Kurtis Heimerl wrote:
I'd consider Rhizomatica the most successful, but
we have run Osmocom
with paying subscribers a few times, notably in Indonesia
Thanks for the links. I imagine that most people on this list probably
already know that my use case is very different, but perhaps some
people might not know, so let me restate it briefly: I don't have the
luxury of operating in a remote/rural geographical area that has NO
existing cellphone services, none of any G at all - instead I live in
an area where every square mm of Earth's surface is carped-bombed with
super-strong 4G and 5G signals from at least 3 major carriers. Given
this super-strong major carrier coverage, all mainstream sheeple with
their 4G/5G smartphones are perfectly happy and satisfied, creating
zero demand for community cellular networks. Instead the only people
who desire a community cellular network are those who have drawn a
line in the sand, said a resolute 'NO!' to all those later-G technologies
and want GSM/2G service instead, for use with either historical
handsets, specifically procured and acquired for nostalgia purposes,
or self-made ones.
However, each included significant engineering efforts
that may be
useful for your own deployment. It can definitely be done and we used
a bulk SIP-Based Voice/SMS provider (Nexmo, now vonage) to do it.
I noticed one element of strangeness: in all of your articles you
talked about the difficulty of acquiring real phone numbers in the
operating country's PSTN numbering plan. In the first paper you wrote
how a deployment in Indonesia used Swedish phone numbers instead of
native Indonesian ones, in the second paper you wrote about your
specific partner MNO lending you not only spectrum, but also
maintaining a chokehold over your pool of phone numbers, and in the
last LTE paper you wrote about specifically opting out of standard
VoLTE because getting phone numbers and PSTN interconnection is so
difficult and expensive.
Why is it that your experiences with obtaining real phone numbers and
PSTN interconnection are so sharply different from mine? Right now I
can go to
bulkvs.com and buy USA phone numbers (real 10-digit NANP, in
the same numbering space as decades-old land lines and mainstream cell
carriers) for something like $0.06 (yes, 6 _cents_) per month, with
super-cheap PSTN interconnection included - super-cheap meaning
$0.0003 (yes, count the zeros) per minute for inbound and $0.004 per
minute for outbound. When I saw these prices for the first time about
a month ago, I was shocked, I couldn't believe my eyes that traditional
PSTN interconnection can be _this_ cheap - but there it is... Is this
type of deal a USA-only phenomenon? Are you saying that similar deals
for PSTN interconnection with real phone numbers are not available in
other countries that are primary targets for building community
cellular networks?
Labs <rp.labs(a)gmx.ch> wrote:
The numbering format is the last of your issues.
Yet in the context of this technical mailing list, providing correct
user experience with respect to phone numbers is one of the reasons
why I need to develop custom sw beyond what Osmocom already provides.
At first glance, in both OsmoHLR and OsmoMSC the legacy term "extension"
has already been replaced with the standard term MSISDN - thus my
first naive thought was that I could enter NANP numbers as MSISDNs in
the OsmoHLR database and have these MSISDNs seen natively by OsmoMSC.
But the way in which the MSISDN concept is implemented in OsmoHLR/MSC
currently disregards all notions of TON and NPI (type of number,
numbering plan indicator), instead treating the MSISDN datum as a raw
string of digits only.
With all mainstream cell operators in USA, a user can dial a domestic
number in 3 ways: as just 10 digits NPANXXxxxx, as 1NPANXXxxxx, or as
+1NPANXXxxxx, and the network is smart enough to recognize all three
as valid ways of dialing the same number. I insist on implementing
the same user experience on my Themyscira Wireless network - but
neither OsmoMSC's built-in MNCC nor its built-in SMSC are smart enough
to handle such number nuances, hence I will need to replace both the
built-in MNCC and the built-in SMS routing mechanism with my own
external implementations. (I am not smart enough to implement a patch
to OsmoHLR and OsmoMSC that would handle TON and NPI for MSISDN
generally, beyond my own specific use case, hence I am going for
external augmentation instead.)
If you provide a
telephone service with a real number for dial out/in you must have also
emergency services working.
bulkvs.com provides E911 support, and so do most VoIP/SIP providers.
What is happening if your subscriber is outside of
your coverage area
and wants dial another number?
Naturally, providing any kind of service outside of coverage area is
impossible: even in those areas where T-Mobile have not shut down their
GSM service yet, they will never allow my made-up IMSIs (starting with
my own made-up PLMN code, *not* stepping on any assigned one) to
register on their network - hence no services of any kind are possible
then.
Maybe you will need some local roaming
agreements with the big operators.
The same T-Mobile who are itching to kill the last remains of their
GSM network? I don't see how there can be any possibility of a roaming
agreement with them - if the sole reason for We the People needing to
set up our own community GSM networks is because T-Mobile are killing
theirs, they certainly won't be inclined to give us roaming access to
that same GSM network of theirs which they are itching to kill.
There is a reason why the other networks running
opensource mobile
networks like Rhizomatica don't use real numbering plan and keep the
networks isolated.
Yes, they have their reasons - but I continue to argue that their
approach would not be right in USA. Even if someone were building a
network in a desert location with no existing services, as opposed to
my very different use case, as long as that desert is located
somewhere within USA where NANP phone numbers would be culturally
appropriate, I argue that they should give their subscribers
"the real thing".
I imagine one could probably connect to/with USA phone numbers from
anywhere in the world, by connecting over the Internet to a USA-based
VoIP provider, but it would "feel wrong" to do it outside of USA, it
would be like those Swedish phone numbers in Indonesia.
Good luck, I hope you succeed.
"I hope you succeed" is rather meaningless without defined criteria
for what counts as success. For the present phase of my project, I
have set a very low bar for what I would count as success. In the
present phase of my project, my goal is to demonstrate some really,
really old phones to an assembled audience and show them working.
Right now the oldest phone in my collection is a recently acquired
Ericsson I888 (late 1990s), and I already have it connecting
successfully to my Themyscira GSM network with my own Themyscira SIM -
FCSIM1 programmed with my own fc-simtool. But without outside
connectivity it does not "feel real" - hence my continued work to
implement the necessary software for outside connectivity.
Now imagine me handing this ancient 1990s phone (big and heavy as a
brick) to an audience member and inviting that person to call someone
with it - *that* would be impressive, and getting to that stage is
what I am currently working on. All of the necessary work is software,
_not_ political or regulatory etc.
Let us know the results.
Once I have some working code implemented, I will upload it to a
Mercurial repository on
freecalypso.org and post the link.
M~