High Bandwidth Requirement in case of low bandwidth connections

I have been facing issues using Jisti meet on DigitalOcean when a large number of people, usually more than 4, attempt to video conference. Most of my employees are from India and the Philippines who do not sometimes have very high broadband capabilities. Their video freezes and sometimes cuts off completely making it very tedious to have a proper conversation.

It is my understanding that Jisti does not “stitch” together videos streams into one video which is then sent to everyone, which would mean that each person is downloading multiple video streams from everyone else. In a limited bandwidth situation this can cause a problem. What I attempted, is to reduce the maximum video resolution down to 240px. It did reduce the problem allowing 5 people to comfortably converse but not more.

Is there a way that i can force Jitsi to “stitch” the video streams into one video stream on the server side, so as to reduce the load on individual bandwidth?

1 Like

There is no option to switch videobridge to become an MCU by “stiching”.

The options you have to further reduce bandwidth using SFU is to use the LastN setting so that the videobridge only forward video from the last N number of participants.

I am sorry but could you elaborate on that? You must excuse me but I didn’t quite understand your technical explanation.
What is SFU and what does forwarding only last N participates mean?
Will the others appear frozen? (Not that it really matters if those who aren’t speak appear a bit frozen)

Also the link you provided is not opening.

Edit:
I managed to find the document you suggested online. Thank you for your help.

Attempting to decrypt @xranby’s message:

  • MCU - multipoint control unit (is this Jitsi Meet? Or Jitsi VideoBridge?)
  • SFU - Selective Forwarding Unit. That page has a video.

So according to the paragraph below from the abstract of that paper, an SFU is a more efficient type of MCU, by selecting only the most recent N speakers instead of everyone, and… mixing together their video streams into one stream to everyone? Or sending each video stream from those N speakers to everyone (according to the video)?

In this paper, we describe a Multipoint Control Unit (MCU) that identifies and selects the last N dominant speakers and forwards their streams to all the conference participants. We evaluate the performance of this Selective Forwarding Unit (SFU) against a simplistic everyone-to-everyone (full-star) MCU. Our results show that the SFU uses 45% less CPU and 63% less bandwidth when forwarding media for 10 of the endpoints in a 30-participant conference.