GSM Modem Confusion

Wally Ritchie wally.ritchie at
Sat Jul 28 08:47:17 CEST 2007

On 7/27/07, Harald Welte <laforge at> wrote:
> On Wed, Jul 18, 2007 at 11:11:16PM -0400, Wally Ritchie wrote:
> > So it is with calypso based neo1973. I think we have here a
> > good starting point with as near a complete 7.07/7.10 implementation
> > as is possible consistent with other requirements.
> >
> > As for the "modem". IMHO the word "gsm modem" should be reserved for
> > what gsmd provides to external TE e.g. as a laptop, PC, or PDA, that
> > uses a neo1973 for gprs or hcsd access. This is another 7.07 (possibly
> > even with a 7.10 layer) interface accessed through Bluetooth (or irda,
> > or usb serial). It would be less confusing if we didn't refer to
> > the calpyso 7.07/7.10 as a modem (perhaps chipset instead).
> unfortunately that is not our strategy.  Your view seems to be heavily
> influenced by years (decades?) of experience in mobile phone
> development.  But we're not coming from a mobile phone point of view.
Mang Ren Mo Xiang. Openmoko is an elephant and we are all wise but blind men
touching different parts of it. (some are more blind than others). What
part of the elephant you touch and your own personal experience determine
your view of what an elephant really is.

> You see, what we are trying to build here is not something that
> technically resembles 'yet another cellphone', but what we're building
> are open source linux-based mobile communications devices.
You can't really build somethat that's "not" something else. You build
something that is. Sure, less is sometimes more. But focusing on things
to leave out often leads to throwing the baby out with the bathwater. Yes,
there are very often things you can safely drop from the feature set. However,
this has attached risks. In the kind of chaotic environment we have now it is
easy to rationalize away things that could really be unidentified "must haves".
It's not too bad if you can just add them in later if necessary.
However, if this
means re-architecture then you may find yourself in great difficulties. There is
always competition.

The dilemma with Mobile Handsets has always been the fact that any given
person usually uses at most 10% of the features. 5% are very basic features.
The other 5% are drawn from the 95%. A successful (economically) handset
model will often have few features that NOBODY uses. This effect will not
go away just by saying we're not really a handset. In the consumer market if it
walks like a duck and talks like a duck its a duck. Perhaps I am a wrong
in supposing that you're after the mass market? Perhaps you just want to
sell a few thousand a month of niche product to Linux geeks. But I am assuming
somebody wants to make some serious money here. IMHO, you could be
in multiple 100K's per  month by Christmas 2008 if you don't screw the pooch
by ignoring the ultimate mass market consumers of these devices.

> We have no interest to support an AT command interface to any external
> entity outside of our devices, at least certainly not for another 12
> months, if not beyond that.
Do you really mean you want to exclude say a windows-only user from
sending a FAX through your device. Do you really mean you don't want
to tap into KU250 in your other pocket? Do you want to cut off xyz's corp's
win2000 laptops from GPRS?

> Similarly, we don't have any intention to implement MMS, WAP, WML, etc.

Do you really think there is not demand for these, especially MMS. Do you
think there is some factor out there is going to cause $0.01/kilobyte pricing
to disappear from every market in the world? It's about mobility and ubiquity!

> So to us, GSM in the mobile communications device (which we actually
> don't call a phone ourselves!) is a technology to provide one particular
> (minor) feature such a GSM voice calls, and to provide IP access over
> GPRS (and later 3G).
It's not a phone, it's an experience! :-)

You'll eventually find that you have to reduce "what is openmoko" or "what is
a FIC phone" (as my kids call it) to the elevator conversation. (15 second
"what do you guys make"?) I think you might end up closer to something
like "we make an open source mobile phone with any kind of feature you can
imagine including a built in linux desktop". Philosophically you can think
of it as anything you want but I think in the end you'll be unavoidably attached
to "mobile handset" or "cellphone" or whatever localized term for ME. I don't
think you'll be saying "we make great Linux PDA's with a built-in phone". How
many "what's a PDA" versus "what a mobile phone" do you think there are in
the world.

> If somebody wants to provide internet access to another device (i.e. his
> laptop), our strategy is to do this via USB ethernet, BNEP or even WiFi.
> The Neo becomes your portable access point, to which you talk TCP/IP.
> So this requirement for 'another 7.07 layer' really doesn't exist on our
> roadmap.
Unfortunately, the world is filled with legacies and trying to escape from them
leaves you vulnerable to competition that does the same thing as you, maybe
not as good as you, but deals better with the real world legacies. Earliest does
not usually win. Best does not usually win. Early Enough Good Enough usually

All of the above is just my humble opinion.

Cheers, Wally
> --
> - Harald Welte <laforge at>         
> ============================================================================
> Software for the world's first truly open Free Software mobile phone

More information about the gsmd-devel mailing list