| WebRTC(Web Real-Time Communication)is an open real-time communication and transmission technology,which is initially designed to realize real-time audio and video transmission through Web browsers and mobile applications.Inside WebRTC,it provides a better experience for real-time communication between users through adaptive bit rate adjustment and jitter buffer to adapt to jitter in different network environments,and forward error correction and selective retransmission mechanism to adapt to packet loss in different network environments.Due to the generality of the internal design of WebRTC,it can still lead to the problem of video lag during playback when dealing with different network conditions.Based on the in-depth study of WebRTC,this paper analyzes the network jitter and packet loss affecting the video playback process of WebRTC,and proposes a mapping model based on the mapping between the video arrival interval and buffering time and the continuous video playback time,and a mapping model based on the mapping between the packet loss rate and the video buffering time,respectively.On this basis,by analyzing and verifying the long-term characteristics of the network measurement data,the video playback quality of the network link in the future period of time is predicted in a probabilistic sense based on the statistical characteristics of the historical network measurement data.The work accomplished in this thesis is mainly in the following aspects:(1)Research on customized measurement methods for WebRTC.In this paper,we first conduct an in-depth analysis of the process of GCC,the congestion control algorithm in WebRTC,and pro vide a customized active measurement scheme for WebRTC by analyzing the packet-sending behavior of WebRTC,which realizes the evaluation of WebRTC service quality in the network link without the help of the WebRTC video stream itself.Based on this,an optimization scheme for the above active measurement configuration is proposed,which achieves WebRTC network service quality measurement for a large number of IPs with reduced resource consumption at the measurement side.(2)A mapping model of packet arrival interval and video playback quality metrics for WebRTC is proposed.Firstly,the factors that cause lag during video transmission are analyzed,and the main factors affecting video lag,i.e.,network packet loss and network jitter,are clarified.At the same time,based on the in-depth study of the internal video coding method of WebRTC and the buffer at the receiver side of WebRTC,the network jitter is converted into the packet arrival interval,and the packet arrival process and the video playback process are modeled as a queuing system,and the diffusion approximation is used to calculate the probability distributions of the video buffering time and the video continuous playback time in the video transmission process of WebRTC.(3)Propose a WebRTC oriented video transmission system.(3)Propose a mapping model of packet loss rate and video buffering time for WebRTC.Aiming at the phenomenon of packet loss in the network,we analyze the causes of packet loss events at the application layer of WebRTC through the study and analysis of the recovery and retransmission mechanism of packet loss in WebRTC,quantify and analyze the packet loss events at the application layer of WebRTC by using the Markov model,and compute the probability distribution of video buffering duration due to packet loss by using the Markov model.(4)Realization of a measurement system for streaming media playback quality.The above proposed WebRTC-oriented customized measurement method and WebRTC-oriented video playback quality mapping model are transformed into concrete engineering implementation,which mainly includes the design and implementation of the configuration center for storing active measurement configurations,the analysis of measurement data,and the storage of measurement data.On this basis,the mapping model of packet arrival interval and video playback quality index for WebRTC and the mapping model of packet loss rate and video buffering time for WebRTC are tested and verified in a public network environment using actual measurement results. |
