面向客户个性化需求的服务网络定制方法

发布时间：2018-04-24 02:33

本文选题：服务网络 + 个性化需求　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：随着云计算、物联网等技术的发展，互联网上可用的软件服务越来越多，用户需求可以通过服务及其之间的互联加以完成。同时，随着用户群体规模的增大，他们的需求多种多样。如何通过对服务进行组合以满足个性化需求，是当前服务计算领域的研究问题之一。针对该问题，当前研究主要有两类方法：大规模定制和服务组合。传统的大规模定制方法（如软件即服务SaaS）通过修改配置进行灵活定制，但这类服务定制系统在本质上是封闭的，服务能力和范围有限。传统的服务组合方法主要解决单需求的定制问题，可以实现用户个性化定制，但面对大规模需求时的成本较高。为了解决这些不足，本文提出基于服务网络进行用户个性化需求满足的方法。服务网络(Service Network，简称SN)试图将这些形式各样、分布在互联网不同位置上的服务以特定协议连接起来，组成一个开放的、包含各类可定制特征的网络。当用户需求抵达时，对服务网络进行定制，找到满足用户个性化需求的最优方案（即传统服务组合研究中的“组合服务”）。本文的前提是具有一个已构建好的服务网络，在其基础上围绕服务网络的优化定制问题开展以下方面的研究： (1)给出了服务网络的概念模型，，建立了相应的数学描述，阐述了用户个性化需求的表达方式，并给出了服务网络及其定制过程的实际案例。 (2)提出面向单需求的服务网络定制方法。首先建立了该问题的数学模型，其优化目标是在满足用户功能与QoS需求的前提下追求使用成本的最小化。使用人工蜂群算法（ABC）进行求解，详细给出了食物源编码规则、解的优化过程、解的适应度计算的设计方案，通过四个实验证实了该方法的有效性，并分析了影响服务网络定制性能的若干因素。 (3)面向并发需求的服务网络定制方法。考虑了并发需求的特点，将定制成本分解为构造成本和使用成本两类，尽可能提高选中服务的复用度，从而以最少的总成本在满足并发需求。提出了MR-SNC算法对该问题进行求解，并与不考虑复用的SR-SNC算法进行对比，实验证实MR-SNC可有效降低成本。 (4)基于保险机制的服务网络定制方案可靠性增强方法。前两个算法得到的定制结果不包含任何冗余，但现实中用户往往对可靠性有较严格的需求，需要在定制结果中保留某些冗余才能满足。提出了提高服务可靠性的三种方法（并行冗余、串行冗余和路径冗余）。针对已有定制方案和尚未存在定制方案两种情况，分别使用启发式算法和ABC算法进行求解，核心是在定制方案中寻找最优的冗余方案（并行冗余、串行冗余和路径冗余），进而通过实验证实了算法的有效性。最后，设计和实现了一个支持服务网络部署、定制和执行的原型系统，介绍了该系统的体系结构、主要模块和工作流程。
[Abstract]:With the development of cloud computing, Internet of things and other technologies, more and more software services are available on the Internet. At the same time, as the size of the user community increases, their needs are diverse. How to compose services to meet individual requirements is one of the current research problems in service computing field. In order to solve this problem, there are two main methods: mass customization and service composition. Traditional mass customization methods (such as software as a service SaaS) can be customized flexibly by modifying the configuration, but this kind of service customization system is closed in nature, and its service capability and scope are limited. The traditional service composition method mainly solves the customization problem of single demand, which can realize the customization of users, but the cost is high when facing the mass demand. In order to solve these problems, this paper proposes a method based on service network to satisfy users' personalized requirements. Service Network (SNN) tries to connect these kinds of services distributed in different locations of the Internet with specific protocols to form an open network with various customizable features. When the user's requirements arrive, the service network is customized to find the best solution to meet the user's personalized needs (that is, "composite services" in the traditional service composition research. The premise of this paper is to have a well-constructed service network, on the basis of which the following research is carried out around the optimization and customization of the service network: 1) the conceptual model of service network is given, the corresponding mathematical description is established, the expression of user's individualized requirements is expounded, and the actual cases of service network and its customization process are given. A service network customization method based on single demand is proposed. Firstly, the mathematical model of the problem is established, and its optimization goal is to minimize the cost of use under the premise of satisfying the user's function and QoS requirements. The artificial bee colony algorithm (ABC) is used to solve the problem. The coding rules of food source, the optimization process of the solution and the design scheme of the fitness calculation of the solution are given in detail. The validity of the method is verified by four experiments. Some factors affecting the performance of service network customization are analyzed. A service network customization method for concurrent requirements. Considering the characteristics of concurrent requirements, the customization cost is decomposed into two categories: construction cost and usage cost, so as to improve the reuse of selected services as much as possible, so as to satisfy the concurrent requirements with the minimum total cost. The MR-SNC algorithm is proposed to solve the problem and compared with the SR-SNC algorithm without multiplexing. The experimental results show that MR-SNC can effectively reduce the cost. 4) the reliability enhancement method of service network customization scheme based on insurance mechanism. The results obtained from the first two algorithms do not contain any redundancy, but in reality, users often have strict requirements for reliability, and some redundancies should be retained in the customization results to be satisfied. This paper presents three methods to improve the reliability of service: parallel redundancy, serial redundancy and path redundancy. The heuristic algorithm and ABC algorithm are used to solve the existing customization scheme and the existing customization scheme. The core is to find the optimal redundancy scheme (parallel redundancy) in the customization scheme. Serial redundancy and path redundancy are used to verify the effectiveness of the algorithm. Finally, a prototype system supporting service network deployment, customization and execution is designed and implemented. The architecture, main modules and workflow of the system are introduced.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【相似文献】