云媒体中基于效用最大化协商机制的资源分配算法研究

发布时间：2018-04-27 00:38

本文选题：云媒体 + 资源分配　；参考：《中国海洋大学》2014年硕士论文

【摘要】：云媒体是云计算与多媒体服务相结合而形成的一种新兴的服务模式。它是将各种形式，如电视、网络、手机等媒体信息聚合成一个媒体资源空间，进而构建一个超媒体传递、多样化服务的信息资源集成共享平台。云媒体可广泛应用在办公自动化、广电通信、电子商务、健康监护等领域，具有十分广阔的应用前景。云计算能够提供动态资源池、虚拟化服务的计算平台，具有强大的数据处理能力，因此，随着互联网技术的发展，大量媒体数据的应用系统不断涌现。本文研究如何利用云计算技术处理多媒体数据，从而更好的为云媒体用户提供更高质量的服务，具有重要的理论意义和应用价值。由于媒体本身的固有属性，使得原有的媒体资源分配和调度算法已无法应用在复杂的云媒体环境中。同时，提高云媒体的服务质量面临着巨大的挑战，包括异构资源统一管理、满足不同用户服务等级协议（SLA，Service Level Agreement）的合理云资源调度等。而如何降低云媒体数据中心能耗；提高系统资源利用率以及建立有效的资源管理模型，进而，提高云媒体服务质量和用户服务满意度，是云媒体资源分配中亟待解决的问题。本论文研究了云媒体资源分配中的主要问题，包括：最小化任务总执行时间，降低任务的处理成本，最大程度的满足用户需求，从而，实现满足云媒体服务性能的最优化云资源分配调度。论文首先研究了云媒体资源分配的相关技术，，结合云媒体环境中执行时间、任务成本、数据中心能耗和资源利用率等问题，使用效用计算模型，将上述问题转化为服务效用问题；从云媒体服务提供者和云媒体服务请求者出发，提出相应的效用函数模型；使用服务价格，服务响应时间和服务带宽等属性对该模型进行统一化描述；然后利用云媒体服务双方的效用函数值和让步机制，得到最大效用化的云媒体资源分配策略，最后提出基于协商机制的资源分配算法RAANM(Resource Allocation Algorithm based on Negotiation Mechanism)。本文提出的资源分配策略使得云媒体资源分配的目标函数不再是最小化最大完成时间，而是以达到效用值最大为目标，这样，可进一步提高用户的服务满意度。仿真实例验证了该算法的有效性。本文的创新点是，提出了云媒体服务的效用函数模型，结合效用函数、让步策略和相应的协商机制效用模型，提出了RAANM算法。
[Abstract]:Cloud media is a new service model formed by the combination of cloud computing and multimedia services. It aggregates all kinds of media information, such as TV, network, mobile phone and so on, into a media resource space, and then constructs a platform for information resources integration and sharing of hypermedia transmission and diversified services. Cloud media can be widely used in office automation, radio and television communications, electronic commerce, health monitoring and other fields, with a very broad application prospects. Cloud computing can provide dynamic resource pool, virtualization service computing platform, and has powerful data processing ability. Therefore, with the development of Internet technology, a large number of media data applications are emerging. This paper studies how to use cloud computing technology to deal with multimedia data, so as to provide a better service for cloud media users, which has important theoretical significance and application value. Due to the inherent properties of media itself, the original media resource allocation and scheduling algorithms can not be applied in the complex cloud media environment. At the same time, improving the quality of service of cloud media faces great challenges, including unified management of heterogeneous resources, reasonable scheduling of cloud resources to meet different user service level agreements (SLA-Service Level agreements) and so on. However, how to reduce the energy consumption of cloud media data centers, improve the utilization of system resources and establish an effective resource management model, and then improve the quality of cloud media services and customer service satisfaction, is an urgent problem in the allocation of cloud media resources. This paper studies the main problems in the allocation of cloud media resources, including: minimizing the total execution time of the task, reducing the processing cost of the task, and meeting the needs of the user to the greatest extent. The optimal cloud resource allocation scheduling can satisfy the performance of cloud media service. Firstly, the paper studies the related technologies of cloud media resource allocation, and uses utility calculation model, combining with the problems of execution time, task cost, data center energy consumption and resource utilization in cloud media environment. From the cloud media service provider and the cloud media service requester, the corresponding utility function model is put forward, and the service price is used. The model is described uniformly by the attributes of service response time and service bandwidth, and then the maximum utility allocation strategy of cloud media resources is obtained by using the utility function value and concession mechanism of both sides of cloud media services. Finally, a resource allocation algorithm based on negotiation mechanism, RAANM(Resource Allocation Algorithm based on Negotiation mechanics, is proposed. The resource allocation strategy proposed in this paper makes the objective function of resource allocation in cloud media not to minimize the maximum completion time, but to achieve the maximum utility value, which can further improve the service satisfaction of users. The effectiveness of the algorithm is verified by a simulation example. The innovation of this paper is that the utility function model of cloud media service is proposed, and the RAANM algorithm is proposed by combining utility function, concession strategy and corresponding negotiation mechanism utility model.
【学位授予单位】：中国海洋大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】