FreeCalypso, Themyscira Wireless and American 2G Cooperative

[Mychaela Falconia]

Hello FreeCalypso and Themyscira community,

I've been silent on this ML for a year and a half (since 2024-10), so
I figure it is time I give everyone an update on what I've been up to
lately, in areas related to GSM/2G retro-cellular phone technology.

The original FreeCalypso project, which was/is a desire to design and
build my very own Libre Dumbphone handset (2G-only) based on my beloved
TI Calypso chipset, running firmware based on TI's reference fw for
this chipset, is still on indefinite hiatus.  The desire is still
there, every day I use a pair of Pirelli DP-L10 phones (one with a
T-Mobile SIM, the other with my own SIM for my own home network) as my
"daily driver" personal phones, I wish with every fiber of my being
that they were FreeCalypso phones rather than Pirelli.  However, since
USA's only remaining GSM operator (T-Mobile) is viciously hostile to
2G and they keep itching to shut their GSM network down - and the
network that still exists at the moment exhibits atrociously poor
quality - right now it would be pointless to work on a better GSM
handset without building a replacement GSM network first.

The desire to build a new GSM/2G network in USA has been occupying me
since 2022, the year when TMO made an all-out declaration of war
against 2G.  This desire has had several iterations, but about a year
ago in early 2025 we (I and some supportive family members) got
serious about it and formed a proper non-profit organization.

American 2G Cooperative has been a legal entity (incorporated in
Delaware, USA) since 2025-05-01.  We haven't filed paperwork with IRS
for tax-exempt status yet, but we are getting ready to do so - the law
firm which we hired in 2025 to guide us through non-profit start-up is
currently reviewing a draft of our application.  We are not pursuing
the usual 501(c)(3) status used by charities and churches - instead we
are set up as a mutual/cooperative telephone company per 501(c)(12),
a very rarely used type of tax-exempt non-profit org.  Compared to the
more common 501(c)(3), the special 501(c)(12) status we seek has a
downside in terms of ability to attract donations: when you donate to
a 501(c)(3) charity, you can claim that donation as a tax write-off,
but the same benefit does not extend to 501(c)(12) and most other
501(c) sections.

However, IRS (the tax agency in USA) is not the only govt agency we
have to please - there is also FCC, the agency that regulates radio
spectrum and business relations (interconnection, roaming etc) between
carriers.  In FCC terminology, all cellular phone services, no matter
which G, are collectively called Commercial Mobile Radio Services or
CMRS.  In order to be eligible to obtain a spectrum license in either
of the two GSM-suitable bands (PCS1900 and GSM850 in USA) and in order
to be able to compel other carriers into letting us roam on them per
FCC rules, we have to be a bona fide CMRS operator ourselves.  A
charity-type organization presenting its GSM operation as educational
or anything along those lines (potential ways to qualify for 501(c)(3)
charity status with IRS) would have a very hard time claiming CMRS
status with FCC - but a mutual/cooperative phone company operating per
501(c)(12) - well, yeah, it's a phone company!  Hence we have to accept
inability to give our donors a tax write-off as the necessary price
for being an acceptable type of company for FCC purposes.

At this point the reader is bound to ask: how are we going to obtain
any GSM-suitable spectrum, either PCS1900 or GSM850, when every bit of
cellular-type spectrum in every band is already claimed and spoken for?
The trick is to pursue geographical areas that are more rural and
remote, farther away from so-called "civilization".  Here is the
situation with spectrum availability in PCS1900 and GSM850 bands
in USA:

In PCS1900 band there are somewhere between 20 and 30 locations across
all of USA (exact number changes depending on what counts as a single
location vs separate locations) that exhibit a vacancy.  In these
20-to-30 peculiar locations some portion of PCS1900 band (a certain
frequency range out of the full band) is not assigned to any company
at all, but rather sits orphaned.  This situation exists in locations
where previously issued licenses for those band portions got terminated
by FCC for failure to construct, were voluntarily canceled (surrended
to FCC) by the licensee, or expired without renewal - e.g., the company
quietly disappeared at some point between decennial renewals of the
license.

We as in A2GC definitely plan to pursue at least one of these PCS1900
band vacancies at some future time - but not right now.  Current FCC
provide no pathway for any new company to obtain a new license in any
of these vacant spots except by way of FCC holding a reauction of this
spectrum - but the last time they did such reaction in PCS1900 band
was in 2008.  All current vacancies in this band result from license
terminations, cancellations or expirations that happened since 2008 -
and it's been quite a while now, hence orphaned chunks of rural PCS1900
spectrum have been piling up.  Of course a situation where some piece
of spectrum sits fallow, serving no one, while at least one company
desires that spectrum and has a concrete plan to put it to use is
indefensible - hence we are hopeful - but the fight will be anything
but easy.  Therefore, my current plan is to pursue GSM850 spectrum
instead, as a way of getting our metaphorical foot in the door, and
then revisit PCS1900 some years later.

In GSM850 band the situation is totally different, thanks to very
different FCC rules that were promulgated a decade or longer before
PCS.  In all newer bands FCC took the map of USA, divided it into
arbitrarily drawn chunks of territory (called BTAs, MTAs, EAs etc) and
auctioned each chunk to the highest bidder.  Each winning bidder got
exclusive right to the spectrum in the entire BTA/MTA/etc chunk of
territory that was auctioned, and they get to keep this license (with
exclusivity over the full territory) as long as they serve some
required percent of area population.  But the old Cellular
Radiotelephone Service, aka 800 MHz Cellular, frequencies coincide
with the band that is GSM850 to us, has very different rules.  In this
band each licensee gets exclusive right to the spectrum only in areas
where they claim coverage (they have to file maps with FCC), as opposed
to fixed chunks of territory that were defined once and for all.

The practical effect is that every current licensee in 850 MHz band
has a claimed area of coverage (area in which they hold exclusive use
of spectrum) that is somewhat amoeba-shaped.  In major metropolital
and otherwise densely populated areas these "amoeba" shapes of
different carriers' 850 MHz spectrum claims join each other as
perfectly fitting jigsaw puzzle pieces, without any gaps - this effect
is probably a result of geographically adjacent carriers in each
frequency block mutually agreeing on the boundary between them as each
carrier expanded its area.  But as we go farther away from civilization,
deeper into rural areas, the picture changes: there are many rural
areas where either Block A or Block B exhibits a vacancy, or even both
blocks.  (FCC's "800 MHz Cellular" band is divided into just two
license-able frequency blocks, A and B - whereas all newer bands are
divided into more blocks.)

And here is the best part: in any so-called "unserved area", meaning
an area where either Block A or Block B is unclaimed, a new company
can straight-up apply for a new license without having to wait for a
govt-initiated auction!  But there are some significant teeth in this
licensing process too:

1) One cannot get a license simply by saying "we wish to build a new
cellular system in this town" - instead every application must be a
lot more detailed.  In order to be acceptable for consideration, an
applicant for license has to specify the exact number of proposed cell
sites (one or more), the exact GPS coordinates (to the second of lat
and long) of each proposed site, the type of support structure (building
or tower), whether existing or proposed to be built, exact height
above ground, exact equipment types (they really care about the model
of antenna and its radiation pattern diagrams), exact power level that
will be used, and expected signal reach distances along 8 cardinal
radials.  The practical effect is that we have to have agreement with
people on the ground (hosts for our proposed cell site) _before_
applying for a license with FCC, we have to have our proposed equipment
tested in a lab and known working, and we have to have a realistic
propagation model.

2) In order to be eligible for a license, every proposed new system
has to have a coverage area of at least 130 square km, computed by
taking submitted signal reach distances along 8 cardinal radials and
applying linear interpolation between them.  (The only upper limit is
not stepping on territory that is already claimed by some other
carrier.)  This coverage area requirement has two implications:

2a) Low-power femtocell hardware is not sufficient - a traditional
full-power BTS is needed.

2b) On flat terrain, mast height requirement is minimal: having the
antenna 3.2 m above ground is enough to reach beyond the required
6.43 km in every direction.  But if mountainous terrain is involved,
the set of viable cell site locations may be reduced to mountaintops.

3) Once a license is granted, the licensee has to complete construction
and make the system operational (defined as providing service to
subscribers and roamers) within 1 y of license grant.  The practical
implication is that we need to complete as much of the needed work as
possible _before_ applying for the license, before the clock starts.

Toward this goal of putting together an acceptable application for a
new GSM850 license, we are now looking at some GSM850 BTS hw made by
Nokia.  Originally my focus was on Ericsson RBS6k hw that once enjoyed
a burst of popularity in Osmocom, and I already obtained some DUG20
and RUS01 B2 modules, both former property of T-Mobile USA, as evidenced
by very distinctive inventory stickers.  However, this RUS01 B2 hw is
usable only for PCS1900 band and not for GSM850.  Unlike RUS01 B2,
RUS01 B5 units appear to be unobtainium!  (I know that Rhizomatica
scored some in the past, as they operate in 850 MHz band, but no more
of them seem to be available today.)  But with the help of some other
people in Osmocom, my eyed were opened to the alternative of Nokia BTS.
I was told what part numbers to look for, and lo and behold, GSM850
radio units are available!

We are now actively in the process of buying this Nokia GSM850 hw -
only two days ago we finally recovered a big chunk of our non-profit
co-op money that up until that point seemed unrecoverably lost.

And now I have to explain the part about lost-till-now money and about
the massive delay which we suffered.  We lost about 10 months of
schedule, from 2025-05 to the present, and almost lost $7500 of our
non-profit money, to a risky business deal that went very bad in 2025.
There was a company (which I shall not name) that held some blocks of
PCS1900 spectrum and advertised that spectrum as being for sale - and
we made a good-faith attempt to buy that spectrum.  The single
individual behind the shell company in question demanded that we pay
him $2500 upfront, then another $5000 in 2025-07, and thus he ended up
holding $7500 of our money in total.  With the first $2500 there was
extreme time pressure and lack of ability to take the matter calmly
and slowly - the license was approaching expiration, and the guy said
he won't renew it (would let it expire) unless we paid him $2500
upfront.  With the second $5000 in 2025-07 I was thinking with my heart
instead of my head - that was before I discovered widespread vacancies
in GSM850 band, and of course PCS1900 is much better for old phone
compatibility - and I felt that we had no choice but to submit to that
guy's capricious demands if we wanted that PCS spectrum.

But mere days after receiving the second chunk of money from A2GC, the
guy reneged on his contractual obligation (we had a little one-page
contract drawn up) to follow through with Assignment of Authorization,
and started trying to rope us into a leasing deal instead.  Needless
to say, it would be business suicide for us to invest massive time,
effort and money into a new GSM build-out in an extremely remote rural
area (so remote that slang expression "BFE" would be fitting) having
only a lease to the spectrum rather than ownership, a lease than can
be revoked by the license-holding company at any second.  We got into
a stalemate and went around in circles for several months, until in
2025-11 (the month in which we finally got to explore an area with
PCS1900 and GSM850 spectrum vacancies with boots on the ground) we
finally told the guy we were no longer interested in his PCS spectrum
and wanted our money back.

Per our contract, the guy was obligated to return our money if he did
not file for Assignment of Authorization (transfer of spectrum) as
agreed.  But it took us several months of battle, from 2025-11 till
now, to recover our money.  We were preparing to take the guy to court
for recovery of our money, but that process is much easier said than
done - obstacles and barriers at every step.  Finally, by applying
some additional leverage that cannot be discussed in detail on a
public ML, we got him to return a big part of our money - $5000 to be
exact.  He still owes us the remaining $2500, but since he already
paid the bigger portion, we'll give him more time to pay the remainder,
before we resort once again to further collection actions.

The important part is that with $5000 out of $7500 returned back to
our Cooperative, we are now unblocked in terms of proceeding with our
plans.  Therefore, no matter how long it takes us to recover the
remaining $2500, there should be no more stalling of our plans - we
have enough to acquire GSM850 BTS hw and proceed with lab experiments
and sw development.

Also to clarify: while I count the 10 month period from 2025-05 to the
present as time lost to the miscreant who screwed us over, we still
accomplished a few things during this time:

* We got most steps of our non-profit org setup completed, including
a set of bylaws which the law firm OKed as acceptable.

* I and another member of our team physically traveled to one of the
areas we identified that has spectrum vacancies in both PCS1900 and
GSM850 bands, and explored that area with boots on the ground.  Among
other exploratory steps, I confirmed with a handheld SA that those
spectrum vacancies are real - the SA showed noise floor in those
frequency ranges.

* If you look in Osmocom commit logs, you will see some sw improvements
I managed to produce during this period. :)  For those without E1 BTS
hardware, using native OsmoBTS instead, my patches fix the bug of
"half-block" artifacts during DTXu with FR & EFR codecs, and the new
Themyscira Wireless RTP jitter buffer and RTP endpoint implementations
are now merged into Osmocom and available for use in mainline OsmoBTS.

* Not quite Osmocom proper but Osmo-adjacent, tw-e1abis-mgw works
together with OsmoBSC and E1-based BTS.  While I don't have any "big"
E1 BTS hw running in my own lab yet, between the InSite unit donated
by Harald & Sarah and some testing that was done by other Osmocom
community members, the new MGW already provides significant
improvements in the quality of operation with Osmocom + E1 BTS combo.

With new-to-us Nokia BTS hw expected to arrive soon, I'll be working
to bring it up with OsmoBSC - a combination that hasn't been tried by
anyone in the community, as far as we can tell.  Further developments
in OsmoBSC and tw-e1abis-mgw will likely be needed at that point.

Besides E1 BTS work, the plan is still active to produce a new MSC in
which existing osmo-msc, suitably modified with a special mode of
operation enabled via vty, will constitute just one component.  I will
write more about these plans later when I am ready to start actually
working on this part.

I should also mention one further reason why I put so much emphasis on
A2GC becoming a physical GSM operator (be it PCS1900 or GSM850) in some
tiny corner of rural America - doing so will give us access to roaming.
FCC rules section 20.12(d) requires licensed CMRS operators to grant
roaming access to others who are similarly licensed, and do so on FRAND
terms.  Therefore, once we become a physical operator, providing
physical GSM services somewhere under a license from FCC, we'll be
able to compel T-Mobile to let us roam with our own IMSIs (the license
will also allow us to obtain an official MCC-MNC assignment) on their
very unloved GSM/2G network!

Right now there is a very peculiar situation with GSM/2G in USA: users
with foreign SIMs, issued by operators in some other country, get a
much better experience on T-Mobile's 2G network than native users who
got their SIM directly from T-Mobile or from a TMO-hosted MVNO that
operates with IMSIs out of TMO's range.  With TMO-owned IMSIs, the
network applies some kind of artificial blocking where non-VoLTE
phones that aren't sufficiently "grandfathered" get booted from the
network.  I don't have air interface traces showing exactly how the
"booting" happens, thus I can only speculate, but the criterion for
the artificial blocking decision appears to be the TAC part of the
IMEI.  It seems to involve the IMEI and not just the fact of connecting
to GSM RAN instead of LTE - I say so because others in Vintage Mobile
Phone community have observed that "modern" phones configured to
connect in 2G mode ("use 2G networks only") connect successfully and
don't get booted out.

But here is the sweet part: foreign IMSIs roaming on T-Mobile USA
never get booted out or otherwise artificially denied service, no
matter what phone model is used, including various 2G-only and 2G+3G
phones beloved by the Vintage Mobile Phone community.  But of course
there are obvious difficulties with obtaining a SIM from some other
country, having a phone number in a different country code (which most
people would be reluctant to call, fearing international charges),
dealing with foreign-currency and foreign-language billing systems,
and receiving various texts from the "home" operator in a foreign
language and wondering each time if it is important or not...

But I reason that if we approach T-Mobile with a demand to be given
roaming access per FCC rules section 20.12(d), we'll be able to demand
(with FCC enforcement if necessary) that we be given the same level of
access which they give to Mexican Telcel and other foreign home
operators, i.e., no artificial blocking of users who intentionally use
a Vintage phone!

And yes, there will be some very exciting Core Network software
development if we accomplish this feat: SS7 MAP protocols for roaming.
In GSM architecture, MSC+VLR entities and HLRs are supposed to
communicate with each other via standardized SS7 MAP, so that MSC+VLR
of operator A can talk to HLR of operator B, as needed during roaming.
But in current Osmocon CNI suite, OsmoMSC talks to OsmoHLR via GSUP
instead of SS7 MAP, a greatly simplified protocol that is private to
Osmocom.  Per my understanding, the primary reason for the current lack
of SS7 MAP support in Osmocom is lack of a business use case: why
invest the work to implement those complex protocols if no one will
allow our hobby networks to interconnect with real commercial ones for
roaming.  But if we can use FCC rules section 20.12(d) to *compel* an
incumbent commercial operator into letting us roam on them, then
suddently we'll have a real use case for SS7 MAP in Osmocom, and a
reason to implement it. :-)

Hasta la Victoria, Siempre,
Mychaela aka The Mother