RE: [Megaco] Local/Remote descriptors and Permanent terminations

["Tom-PT Taylor"<[e-mail]>]

[Stephen Casner copied because this potentially affects AVT WG work in
progress.]

To summarize: we are trying to define how to indicate that silence
suppression is desired for a given audio flow.  We could do it through an
a=silenceSupp attribute, through codec-specific a=fmtp: parameters, or
implicitly by specifying the Comfort Noise codec as an alternative.  Markus
shows the sort of contradictions we could arrive at if we are not careful.

I think it's quite clear that we do NOT want to tie silence suppression to
the TDM package.  

The cleanest way would be to ensure that every codec where silence
suppression might apply has a silence suppression parameter.  For RTP, this
requires revisiting draft-ietf-avt-rtp-mime-06.txt, which is currently in
IETF Last Call.  Not sure what we do about ATM: RFC 3108 defines a profile
that just picks up the RTP mappings, so maybe it would also pick up the
parameters defined in the MIME type registrations and payload format
descriptions.  I find this whole area murky because theoretically there
should be new MIME type registrations for each transport in order to reuse
the profiles in SDP.

A second possibility is to define the a=silenceSupp attribute for all
transports, but specify that it takes second priority to a=fmtp parameters
where these are defined.  This is probably the most practical route.

Stephen, any comments?

-----Original Message-----
From: Pauls Markus [mailto:[e-mail]]
Sent: Wednesday, April 03, 2002 2:49 AM
To: [e-mail]
Cc: 'Carl Rutter'
Subject: RE: [Megaco] Local/Remote descriptors and Permanent
terminations


Suppose you have the following SDP:

m=audio 4444 RTP/AVP 0 18
a=fmtp:18 annexb=no         // G.729, Silence Suppression off (see
draft-ietf-avt-rtp-mime-05)
a=silenceSupp: on - - - -

(that means the GW may either use G.711 µ-law or G.729)

The silenceSupp attribute is not specific for a payload type, the fmtp
attribute is. So which one has precedence in this case?

In my opinion you should have the ability to switch on silence suppression
depending on the payload type. If you want to use it with G.711, it does not
mean that you want it with G.729.

Beyond that I believe generally it is more suited to an ephemeral
termination. Therefore it should be expressed by means of SDP.

However, if you want to transport the comfort noise payload (according to
draft-ietf-avt-rtp-cn-05), you should add the payload type 13 to the SDP
media line as well (IMO).

From my point of view the SDP should therefore look something like this:

m=audio 4444 RTP/AVP 0 18 13
a=fmtp:18 annexb=no
a=fmtp:0 silSup=yes

where this PCMU specific parameter silSup must still be defined (for PCMA as
well).

Any comment is appreciated.


	Kind regards

		Markus Pauls


________________________________________________________
Markus Pauls

SIEMENS
Information and Communication Networks

ICN WN CC NA B13   	Tel.:  	[phone]
Hofmannstr. 51     	Fax:   	[phone]
81359 Munich        	Email: 	[e-mail]
Germany




-----Original Message-----
From: Carl Rutter [mailto:[e-mail]]
Sent: Wednesday, April 03, 2002 3:36 AM
To: Chuong N. Nguyen
Cc: [e-mail]
Subject: RE: [Megaco] Local/Remote descriptors and Permanent terminations


You bring up a good point, 
Is/Should Silient Suppression be tied to the TDM package or
SDP TDM, if in fact its really more applicable to the Ephemral side.
If no can think of a good reason then your right it would make more sense to
be in the SDP.
For now we could use 0 13 for G711 and long-term use the ATM approach for
SDP.
Trying to drive towards closure as we have a need to implement this.
Yea, the hairpinning would be in the MG Switch.
Comments?

Carl
-----Original Message-----
From: [e-mail] [mailto:[e-mail]]On Behalf Of
Chuong N. Nguyen
Sent: Tuesday, April 02, 2002 8:05 PM
Cc: [e-mail]
Subject: Re: [Megaco] Local/Remote descriptors and Permanent terminations


  
That is what  I initially thought silence suppression was about. 
But then I was told differently. 
It could be applied to TDM. 
If not, then why are we discussing adding silence suppresion to the TDM
package? 
  
By the way,  I thought TDM-TDM calls are being made at the Gwy too. 
Aren't you guys doing that now? 
  
  
Carl Rutter wrote: 
I can't think of an application where you'd want to do that. Silence
suppression really wouldn't make sense for a TDM-TDMwhich is hairpined at
the CO. The DS0 is dedicated. The advantage of Silent suppression it to
limit the RTP traffic on the IP side, which you wouldn't have in that
case.Maybe if you wanted to inject Comfort Noise instead of backgroundnoice
in the TDM-TDM call??Carl 
-----Original Message----- 
From: [e-mail] [mailto:[e-mail]]On Behalf Of
Chuong N. Nguyen 
Sent: Tuesday, April 02, 2002 7:16 PM 
To: [e-mail] 
Subject: Re: [Megaco] Local/Remote descriptors and Permanent terminations
Tom-PT Taylor wrote: 
I've had a message pointing out the existence of 
draft-ietf-avt-rtp-cn-05.txt.  Specifying this payload type as well as G.711

would indicate the use of silence suppression implicitly.  That covers RTP 
transport.  For ATM we have the a=silenceSupp attribute defined in RFC 3108.

Do we need anything for TDM?

What if someone wanted to make a TDM-TDM, hairpin call? 
Would silenceSupp be used for such a call? 
If yes, then what? 
  
  
-- 
  Alcatel USA, Inc             Internet: [e-mail]
  1000 Coit Road Plano, Texas 75075           Phone:    [phone]
  **** The opinions expressed are not those of Alcatel USA, Inc **** 
-- 
  Alcatel USA, Inc             Internet: [e-mail]
  1000 Coit Road Plano, Texas 75075           Phone:    [phone]
  **** The opinions expressed are not those of Alcatel USA, Inc ****  


_______________________________________________
Megaco mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/megaco

_______________________________________________
Megaco mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/megaco