Research And Implementation Of Streaming Media Distribution System Based On Video Quality

With the popularization of the mobile Internet and the superposition of the COVID-19 epidemic,enterprises and individuals have an increasing demand for real-time video communication.Video conferencing solutions are highly interactive and are an important scene in real-time video communication.Due to the complex network situation and the instability of the scene,the transmission of high-quality and low-latency video streams has become an important challenge in video conferencing scenarios.Bitrate adaptation is a method that automatically adapts to the capacity and performance of the network by dynamically adjusting the bitrate of the video stream,which has become a prevalent solution to provide users with high-quality video.At present,the main strategy of bitrate adaptation methods is to select the highest bitrate when network conditions allow,but because the bitrate has a marginal decreasing effect,that is,as the video bitrate increases,the growth rate of video quality is decreasing.This strategy will cause waste of network resources and encoding computing resources,and ignore the impact of video content characteristics on video quality.In addition,with the increase of business volume,a large amount of data flow brings bandwidth pressure and computing pressure to enterprise equipment,which not only increases the cost of economy,but also leads to the degradation of real-time video quality and affects the Quality Of Experience(QOE)of users.In view of the above problems,this task designs a streaming media distribution system based on video quality in video conference scenarios,mainly to improve video quality from two aspects:one is to integrate a bit rate adaptive algorithm that considers video content,this algorithm combines the method of DRL(Deep Reinforcement Learning),through the network conditions and the characteristics of video frames,choose a high video quality bit rate instead of always choosing a high bit rate,so as to reduce the bandwidth waste caused by the marginal decreasing effect of the video bit rate;the second is to design a high-performance streaming media topology structure,combining the mainstream MCU structure(MultiPoint Control Unit)and SFU(Selective Forwarding Unit)structure considering the real scene,it can be optimized from the aspects of calculation amount,bandwidth consumption,low latency and controllable video quality,and in the In the case of large business volume,the advantages of this structure are more obvious.Experiments were designed to verify the effectiveness and practicability of the above method,and streaming media distribution system based on video quality in the video conference scene is built,including the design and implementation of the server and the client,and the final system function test and performance test meet the expected goals.