Design And Implementation Of User's QoE Evaluation System For IPTV

Nowadays with the rapid development of computer networks,a variety of video services continue to emerge.As a concern of the business,IPTV has been pushed to the forefront of this era.With the rapid development of IPTV services at the same time,what followed is the quality requirement by the users.The development trend requires service providers to actively evaluate and predict the user's satisfaction with IPTV services,so as to improve the defective user experience.However,the traditional Quality of Service(QoS)monitoring can not meet this requirement.Quality of Experience(QoE)is introduced to describe the user perceived quality,and has become an attractive topic in recent years.To find a solution to improve the QoE and the demand of building QoE evaluation models are becoming more and more urgent.In this thesis,we expound the user subjective indicator and subjective evaluation method improvement,algorithm selection and improvement,design and implementation of users' QoE evaluation system.The specific contents are as follows:First of all,based on the user's viewing record dataset collected from the set-top box,the author puts forward the user's viewing custom indicator creatively from the user's subjective point of view which reflects the user's interest in watching TV.Aiming at the shortcomings of the traditional subjective evaluation methods,such as complexity,high cost and non-real time,the subjective evaluation method is improved according to the user's behavior indicator which map the user's viewing time ratio to the MOS value,quantifying the quality of experience.Secondly,according to the characteristics of video data analysis and actual scene tasks,comparing several classic algorithms,including regression,kNN and CART,the results showed that the CART algorithm is more accurate than the others.Therefore,we propose the algorithm based on the CART algorithm.Specifically,the kernel function method is used to improve the output value of the CART algorithm.The accuracy of the new model is verified by experiments.The results show that the proposed algorithm can improve the prediction accuracy of QoE model.Finally,for the set-top box dataset,we perform distributed data storage,processing and analysis on the big data platform,combining with the proposed algorithm to program and implementing the user's QoE evaluation systerm.This system is not only beneficial for service providers to monitor the service quality of real-time,optimize the network environment,upgrade the user experience and promote the brand value,but also help the users to better understand the service,know their satisfaction with the service and watching interest.Meanwhile,users can also put forward the suggestion on this service for its' shortcomings.