Re: [MMUSIC] RFC 6190 Single Session Transport

[Jonathan Lennox <[e-mail]>]

This was certainly my understanding of RFC 6190 as well, from heavy participation in its standardization in AVT.

I don't understand how single-session multi-source transmission will work without either a) signaling "a=ssrc-group DDP" decoding dependencies, or b) making various implementation-specific assumptions about the structure of the SVC streams.

On Jun 17, 2013, at 2:38 PM, Mo Zanaty (mzanaty) <[e-mail]<mailto:[e-mail]>> wrote:

My understanding of 6190 SST is a single SSRC for all layers, not multiple SSRCs (one per layer) within one RTP session. 6190 doesn’t state this clearly, but otherwise SST would need all the same mst-mode stuff as MST, since mst-mode defines how receivers put packets into decoding order from multiple streams based on timestamps or Cross-Session Decoding Order Numbers (CS-DON).

I think you may be referring to the common industry practice of MST using the same port and thus the same RTP session, i.e. Multi-SSRC (rather than Session) Transmission. That is not standardized in 6190, but often used in practice. This often uses a single m-line, so it has no BUNDLE dependency. If it used multiple m-lines, your argument would be true for that case. BUNDLE (or some similar mechanism to mux m-lines onto one port) would be required for that type of MST to use the same port. But that does not seem to be a limitation of Plan A, since some mux mechanism is always required to do that type of MST regardless of Plan A.

Mo

From: [e-mail]<mailto:[e-mail]> [mailto:[e-mail]<mailto:[e-mail]>] On Behalf Of Bernard Aboba
Sent: Monday, June 17, 2013 1:20 PM
To: [e-mail]<mailto:[e-mail]>
Subject: [MMUSIC] RFC 6190 Single Session Transport

At the virtual interim, a question arose about RFC 6190 Single Session Transport.

Here is what RFC 6190 Section 1 says:


   This memo defines two basic modes for transmission of SVC data,

   single-session transmission (SST) and multi-session transmission

   (MST).  In SST, a single RTP session is used for the transmission of

   all scalability layers comprising an SVC bitstream; in MST, the

   scalability layers are transported on different RTP sessions.

Section 1.2.2 clarifies the usage scenarios of SST and MST:



   SST should be used in point-to-point unicast applications and, in

   general, whenever the potential benefit of using multiple RTP

   sessions does not justify the added complexity...



   MST should be used in a multicast session when different receivers

   may request different layers of the scalable bitstream.

Since SST implies (by definition), use of a single RTP session to transport multiple layers, were plan A to be used for signaling, negotiation of SST would only be possible if both peers support BUNDLE, since otherwise distinct RTP ports would be required for each layer, which is incompatible with SST.

Today, existing SST implementations typically utilize a single m line.  As Jonathan noted, some implementations utilize the same SSRC for multiple layers, while others utilize distinct SSRCs.  Utilization of the same SSRC is possible, since separation of the layers can be handled within the H.264/SVC implementation.



_______________________________________________
mmusic mailing list
[e-mail]<mailto:[e-mail]>
https://www.ietf.org/mailman/listinfo/mmusic

--
Jonathan Lennox
[e-mail]<mailto:[e-mail]>