulation experiment is verified by the simulation experiment.
4. on the basis of the above research, a multi terminal collaborative demonstration platform for service combination is built, which has been studied and applied to interconnect and interconnect heterogeneous networks, heterogeneous terminal service discovery, and service composition based on user preferences. It provides a feasibility verification for the key technologies studied in this paper.
【学位授予单位】：华东理工大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】