It shouldn't be that hard. A million dollars, a few people who know about RF hardware, and the (possibly leaked) GSM specs should be enough.
There might be some trouble with encryption / carrier's private keys, but I think that shouldn't be an issue, since a multiband GSM phone works with every strange operator in every remote country, as long as the frequency is supported.
Some cool applications for a really free GSM from the top of my head:
- Virtual SIMS, you only use the data from the SIM, not the physical card to login. (Might be tricky because the SIM protects its information. If you don't want to resort to attacks ala chip shaving and electron microscopes to get the keys, you might just setup a simless private network with a DIY SIM tower (that already exist, AFAIK), as a kind of proof-of-concept.)
- You can see when you recieve a silent SMS ("type 0"), that authorities use to ping your phone to trace you. You can also send them.
- You can record audio from and play it to the phone, which is not possible in most consumer phones
- You can implement hardware encryption (which is tricky for voice, since GSM uses psychoaccoustical compression techniques that might not work on a encrypted stream - nevertheless this is a solved problem and there should be papers on this)
- You would find a bunch of security holes in the network, since its been relying on obscurity and trusted devices for so long.
- You could have a phone that's bottom to top trusted and open-source.
Many of the other things you mention can already be done by using off the shelf GSM baseband chips, since they don't necessarily do much access control of what gets sent.
Another other issue is the carriers want the baseband chisets and protocol stacks certified before allowing it on their network.
Personally, I think it could be OK as long as the processor is contained (i.e., no open access to the main memory, reliable mechanism for fully shutting it down, etc).
They would likely not be happy about it. I assume it would be something like the relationship between tivo and people running hacked tivos ... a cat and mouse game that a bit of effort can keep you ahead of.
BUT, instead of trying to bake up a ground-up open firmware, which would be wonderful but VERY difficult, better to just pick an existing firmware, like osmocom did with the (very old) calypso chipset. You would pick a firmware with very wide market penetration so that it is easy to get handsets, and then try to pick one that was easy to hack.
So ... whatever very widely adopted firmware has the most holes in it.
Hardly unprecedented. Cable modems work the same way. It seems to make good sense to me; the provider is responsible for the performance of the network, and the client interface is a part of that network.
The interesting thing was that several other people at the workshop were very gung-ho about how building their own phone in this way makes them in some way free from oppressive phone manufacturers. This didn't really strike me as quite correct given this is mostly just assembling parts from similar manufacturers.
Edit: sorry, wolfgke informed us that still we can't have the cake.
1. In a sea of people carrying iPhones and/or other various touchscreen smart phones, the minority of people using any kind of alternative seem to possess minscule flip phones with full-color displays. I haven't seen a candy bar phone in ages, unless it happens to be some venerable variety of nigh-defunct BlackBerry. I can't imagine this kit being resized down to a scale that would ever fit into what might be regarded as "common", especially since DIY kits need to be produced with parts big enough to man-handle, lest they risk total obscurity. No DIY kit will ever be successful if the parts involved are small enough to warrant a jewler's or watch maker's precision. Would you sell the kit with a loupe and tweezers?
2. Why should an unorthodox handset be suspicious? I guess it depends on the country one might live in? Also passing through customs and airport security, I guess?
I still use a candy bar Samsung with a slide-out keyboard.
Do you think it would be possible to gut the phone, and replace it with your own home-brew parts? Would the slide-out keyboard be likely to integrate into whatever DIY parts you select?
Though now that Nokia got sold, i have a relic!
In fact, I switched from Verizon to T-Mobile after a decade just so I could have a GSM phone. Haven't gotten the pieces together yet, but I'm so down with this because I totally want a wooden phone.
And all the hipsters will squeal with jealousy.