基于OpenFlow的视频流控制原型系统实现

发布时间：2018-04-18 08:50

本文选题：视频流 + 兴趣挖掘　；参考：《电子科技大学》2013年硕士论文

【摘要】：中国互联网数据平台的相关数据显示，网络视频是用户人均单日访问时间最长的应用，同时在线视频已成为互联网第一大应用。对于网络运营商而言，其主要收入来源于向用户销售使用时长，大量的视频流量并没有带来对称的收入增长，同时由于市场竞争加剧，整个产业资费水平下降，运营商亟需改变传统的盈利模式，由业务向服务的转变是关键。而对于视频服务提供商而言，，目前提供的都是免费视频，广告收入拿捏着他们发展的命脉，而广大、稳定的视频用户群体是其营收的保证。当前视频资源异常丰富，要想培养忠实的粉丝，具有吸引力的内容固然重要，如果能够准确根据用户兴趣主动推荐视频，将深深俘获视频用户的心，但若因为网络状况而影响到最终的视觉体验，将功亏一篑。另一方面，如果运营商的网络设备能够对业务进行感知，可以识别传输网络中的视频流并进行一些用户兴趣挖掘工作，为个性化服务提供一定的依据，差别服务及收费模式的问题就可以迎刃而解，视频服务提供商也可以订制更高服务质量的网络而提供更好的用户体验，所以由运营商提供增值服务，由视频服务提供商支付增值费用，可以说是一个双赢的局面。本论文正是针对上述问题，以P2P及Web视频流为研究对象，在深入分析当前视频流识别及兴趣挖掘技术原理的基础上，提出在网络汇聚层环境建立一个基于OpenFlow的视频流控制系统，主要内容为： 1．研究视频流识别及兴趣挖掘技术原理，接合本控制系统特点，分析OpenFlow技术架构原理及组成，确定了原型系统在OpenFlow架构下实现具有可行性。 2．对系统进行设计，主要功能是当网络视频流在传输过程中经过汇聚设备时，路由器根据一定的规则、算法对P2P、Web视频流进行视频识别，在此基础上提取用户所观看视频的相关信息，根据所得信息结合用户兴趣模型，对视频流用户进行兴趣挖掘，同时为对识别出的视频流进行优先级调度，设计了相应的计费功能。 3．在OPNET网络仿真平台下搭建测试环境，对系统进行测试并分析。
[Abstract]:According to the related data of Chinese Internet data platform, online video is the longest application per user in a single day, and online video has become the largest application of the Internet.For network operators, their main revenue comes from the length of time spent selling to users, and a large amount of video traffic does not bring symmetrical revenue growth. Meanwhile, due to the intensification of market competition, the entire industry tariff level drops.Operators need to change the traditional profit model, from business to service is the key.For video service providers, all they offer is free video, advertising revenue is their lifeblood, and a large, stable video user base is its revenue guarantee.At present, video resources are extremely rich. If you want to cultivate loyal fans, it is important to have attractive content. If you can accurately and actively recommend videos according to users' interests, it will capture the hearts of video users.But if because of the network situation and affect the final visual experience, will fall short.On the other hand, if the network equipment of the operator can perceive the service, it can identify the video stream in the transmission network and do some work of user interest mining, which can provide some basis for the personalized service.The problem of different services and charging modes can be easily solved. Video service providers can also customize networks with higher service quality to provide a better user experience, so operators can provide value-added services.It is a double-win situation that the video service provider pays the value-added fee.Aiming at the above problems, this paper takes P2P and Web video stream as the research object, on the basis of deeply analyzing the technology principle of current video stream identification and interest mining, this paper proposes to establish a video flow control system based on OpenFlow in the network convergence layer environment.The main contents are:1.This paper studies the principle of video stream recognition and interest mining, integrates the characteristics of this control system, analyzes the principle and composition of OpenFlow technology architecture, and determines the feasibility of implementing the prototype system under the OpenFlow framework.2.The main function of the system is that when the network video stream passes through the convergent device in the course of transmission, the router carries out video recognition of P2PNs Web video stream according to certain rules.On this basis, the relevant information of the video viewed by the user is extracted. According to the information obtained and the interest model of the user, the interest mining of the video stream user is carried out, and the priority scheduling of the identified video stream is carried out at the same time.The corresponding billing function is designed.3.The test environment is built on the OPNET network simulation platform, and the system is tested and analyzed.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP273;TN919.8

【参考文献】