See the modifications proposed below to Karl's Implementor's Guide text for Endpoint B exceptional procedures. ========================================================== Randy Wuerfel Enterprise Network Systems Development Unisphere Solutions, Inc. E-mail: Randy.P.Wuerfel@icn.siemens.com 4900 Old Ironsides Drive Fax: (408) 492-4666 M/S 200 Tel: (408) 492-4375 P.O. Box 58075 Santa Clara, CA 95052-8075
-----Original Message----- From: Klaghofer Karl ICN IB NL IP 7 [mailto:Karl.Klaghofer@ICN.SIEMENS.DE] Sent: Friday, March 26, 1999 2:33 AM To: ITU-SG16@mailbag.cps.intel.com Subject: AW: Call Transfer
Transfer with Consultation: As Ernst pointed out, we need the ctSetup Invoke as it is defined in case of Transfer with Consultation. Transfer without Consultation (i.e. no secondary call exists): Some history: H.450.2 has chosen to use the same operation (ctSetup Invoke) in both cases. Transfer with and Transfer without Consultation in order to avoid having different procedures / state events within endpoint B for those two cases (with and without Consult). However, the point Espen raised is a valid question. We should put some exceptional procedure handling to the implementors guide (based on the suggestions made by Ernst) that clarify the usage of Transfer without Consultation in conjunction with an endpoint C that does not support Transfer. If endpoint C does not support H.450 at all, a basic call ALERTING, CONNECT would be received in endpoint B. This would lead to a FACILITY with ctInitiate Return error being sent to A (since ctSetup operation needs explicit confirmation). Endpoint B would end up with two calls A and C. To avoid this, the implementors guide should state some exceptional procedure handling that would apply to Transfer without Consultation (i.e. no secondary call exists).
" Endpoint B exceptional procedures: a) When receiving an ALERTING or CONNECT message from endpoint C (that does not include a response to the ctSetup Invoke) while being in state CT-Await-Setup-Response, the transferred endpoint B shall (or may ?) continue as if a ctSetup Return Result had been received. This allows endpoint B to successfully continue with the Call Transfer procedures (i.e. appropriate internal call transfer state handling and clearing of the primary call to the transferring endpoint A). This exceptional procedure enables successful Call Transfer when the transferred-to endpoint C does not support H.450. b) When a RELEASE COMPLETE message is received (possibly containing a ctSetup return error APDU) in response to a SETUP message that contained a ctSetup Invoke sent by endpoint B (which happens if endpoint C supports H.450.1, but not H.450.2), then endpoint B may retry call establishment to endpoint C using a normal basic call (i.e. by sending a SETUP message without a ctSetup invoke APDU). Upon receiving an ALERTING or CONNECT message from endpoint C on this new call, the procedures as described above in a) shall apply. "
Remark: If endpoint C supports H.450.1 but no H.450.2, I would NOT start ignoring the i-apdu "clear call". I-apdu always shall be honoured (if H.450.1 is supported), since C does not know that this is about call transfer (so it can not really be an exceptional procedure as part of a Call Transfer standard). This would cause problems when further supplementary services are inroduced that also use i-apdu "clear call".
Any comments ? Regards, Karl Klaghofer Siemens AG
------------------------------------------------------
The simplest option is to just relax the 'clearCall' requirement in case of single-step transfer. The effect would be that a ctInitiate error is returned to user A and the primary call is not cleared. User B would have two calls to handle. Another alternative is the simple option plus adding the clearing of the primary call to the exceptional procedures for the transferred endpoint. This leaves a clean situation for all users, with B and C having an active call. For the consultation case H.450.2 should stay as it is.
> -----Ursprüngliche Nachricht----- > Von: Ernst Horvath [SMTP:ernst.horvath@SIEMENS.AT] mailto:[SMTP:ernst.horvath@SIEMENS.AT] > Gesendet am: Mittwoch, 24. März 1999 10:58 > An: ITU-SG16@mailbag.cps.intel.com mailto:ITU-SG16@mailbag.cps.intel.com > Betreff: AW: Call Transfer > > We must consider two cases: > > First case: a secondary call exists. Here the new call from the > transferred endpoint is to replace an existing call which is identified > by the callIdentity value in the callTransferSetup invoke APDU. It is > essential that the transferred-to endpoint matches the new call and the > existing call, accepts the ne and releases the old call. Otherwise, if > the new call is treated just as another basic call, the called user may > be considered busy (with the old call). > > > Second case: No secondary call exists ("single step transfer"). Here > the requirement could be relaxed, the new call could be treated as just > a basic call if callTransferSetup is not understood. This would be the > case anyway if H.450.1 is also not supported. > The main consequence would be that the additional information about the > call having been transferred is lost. > Of course, also with the existing H.450.2, the transferred endpoint can > always repeat the SETUP without a callTransferSetup APDU when the first > SETUP fails. > > Ernst Horvath > Siemens AG > maito: ernst.horvath@siemens.at mailto:ernst.horvath@siemens.at > > > > -----Ursprüngliche Nachricht----- > Von: Espen.Skjaeran [SMTP:Espen.Skjaeran@ERICSSON.COM] mailto:[SMTP:Espen.Skjaeran@ERICSSON.COM] > Gesendet am: Dienstag, 23. März 1999 20:44 > An: ITU-SG16 > Cc: Espen.Skjaeran > Betreff: Call Transfer > > Hi, > > I have a specific question to H.450.2 Call Transfer. > -Why do the transferred-to endpoint have to support the service? > > Section 6 of H.450.2 states > "When conveying the invoke APDU of operation callTransferSetup, the > Interpretation APDU shall contain value > clearCallIfAnyInvokePduNotRecognized" > which I find to mean "if you don't support the service, forget it" > > The session between the transferred and the transferred-to endpoint > follows a basic call, and no useful information is returned > in the ctSetup. result operation. Why can't it be "dumb" and only > believe it is doing a basic call? > Isn't the receipt of ALERTING/CONNECT instead of RELEASE COMPLETE > acknowledgement enough that the transfer was successfull? > > Anyone to clear up this? > > Thanks, > Espen