Research On QoE Parameter Measurement Method Based On Analysis Of Video Behavior

Video services under the mobile network are developing rapidly,to balance hardware investment and user experience,mobile network operators need to make accurate assessments of video QoE.The behaviors such as playback state switching and traffic data generation of the video terminal are controlled by the terminal application behavior model.The analysis shows that the model consists of three parameters:initial buffer queue length??₁?,buffer queue lower limit??₂?and trigger buffering queue length??₀?,which is the key to evaluate QoE.Most of the related researches on the three parameters of the model are estimated by using the relevant feature values of the video packets'message.The accuracy of the results is low and cannot meet the needs of QoE evaluation.To solve the above problems,this thesis studies the two types of video platforms,encryption and non-encryption,and proposes a method based on video data behavior,to accurately measure the above three parameters,and the result is accurate to the video frame.Then construct the behavior model of the video platform,and finally measure the video buffer delay with the help of the behavior model.This thesis mainly includes the following contents:?1?Build the behavior model of non-encrypted platform Youku.For Youku with non-encrypted transmission protocol,analyze the Youku traffic characteristics,fragmentation mode and transmission mechanism by collecting Youku mobile video traffic.On this basis,to identify Youku video traffic,quantitatively control the amount of video data actually received by the terminal application,establish the association between playback status and video traffic,using video fingerprint information,and use customized video with tag attributes,to accurately measure the behavior model's parameters,and the accuracy of results is to the video frame.?2?Build the behavior model of encryption platform YouTube.For YouTube with encrypted transmission protocol,the YouTube encrypted traffic is collected on the mobile terminal.To identify the encrypted video traffic,and analyse the impact of the encryption mechanism on the video traffic and playback status.Develop measurement applications based on YouTube's official API,establish the relationship between playback status and encrypted video traffic,adopt a new encrypted video fingerprint library construction scheme,and use the customized video with tag attributes,to accurately measure the behavior model's parameters,and the accuracy of results is to the video frame.?3?Measuring the buffer delay based on the video behavior model.Using the YouTube video behavior model,analyze the underlying logic that investigates buffering.Analyze the encrypted traffic characteristics under TCP and QUIC,identify the TCP protocol video fragmentation;define the YouTube video traffic data cluster under the QUIC protocol,and identify the data cluster's fragmentation.The behavioral model parameters are used to determine the time point at which the buffering is started and the playback condition is reached.Finally,the video buffering delay is calculated.