Subject: Re: GSM03.40 Port Addressing Date: Mon, 27 Sep 1999 09:09:53 +0100 From: Holley Kevin To: 3GPP_TSG_T_WG2_SWG3@LIST.ETSI.FR I think that Stefan makes some very valid points here, so my feeling is that we should keep with the scheme of including port addresses with all segments. However I don't understand point 4. If the ME is sending all messages to the TE then there is no need for the ME to keep any of them. Why would the port address make any difference? The port refers to the port on the device to which the message is sent - i.e. the port on the TE unless the message is ME specific (11, 15, F1 etc) or SIM specific (12, 16, F2 etc) in which case it should not go to the TE in the first place. Or am I missing something? > -- > Kevin Holley > BT Adastral Park > Tel: +44 1473 605604 > Fax: +44 1473 623794 > Mobile: +44 7802 220811 > > > -----Original Message----- > From: Stefan Pusl [SMTP:Stefan.Pusl@MCH.SIEMENS.DE] > Sent: Friday, September 24, 1999 3:09 PM > To: 3GPP_TSG_T_WG2_SWG3@LIST.ETSI.FR > Subject: AW: GSM03.40 Port Addressing > > Hello Owen, Ian, Kevin, > hello everybody on this list, > > from the point of a mobile manufacturer things look like this: > > The SM-TL has to decide depending on a couple of features for each short > message how to handle it. > > We have to consider, that segments of a concatenated short message may be > lost or received in a different order than originated. > > If we assume to include the port address only in the first segment of a > concatenated short message, the following problems arise: > > 1. If the SM-TL doesn't receive the first segment of a concatenated short > message, then the MS has to store each segment of the concatenated short > message until the first segment is received. The disadvantage of the > situation is, that the MS must have the capability to store 254 segments. > That's not acceptable. > > 2. If the storage capacity is lower than 254 segments, the transfer of > further segments of the concatenated short message stops until the user > has > deleted some segments. The consequence is, that the deleted segments are > lost for the application. > > 3. Point 1 already arises for one certain combination of message > reference, > originating address and port address. But the SM-TL has to cope with more > than one message reference, more than one originator and more than one > port > address. > > 4. Assume the TE has indicated the ME to transfer all incoming messages. > The > SM-TL doesn't receive the first segment of a concatenated short message, > then the ME sends the received segment to the TE. When the ME receives the > first segment with the port address later, the previous received segments > of > the concatencated short message are lost for the ME (especially for the > application). > > The reported problems can be avoided, if the port address is included in > each segment of a concatenated short message. > > Best regards, > > Stefan Pusl > ICP CD MP GSM RD M 51 > Grillparzer Str. 10 > 81675 München > Tel.: 089/722-31023 > Fax: 089/722-25167 > stefan.pusl@mch.siemens.de > > > > -----Ursprüngliche Nachricht----- > > Von: Owen Sullivan [SMTP:owens@ALDISCON.IE] > > Gesendet am: Freitag, 24. September 1999 13:30 > > An: 3GPP_TSG_T_WG2_SWG3@LIST.ETSI.FR > > Betreff: Re: GSM03.40 Port Addressing > > > > Hi Ian, it is my understanding that the mobile station must use the > > Concatenated short message reference number field from the Concatenated > > short > > messages Information Element (or Concatenated short messages, 16-bit > > reference > > number field from the Concatenated short messages, 16-bit reference > number > > Information Element) to determine to which datagram set a particular > > message belongs to. > > > > As far as I know, the Application Port addressing Information Element is > > irrelevant in making this decision. The Application Port addressing > > Information > > Element would therefore be irrelevant in the case mentioned below where > a > > mobile station must decide that a message that has arrived is a part of > a > > new > > datagram set rather than a previous partially received datagram set. > > > > The Concatenated short messages Information Element must therefore be > > present > > in every message of a datagram set, but it is still not clear to me why > > the > > Application Port addressing Information Element must be. > > > > As you say, this may be easier discussed in person. > > > > Kevin, in terms of the different cases you mentioned, the main case I > had > > in > > mind when asking the question is fixed to mobile terminated messages. > > > > Thanks very much. > > > > Harris Ian wrote: > > > > > The proposal to put the IEI in the first segement of a concatenated > > message > > > only is OK provided there is adequate safeguard in the receiving MS to > > take > > > into account that from time to time, segments may be missing which > could > > be > > > the first or last or all those after the first. If the first segment > has > > > been received then the last may never arrive but the sending entity > may > > have > > > decided to send a different set of concatenated messages with a > > different > > > IEI. The receiving entity may not have suficient information to > > understand > > > that a new set of messages are about to arrive - possibly for a > > different > > > port. > > > Clearly, we know from experience that out of sequence messages and > > missing > > > segments in a sequence of messages is unlikely but it is likely that > > > segments after the first will never get delivered because of absent > sub > > and > > > validity period conditions. > > > IT is difficult to have this sort of debate by e-mail but I support > any > > > method to make best use of the bandwidth provided that the integrity > of > > SMS > > > is not compromised by assuming a perfect environment. > > > This is worthy of debate in Korea I feel. > > > > > > > -----Original Message----- > > > > From: Holley Kevin [SMTP:kevin.holley@BT.COM] > > > > Sent: Thursday, September 23, 1999 2:34 PM > > > > To: 3GPP_TSG_T_WG2_SWG3@LIST.ETSI.FR > > > > Subject: Re: GSM03.40 Port Addressing > > > > > > > > I think that this is an area where we need some more thought. My > > feeling > > > > is > > > > that we have a number of applications which assume transparency in > the > > > > SMSC > > > > for mobile to mobile messages, although this is not stated in 03.40. > > > > > > > > I therefore think that we should split this into two cases - for the > > > > mobile > > > > to mobile case then either the SMSC makes no changes and passes > > through > > > > transparently or we need some method to indicate whether this is > > desired > > > > or > > > > not. > > > > > > > > For the mobile to fixed case then you might want something different > > or > > > > you > > > > might not. > > > > > > > > Whether or not the SMSC is involved, some bandwidth can be saved as > > Owen > > > > says by including the IEI for port addressing only in the first of a > > > > series > > > > of messages. > > > > > > > > Comments are invited! > > > > > > > > Regards, > > > > > > > > > -- > > > > > Kevin Holley > > > > > 3GPP TSG T Vice Chairman > > > > 3GPP TSG T2 & SMG4 Chairman > > > > > BT Adastral Park > > > > > Tel/Fax: +44 20 7519 9028 > > > > > Mobile: +44 7802 220811 > > > > > > > > > > > > > > > -----Original Message----- > > > > > From: Owen Sullivan [SMTP:owens@ALDISCON.IE] > > > > > Sent: Wednesday, September 22, 1999 3:36 PM > > > > > To: 3GPP_TSG_T_WG2_SWG3@LIST.ETSI.FR > > > > > Subject: GSM03.40 Port Addressing > > > > > > > > > > If two or more mobile terminated short messages are being sent via > > GSM > > > > SMS > > > > > the Concatenated short messages Information Element must be > included > > in > > > > > each so > > > > > that the handset can reassemble the messages into a larger > datagram. > > > > Must > > > > > the > > > > > Application Port addressing scheme Information Element be included > > in > > > > each > > > > > message, or is it sufficient to include it only in the first > message > > of > > > > > the sequence? > > > > > > > > > > Including the Port only in the first message means the handset > must > > > > > reassemble the > > > > > entire datagram, then forward the entire datagram to the > appropriate > > > > > application > > > > > based on the Port address in the first message. > > > > > > > > > > Sending the Application Port addressing scheme Information Element > > only > > > > in > > > > > the > > > > > first message/fragment of a datagram would save 4 octets in each > > > > > subsequent > > > > > message/fragment of the datagram (assuming 16 bit addressing is > > used). > > > > > > > > > > I cannot find a statement in GSM03.40 that clarifies the issue > > either > > > > way. > > > > > > > > > > Thanks very much. << File: Card for Owen Sullivan >> > > << Datei: Card for Owen Sullivan >>