IPTV用户QoE评测系统设计与实现

发布时间：2018-06-15 05:33

本文选题：用户体验质量 + 用户观看习惯　；参考：《南京邮电大学》2017年硕士论文

【摘要】：计算机网络快速发展的当下,多样性的视频服务源源不断的涌现,其中IPTV作为一个广受关注的业务,被推向了这个技术时代的尖端。在IPTV业务快速发展的同时,随之而来的是用户对于服务质量的要求。目前的发展态势要求服务提供商们能够主动评价并预测用户对IPTV服务的满意度,从而能够及时改善不良的用户体验。然而传统的服务质量(QoS)监测体系无法满足要求,需要引入体验质量(QoE)来描述用户感知状况。因此,QoE已经成为当下的研究热点,寻找提升QoE的解决方案和建立QoE评估模型的需求也越来越迫切。基于此,本论文主要在用户主观指标及主观评价方法改进、算法选择与改进、用户QoE评测系统的设计与实现三方面开展研究工作,具体内容如下:首先,基于机顶盒采集到的用户观看记录数据集,从用户主观角度出发,提出用户观看习惯指标,此指标反映用户在观看节目时的兴趣或爱好。针对传统主观评价方法复杂,成本高,非实时的缺点,根据用户行为指标来改进主观评价方法,将用户观看时间比指标映射为MOS值,实现了QoE的量化。其次,针对所分析的数据特点和实际场景任务的需求,首先比较多种经典算法,包括回归、k近邻(kNN)和分类与回归树(CART),结果表明CART算法的预测准确度更高。基于此,为了进一步提高模型的预测准确度,对CART算法进行了改进。即,采用核函数的方法改进CART算法的输出。结果表明,本论文所提出的CART改进算法可以有效地提高QoE模型的预测准确度。最后,基于大数据平台,针对机顶盒数据集,进行分布式数据存储、处理和分析,并综合本论文的评价模型,设计和实现可视化的用户QoE评测系统。此系统有利于服务提供商及时监测服务质量,优化网络环境,提升用户体验,推广品牌价值,也有利于用户及时了解服务以及自己对服务的满意程度和观看兴趣,并可针对服务存在的不足提出自己的建议。
[Abstract]:At the moment of the rapid development of the computer network, the diversity of video service has come to a steady stream. As a widely concerned business, IPTV has been pushed to the tip of the technical age. At the same time, the rapid development of IPTV service is the requirement of the service quality of the users. The current development trend requires service providers It is able to actively evaluate and predict users' satisfaction with IPTV services so that the bad user experience can be improved in time. However, the traditional quality of service (QoS) monitoring system can not meet the requirements and needs to introduce the experience quality (QoE) to describe the user perception. Therefore, QoE has become the focus of the present research, looking for solutions to improve the QoE. And the requirement of establishing QoE evaluation model is becoming more and more urgent. Based on this, this paper mainly focuses on the improvement of user subjective index and subjective evaluation method, algorithm selection and improvement, the design and implementation of user QoE evaluation system in three aspects. On the subjective angle of the user, the user's viewing habits are put forward. This index reflects the interest or hobby of the user when watching the program. The subjective evaluation method is complex, high cost and non real time. According to the user behavior index, the subjective evaluation method is improved, and the user viewing time is mapped to the MOS value, and the quantization of QoE is realized. Secondly, in view of the analysis of the characteristics of the data and the needs of the actual scene task, first compare a variety of classical algorithms, including regression, k nearest neighbor (kNN) and classification and regression tree (CART), the results show that the prediction accuracy of the CART algorithm is higher. Based on this, in order to further improve the prediction accuracy of the model, the CART algorithm is improved. That is, the kernel is adopted. The method of function improves the output of the CART algorithm. The results show that the CART improved algorithm proposed in this paper can effectively improve the prediction accuracy of the QoE model. Finally, based on the large data platform, the distributed data storage, processing and analysis are carried out for the set of set-top box data sets, and the evaluation model of this paper is integrated, and the visualization is designed and implemented. User QoE evaluation system. This system helps service providers to monitor service quality in time, optimize network environment, improve user experience, promote brand value, also help users to understand service as well as their satisfaction and interest in service, and put forward their own suggestions on the shortage of service.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN949.292

【参考文献】