Re: [MMUSIC] Simulcast and 5576 source attributes?

Hi,

The text in 06 was " It is possible, but not required to use source-specific signaling

      [RFC5576] with the proposed solution."

I made the comment at the time about identifying which SSRC is related to each simulcast stream. The current text tried to address my comment in 07

I agree with Jonathan here.   Note the SSRC may change in the RTP due to collision for example.

BTW:  repeating myself, this is what we tried to address in https://tools.ietf.org/html/draft-even-mmusic-application-token-03  see section 5

Roni

From: mmusic [mailto:mmusic-bounces@ietf.org] On Behalf Of Jonathan Lennox
Sent: יום ב 13 נובמבר 2017 08:49
To: Bo Burman
Cc: mmusic
Subject: Re: [MMUSIC] Simulcast and 5576 source attributes?

a=ssrc:<ssrc> fmtp:<pt> doesn’t constrain what payload types the SSRC can use; it just gives additional format parameters for that payload type as applied to that SSRC.  (The motivating use case was H.264 sprop- parameters.)

On Nov 13, 2017, at 11:49 AM, Bo Burman <bo.burman@ericsson.com<mailto:bo.burman@ericsson.com>> wrote:

One "certain case" is when using a 1:1 mapping of rid-id and RTP PT, in combination with a 1:1 mapping of SSRC and RTP PT. This would e.g. be:
a=rid:<rid-id1> pt=<pt1>
a=ssrc:<ssrc1> fmtp:<pt1>
a=rid:<rid-id2> pt=<pt2>
a=ssrc:<ssrc2> fmtp:<pt2>
a=simulcast:send 1,2

This way, ssrc1 can be known to use rid-id1 and ssrc2 rid-id2, via knowledge what RTP PT they use. Therefore, which simulcast stream uses which SSRC is also known.

Cheers
Bo
(as individual)

-------- Original Message --------
Subject: [MMUSIC] Simulcast and 5576 source attributes?
From: Jonathan Lennox <jonathan@vidyo.com<mailto:jonathan@vidyo.com>>
Date: 9 nov. 2017 15:25
To: mmusic <mmusic@ietf.org<mailto:mmusic@ietf.org>>

The latest Simulcast draft (draft-ietf-mmusic-sdp-simulcast-10) has a paragraph at the end of Section 5.1 that says:

   It is possible to use source-specific signaling [RFC5576] with
   "a=simulcast", but it is only in certain cases possible to learn from
   that signaling which SSRC will belong to a particular simulcast
   stream.

However, there’s no other mention of 5576 in the document, and no registration of any 5576-style source-specific attributes.

What are the “certain cases”? It’s not clear to me there’s any way for the currently-registered 5576 attributes to tell you which SSRC corresponds to which simulcast stream, prior to receiving RID in RTCP or header extensions.  (Notably, there’s no rid source attribute.)

_______________________________________________
mmusic mailing list
mmusic@ietf.org<mailto:mmusic@ietf.org>
https://www.ietf.org/mailman/listinfo/mmusic