Research Of Echo Cancellation Algorithm And Design Implementation For WebRTC Conference System

With the continuous innovation of the Internet,people hope to use a more convenient and simple way to enhance the communication between each other through audio-video conferencing.And the audio-video conferencing system based on the browser has the advantage of realizing cross-platform,installation free and convenient access,which just meets this demand.At the same time,due to the endless emergence of various technologies,especially the open source WebRTC project of Google,the implementation of the browserbased multi-person meeting system maybe will be achieved,so its WebRTC technology has gradually received more and more attention and research.Since WebRTC technology has been open source,on the one hand,W3 C and IETF were responsible for the formulation and development of its standard work,and on the other hand,the WebRTC Alliance leaded the upgrade and update of its technical level.In recent years,WebRTC technology has been on the rise.For example,foreign YouTube applications,Chrome,Mozilla,Opera,Edge,Safari browsers,and domestic operators,equipment vendors,the Internet companies are actively researching and optimizing WebRTC technology.For the WebRTC project,whether within its framework the core algorithm research is carried out,or the audio-video application is realized based on WebRTC,the research on WebRTC has been gradually increasing in recent years.The thesis focuses on the research and analysis of WebRTC technology,and the main work contents are as follows: firtly,this thesis introduces the core parts according to the WebRTC framework: audio engine,video engines,transmission module.The audio engine includes codec,NetEQ,echo cancellation and noise suppression,and echo is a problem that the conference system has not been able to solve completely.Through the study of the current situation of echo cancellation,focusing on the analysis of adaptive echo cancellation algorithm in WebRTC,this thesis studied the core part adaptive filter algorithm,and discussed the principles of LMS algorithm,NLMS algorithm and block LMS algorithm respectively.By studied and summarized the principle of its echo cancellation algorithm,on the basis of the fixed step size NLMS algorithm and combining with the literature,an improved variable step size NLMS algorithm was proposed,and verified by simulation experiment.Secondly,through the research on the development status of WebRTC,this thesis studied the audio and video conferencing system technology provided by WebRTC,and analyzed WebRTC related signaling,NAT through technology.According to project requirements,this thesis realized the WebRTC multi-person audio and video conference through the session management mechanism based on socket.io,and on the basis of this,realized instant message communication,and completed the screen sharing function through the extension plug-in.And then test and analyze according to the system shows that in the process of multi-person audio and video communication,the system runs normally,the audio and video quality is good,the system basically meets the communication and usage needs of participants.