Research On Statistical Mean Opinion Score Of Network Video Service And Its Application To Traffic Classification

With the widespread use of network video services and the development of wireless network technology,people's demand for video services is continuous increasing.How to predict and guarantee the users' quality of experience(QoE)is a urgent problem to be solved by Internet Service Providers.On one hand,the limited resources of the network and the bit error rate of the channel will affect the quality of network video transimission.The traditional quality of service(QoS)can't directly reflect the user's true feelings of video service.On the other hand,in order to better control and manage the network video traffic,how to find efficient and simple combination of statistical features of the network video traffic for effective classification has been the focus of research.The main contents of this thesis are as follows:Firstly,this thesis introduces the theory of subjective quality assessment and objective quality assessment,the definition of QoE,and some existing QoE evaluation methods for video services.Then,it introduces the objective quality assessment model proposed by ITU-T P.1201 standard and vector quantization algorithm.According to the ITU-T P.1201 QoE prediction model,this thesis analyzes the influence of video bitrate,video average playback interruption time,video play interruption frequency and initial buffering time to users' QoE through vector quantization algorithm.According to the above QoS parameters,the Probability Density Function(PDF)of three kinds of video services is studied in this thesis.Finally,according to the Mean Opinion Score(MOS)values of three kinds of quality video,we propose a new traffic classification method based on the PDF mean value of video MOS.This method studies the classification of network video with MOS related characteristics for the first time.This method takes into account the MOS mean value of the video services and the downlink byte rate selected based on the CON-GR(Consistency Feature Selection-Gain Ratio)feature selection method.Support Vector Machine(SVM)classifier is used to implement the classification of three types of video services.Experimental results show that the proposed method can achieve higher classification accuracy compared with existing video stream classification methods.