![]() |
2009-10-16
, 18:58
|
|
Posts: 3,397 |
Thanked: 1,212 times |
Joined on Jul 2008
@ Netherlands
|
#52
|
All that all of you talking about implementation MMS support is dirty hack.
1. MMS is not unique service that requires separate interface. Just see the OMA and 3GPP's R5 requirements. All operator's services should be provided thru IMS. MMS is one of. So, the MMS feature is SIP.
2. IP-address spaces of Internet's connection (Inernet's PDP or Wi-Fi) and operator's PDP could be the same. There is collision.
The Following User Says Thank You to allnameswereout For This Useful Post: | ||
![]() |
2009-10-17
, 11:19
|
Posts: 18 |
Thanked: 18 times |
Joined on Oct 2009
@ Russia
|
#53
|
SIP? Some comments in this thread point out that in some cases a seperate connection is required. I'm not sure on the statistics though, thats quite foggy.
In my case MMSC has public IPv4. If you'd sent the MMS and then give target device the link to the content you'd need to give a public IPv4. So the MMSC must have at least one public IPv4. If operators use a private IPv4 for their MMSC that is their stupidity because there is no need for that.
The Following 4 Users Say Thank You to Performer For This Useful Post: | ||
![]() |
2009-10-17
, 20:39
|
Posts: 182 |
Thanked: 540 times |
Joined on Aug 2009
@ Finland
|
#54
|
The Following 5 Users Say Thank You to abbra For This Useful Post: | ||
![]() |
2009-10-18
, 07:57
|
Posts: 18 |
Thanked: 18 times |
Joined on Oct 2009
@ Russia
|
#55
|
Performer, the things you are describing are same we have been discussing internally already. Dirty hacks could help right now but probably will not work for all cases. Separate TCP-stack is something you can get with Linux 2.6.30+ (separate networking namespaces). Maemo 5 is on 2.6.28 baseline where networking namespaces are not available, thus all this discussing with hacks.
As you are representing an operator business, would you be able to reflect on how important MMS is to operators? I know from statistics that generally users utilize MMS at a rate about 2-4% of SMS use, which is definitely very low and being replaced by alternative options -- email, direct social services communications, etc -- over time. But how important whole MMS business to operators?
The Following 4 Users Say Thank You to Performer For This Useful Post: | ||
![]() |
2009-10-18
, 08:54
|
Posts: 547 |
Thanked: 1,383 times |
Joined on Sep 2009
@ Stockholm, Sweden
|
#56
|
![]() |
2009-10-18
, 09:39
|
Posts: 18 |
Thanked: 18 times |
Joined on Oct 2009
@ Russia
|
#57
|
Thanks for the great input, glad to see we got someone working for an operator in here to tell us how they look at it
While I do realise fullfledged MMS support is impossible with the current kernel, I do still have a question.
Is it possible to implement it and get it working to such a degree you would be able to receive MMS only when there is no other connection active (and then open a connection to the MMS APN)? Assuming of course we can somehow read the WAP Push messages (if they don't show up in the el.db).
![]() |
2009-10-19
, 13:50
|
|
Posts: 72 |
Thanked: 51 times |
Joined on Jul 2008
@ Germany
|
#58
|
The way I understood it, the first message asks the phone what capabilities it has, but I assume there is some kind of timeout for that response as well as you said tso.
The Following User Says Thank You to Framstag For This Useful Post: | ||
![]() |
2009-10-19
, 14:08
|
|
Posts: 72 |
Thanked: 51 times |
Joined on Jul 2008
@ Germany
|
#59
|
In IMS infrastructure MMS agent shoud request service thru SIP before applying MM1. Just check specs of MM1st3.
Sorry, but my experience shows that most of operators uses private IP-networks for their services. I'm employee of one of them who has about 70 million of subscribers. Most parts of data-interchange between operators performs thru private networks like GRX by reasons of security and stability.
The Following 3 Users Say Thank You to Framstag For This Useful Post: | ||
![]() |
2009-10-19
, 14:32
|
Posts: 18 |
Thanked: 18 times |
Joined on Oct 2009
@ Russia
|
#60
|
The MMSC of my former employer marked an user/device as MMS capable as soon as the device send an MMS on its own. Part of the MMS submit request also can be a device id (string) and a profile URL (XML file), which in turn can be used for getting device capabilities and content conversion when sending an MMS to a device.
Yes, I timeout for retrieval could be used, too.
Gruß...Tim
Just look the truth. In real situation to implement it we should consider next requirements:
1. MMS is not unique service that requires separate interface. Just see the OMA and 3GPP's R5 requirements. All operator's services should be provided thru IMS. MMS is one of. So, the MMS feature is SIP.
2. IP-address spaces of Internet's connection (Inernet's PDP or Wi-Fi) and operator's PDP could be the same. There is collision.
3. MMS PDP usualy have separare DNS (if used to resolve HTTP Proxy / WAP 1.2 Gateway address).
4. 3G network and GSM supports up to 2 separate PDP-conexts at the same time.
5. WAP-Push indicators has specific message type and i'm not sure that messages of that type will be stored at the event database.
So the best idea is to implement multi-instanced TCP-stack like it's implemented at Symbian and qualified support of WAP-Pushes. After that we could speak about MMS frond-end and rendering.
I know nothing about implementation of multi-PDP at Android but the Windows Mobile implementation is too dirty. As i understood the iPhone implementation has second TCP stack for MMS.