In converged communication, we often hear the following terms: MCU service, SFU architecture, MESH architecture, star network and so on. Many customers are confused when they hear these data, and often say that we just want a device that can pull multiple devices into the same conference, why is it so complicated. Today we will talk about where these terms come from, where they are applied, and how our anyRTC does it.
1. Mesh architecture
Due to the popularity of WebRTC, people are no longer as unfamiliar with audio and video communication as before. WebRTC itself is a P2P communication model, and P2P is point-to-point communication. With the help of STUN/TURN services, communication under any network can be realized. The most commonly used architecture for the P2P communication model is MESH: that is, multiple terminals are connected to each other in pairs to form a network-like structure.
As shown in the figure, the four terminals B1, B2, B3, and B4 perform many-to-many communication. When B1 wants to share media (such as audio and video), it needs to send data to B2, B3, and B4 respectively. In the same way, if B2 wants to share media, it needs to send data to B1, B3 and B4 respectively, and so on. This solution is more friendly to the situation where there are few terminals, but as the number of terminals increases, the bandwidth requirements for each terminal will be very high.
2. SFU architecture
It can be seen that the MESH architecture is not suitable for large-scale group meetings, and the SFU architecture came into being.
SFU (Selective Forwarding Unit) architecture: It is composed of a server and multiple terminals. However, compared with MESH, SFU broadcasts a stream on the server side. After receiving an audio and video stream shared by a certain terminal, it directly sends the The audio and video streams are forwarded to other terminals in the room. For example, B1 publishes a stream to the SFU, and the SFU distributes the stream to B2, B3, and B4. At the same time, SFU does not do audio and video transcoding, does not require high server configuration, and has good real-time performance and high flexibility. It is very suitable for large-scale distribution networks, which makes the SFU architecture easily the most widely used in the market. Universal architecture.
3. MCU architecture
At this time, some people will ask, the SFU architecture seems to be perfect, and other architectures are not needed. The answer is no, let's introduce the originator architecture of video conferencing: MCU architecture.
MCU (Multipoint Conferencing Unit) architecture: A server and multiple terminals form a star structure. Unlike SFU, MCU can do codec and video synthesis. B1, B2, B3, and B4 send the audio and video streams they want to share to the server. The server will decode and mix the audio and video streams of all terminals in the same room, and finally generate a mixed audio and video stream. Send it to each terminal, so that different terminals can see/hear the audio and video of other terminals. Due to the addition of codecs, the configuration requirements of the server are very high, so the MCU architecture is not suitable for large-scale distribution, but this does not mean that this architecture is not important.
Due to its early start and mature technology, the MCU architecture is widely used in hardware video conferencing. At the same time, MCU can avoid the differentiation of different codec devices through decoding and re-encoding, meet the integration and access needs of more customers, and improve user experience and product competitiveness. In addition, the terminals connected to the MCU upload all the way and receive all the way, so the requirements for the network are not so high, and the user experience is relatively good.
Four. Summary
Through the above introduction, we have learned about so many architectures of streaming media services. The anyRTC integrated communication system can deliver different solutions according to the actual needs of users, so as to better meet the actual needs.
