"Early media" with fast connect
Dear experts, there seems to be a possible conflict with early media during call establishment using fast connect. Clause 8.1.7.1 of H.323 clearly says that the calling endpoint must be prepared to receive media right away on the ports it indicated in fastStart, and that the called endpoint may start transmitting media as soon as it accepts a fastStart proposal (unless, of course, the calling endpoint indicated also "mediaWaitForConnect"). This is independent of any Progress indicator. Clause 8.1.7.4, on the other hand, requires that the calling endpoint provides tones like ringback locally if it did not receive a Progress indicator #1 or #8. So the calling endpoint is obliged to interrupt/ignore possible media streams from the called endpoint and replace them with its own tones. Isn't this a conflict situation? I would like to hear your opinion or any experience from implementations. Kind Regards, Ernst Horvath Siemens AG
"Early media" with fast connectErnst, I think the key is "prepared to receive." The endpoint may be prepared to receive, but it will not necessarily play the streams. My assumption is that, if the endpoint is playing local ringback, it will do so until either: 1) It is told not to do so any longer 2) RTP streams arrive However, playing RTP streams before receiving CONNECT may have some implications. Perhaps some endpoints send RTP data that contains no real content... thus inadvertently killing the local ringback. Whatever the resolution, I think we need to ensure that local ringback happens as long as is necessary, but that we terminate once "real" media arrives-- we do not want to clip audio. Paul ----- Original Message ----- From: Horvath Ernst To: ITU-SG16@echo.jf.INTEL.COM Sent: Friday, April 26, 2002 7:40 AM Subject: "Early media" with fast connect Dear experts, there seems to be a possible conflict with early media during call establishment using fast connect. Clause 8.1.7.1 of H.323 clearly says that the calling endpoint must be prepared to receive media right away on the ports it indicated in fastStart, and that the called endpoint may start transmitting media as soon as it accepts a fastStart proposal (unless, of course, the calling endpoint indicated also "mediaWaitForConnect"). This is independent of any Progress indicator. Clause 8.1.7.4, on the other hand, requires that the calling endpoint provides tones like ringback locally if it did not receive a Progress indicator #1 or #8. So the calling endpoint is obliged to interrupt/ignore possible media streams from the called endpoint and replace them with its own tones. Isn't this a conflict situation? I would like to hear your opinion or any experience from implementations. Kind Regards, Ernst Horvath Siemens AG
Hi, I think that the preparedness for early media is based on the fact that the media may arrive before the Connect or Progress to which it relates. These signalling messages come over a reliable channel (and may contain the Fast Start) and may be delayed by retransmission requirements, whilst the media may drop a few packets but still arrive. Local Ringback should be applied on receipt of Alerting, whilst remote media should be applied on receipt of an Inband indication (tones, announcements, silence, ...). These situations are pretty clear, but what happens in the other cases: media on its own, Alerting with media? I've often wondered. It probably depends on your capabilities. Does the MGC know from the MG that you are receiving media, does it have any power (or is it silence), can it mix it with local ringback, can it not mix it? Some of it may have to be left to "Implementation Defined", but guidelines as to what *should* happen would be nice. Douglas At 23:36 27/04/2002 -0400, Paul E. Jones wrote:
<fontfamily><param>Arial</param>Ernst, </fontfamily> <fontfamily><param>Arial</param>I think the key is "prepared to receive." The endpoint may be prepared to receive, but it will not necessarily play the streams. </fontfamily> <fontfamily><param>Arial</param>My assumption is that, if the endpoint is playing local ringback, it will do so until either: 1) It is told not to do so any longer 2) RTP streams arrive </fontfamily> <fontfamily><param>Arial</param>However, playing RTP streams before receiving CONNECT may have some implications. Perhaps some endpoints send RTP data that contains no real content... thus inadvertently killing the local ringback. </fontfamily> <fontfamily><param>Arial</param>Whatever the resolution, I think we need to ensure that local ringback happens as long as is necessary, but that we terminate once "real" media arrives-- we do not want to clip audio. </fontfamily> <fontfamily><param>Arial</param>Paul </fontfamily> ----- Original Message ----- <bold>From:</bold> <<mailto:ernst.horvath@SIEMENS.COM>Horvath Ernst <bold>To:</bold> <<mailto:ITU-SG16@echo.jf.INTEL.COM>ITU-SG16@echo.jf.INTEL.COM <bold>Sent:</bold> Friday, April 26, 2002 7:40 AM <bold>Subject:</bold> "Early media" with fast connect Dear experts, there seems to be a possible conflict with early media during call establishment using fast connect. Clause 8.1.7.1 of H.323 clearly says that the calling endpoint must be prepared to receive media right away on the ports it indicated in fastStart, and that the called endpoint may start transmitting media as soon as it accepts a fastStart proposal (unless, of course, the calling endpoint indicated also "mediaWaitForConnect"). This is independent of any Progress indicator. Clause 8.1.7.4, on the other hand, requires that the calling endpoint provides tones like ringback locally if it did not receive a Progress indicator #1 or #8. So the calling endpoint is obliged to interrupt/ignore possible media streams from the called endpoint and replace them with its own tones. Isn't this a conflict situation? I would like to hear your opinion or any experience from implementations. Kind Regards, Ernst Horvath Siemens AG <<<<<<<<
participants (3)
-
Douglas Clowes -
Horvath Ernst -
Paul E. Jones