Research On QoE Evaluation Of Mobile Streaming Media Considering Business Domain

With the rapid development of the operation network and the generation of the era of 5G,mobile streaming media services plays an increasingly important role in our daily life.For those service providers,an accurate evaluation of user experience for streaming media services is necessary to provide higher competitiveness in the fierce market.The Current research on Quality of Experience(QoE,Quality of Experience)evaluation methods mainly focus on the influential factors in the technology domain and the user domain of without the consideration of the factors in business domain factors.Meanwhile,those user-centric QoE evaluation methods highly depend on Personalized user experience quality that those input factors of evaluation algorithm are user’s subjective feedbacks,which cause the negative performance on real-time evaluation model.In order to solve above problems,the study proposes a QoE evaluation model which considering business domain that predict user preference with video popularity so that the QoE evaluation model can be calculated subjectively with objective indicators.Then an experimental platform of online streaming media service is implemented to get validation of the QoE evaluation model.Two kinds of study are done in this paper:1.The combination of factors in business domain with QoE evaluation model.A hypothesis is first proposed that the popularity and user preference have the mapping relationship based on the herd effect.The verification of the hypothesis is developed by quantitative analysis of information entropy.Then the K-means clustering is used to determine the binary video popularity degree with the user diversity,which results in a prediction model between video popularity to user preference in the range of high popularity videos.Then the QoE evaluation model is established with the data collected from experimental platform with video popularity,video bitrate and user preference and the prediction precision is 0.82.The proposed QoE evaluation can fill the gap of less consideration of business domain of current works and the model supports the direct prediction on QoE with objective indicatiors of video popularity under the condition of high popularity.Those traditional QoE evalution which rely on user subjective feedbacks can be substituted so that the real-time capability of evaluation model can be improved.2.The construction of the experimental platform.The experimental system is based on the structure of Nginx-Linux-MySQL-PHP,supporting HTTP Live Streaming techonology and the collection of required input and outputdata for QoE evaluation.Also with the optimization of experimental process,a dynamic experiment task list is implemented.According to the analysis of data collected in experimental platform,the hypothesis that users are convergent to the group preference in the condition of high popularity because of the herd effect can be verified.And then the fact that in the condition of high popularity,the popularity can be used to predict individual preference which is used in QoE estimation model considering bitrates and video popularity can be verified.