Multipoint Conference Unit (MCU)
Each participant of Multipoint Conference Unit (MCU) can publish a WebRTC stream and play a mixer stream with audio and video from the other participants and own video (without own audio)
To organize Multipoint Conference Unit (MCU), real-time stream mixer function is added to WCS. Lack of own audio playback for the conference user eliminates the echo effect.
The key difference between the MCU and the classic conference is the number of video streams received by each participant.
In the case of a regular conference, each participant receives N different video streams and plays them in N <video> elements on the page. As a result, if the user receives 3 streams of 1 Mbps each, then the total bitrate is 3 Mbps.
In the case of the MCU, each participant receives only 1 mosaic stream, assembled from the flows of other participants, with a fixed bitrate, which depends on the output resolution of the mixer (720p 2Mbps). And plays in one <video> element. Mixing is done on the server side and consumes server CPU and RAM resources.
That is, in the first case of regular conferences, we save CPU and RAM of the server, but we spend a lot of network and browser resources on rendering N <video> frames. In the second case (MCU), we save the network and browser resources, but heavily load the CPU and RAM of the server, because mixing 1 room MCU may require 2 physical cores.
Specifications
Stream sources |
Protocols | Playback platforms |
|
|
|
- Chrome
- Firefox
- Opera
- Safari, Mac OS
- Safari, iOS
- Edge
- iOS SDK
- Android SDK
- WebRTC
- Chrome
- Firefox
- Opera
- Safari, Mac OS
- Safari, iOS
- Edge
- iOS SDK
- Android SDK
Scheme of the MCU
Each MCU participant sends video + audio streams to the WCS. WCS gives MCU participants a mixed stream that contains video of all participants and audio of all but their own.
Step-by-step diagram of the MCU
- Browser 1 publishes audio + video stream on WCS
- Browser 2 publishes audio + video stream on WCS
- WCS mixes the published streams into one
- WCS returns mixed stream without its own audio to browser 1
- WCS returns mixed stream without its own audio to browser 2
For information on testing Multipoint Conference Unit (MCU), see Testing
Download Web Call Server 5
System requirements: Linux x86_64, 1 core CPU, 2 Gb RAM, Java
Installation:
- wget https://flashphoner.com/download-wcs5.2-server.tar.gz
- Unpack and install using 'install.sh'
- Launch server using command 'service webcallserver start'
- Open the web interface https://host:8444 and activate your license
If you are using Amazon EC2, you don't need to download anything.