0 with the deep development of the web and mobile video live application, there are more and more platforms to support the interactive video streaming technology, the original RTMP streaming scheme can not meet the need of immediate interaction due to the delay. This paper presents several interactive video live models (just the idea does not represent this in the actual application) to share with you for further discussion.
1 peer
1.1 Model Diagram
1.2 Description
Even if the user sends a request to the signaling server, the signaling server notifies the primary user of the Mac, if accepted, bothTURNrequest the respective local sideIPorTURNa well-distributed public networkIP, by signaling the server to exchange the network information, the two parties prefer toPeerWay to try to join the other party's public networkIPIf the timeout fails, try to join theTURNServer-AssignedIPwith the port number, this time throughTURNthe server relays both media streams to ensure that anywayPeeris successful, and the master user mixes and mixes the multimedia stream toRTMPSend to a weak live streaming server cluster while sending a non-mixing, mixed screen stream to the userA, even MAK usersAcut off from the weak real-time streaming system to obtain the audio and video stream, began to accept the media stream decoding rendering of the main user, and also sent to the main users of the media stream to the master user, for the main user to play output and mixing, mixing screen.
1.3 Features
- Suitable for a single host simultaneously with a few (2 or so) users simultaneously with the Mac
- Increase terminal performance pressure at the host end
- The main implementation focus on the client, can refer to WEBRTC Implementation
- A high-real protocol transmission for the even-grain stream
- Less server relay links, lower latency
- Weak intrusion, no large-scale retrofit for existing weak real-time streaming systems
- No increase in server bandwidth traffic
2
2.1 Model Diagram
2.2 Description
Even MAK user a through the signaling control server to the main user of the MAI, even if the owner agrees, even Mak user a to the high real-time streaming media server to send media streaming, At the same time, the signaling control server is ready to broadcast to all viewers the media stream of the users of the MAI, and the Mac host and all viewers receive the media stream of the MAI user A and decode separately, if the direct voice output is directly to the audio device and the video is rendered in a mixed screen for the main user. If the audience to mix, mixed screen output to the audio device and rendering to the screen; even Mak user a does not need to disconnect the media stream that accepts the main user of the master, as far as it is necessary to display the video only in mixed or mixed screen.
2.3 Features
- Suitable for multiple host and multi-user interaction scenarios
- High real-time streaming media system complexity and coupling low, easy to expand
- Server side due to no need for mixing, mixed screen CPU pressure low
- The main implementation is focused on the server side, facilitate service upgrade
- Too much bandwidth
- For viewers, the difference between the media flow between the master user and the MAI user a
- High-invasive, large-scale transformation of current weak real-time streaming media systems
3
3.1 Model Diagram
3.2 Description
Even MAK user a through the signaling control server to send the main user to initiate a connection with the MAK request, even if the master user agrees, even MAK user a to disconnect the weak media stream, while the high real-time streaming media server to obtain the main user media and publish their own media streams, Even if the main user through the high-real-time streaming media server to obtain the media of the user, after the high-real-time streaming media server received the media stream of the user, start decoding, mixing, mixing screen, re-encoding by the original channel to send media streaming to the weak real-time streaming media server, the audience can sense the mixed media.
3.3 Features
- Suitable for multiple host and multi-user interaction scenarios
- Low flow rate, even the wheat user up and down, the audience few calls no sense
- Increased server-side pressure due to server-side mixing and mixing screens
- Moderate intrusion, weak real-time streaming media system can remain unchanged
- Less real-time than peer mode
- Highest complexity compared to the first two models
Analysis of the model and characteristics of user's continuous-wheat technology in video broadcasting