TURN + Multicast (WebRTC)

[Chris Lercher]

Hi,

I wonder, if it would be (at least in the future) possible that TURN could do a one-to-many multicast (of e.g. WebRTC video streams)?

If I want to keep the confidentiality constraint, that the server can't take a peek into the encrypted WebRTC streams (which is the case with TURN, as answered here https://groups.google.com/d/msg/turn-server-project-rfc5766-turn-server/0CI4lwNwP4w/jzeC52di-7cJ), then using a media server instead of TURN is most probably not a solution (or is it?)

It is possible to encrypt something in a way that multiple recipients can decrypt it (see e.g. http://stackoverflow.com/a/38853/291741), so from a purely theoretical standpoint, it could be done. But are there already any endeavors to make this practically usable? Maybe even for the rfc5766-turn-server?

- Chris

[Philipp Hancke]

If I want to keep the confidentiality constraint, that the server can't take a peek into the encrypted WebRTC streams (which is the case with TURN, as answered here https://groups.google.com/d/msg/turn-server-project-rfc5766-turn-server/0CI4lwNwP4w/jzeC52di-7cJ),

Note that when DTLS-SRTP is used, the TURN server can not decrypt media even when it has access to the signalling path (note however that active attacks are still possible) 

 

It is possible to encrypt something in a way that multiple recipients can decrypt it (see e.g. http://stackoverflow.com/a/38853/291741), so from a purely theoretical standpoint, it could be done. But are there already any endeavors to make this practically usable?

The problem is that you get different feedback from different participants and you can not send the same stream to everyone. http://tools.ietf.org/html/draft-ietf-avtcore-rtp-topologies-update-01 provides a nice explanation of the problem.

[Oleg Moskalenko]

The TURN specs were designed for peer-to-peer communications. I am not aware of any work toward one-to-many communications.

Oleg