Implementing 3G is more complicated than 2G by an order of magnitude (and OsmocommBB is still far from having their 2G support complete). Plus Calypso is one of the only baseband processors out there that accepts unsigned firmware (which is exactly why it's OsmocomBB's target) and 3G on Calypso is not possible.
A major irony is that, in principle, a 4G radio could be made very simple, with all the protocol processing up in the application processor, and sharing code with SIP VoIP and UMA capability in the phone OS. But, as far as I know, no 4G radios are designed that way. Maybe when 3G starts to go away.
I wasn't thinking of pulling the layer 1 up into the AP. But I don't know... do you have such requirements anymore? You have real time requirements for voice calls in 2G/3G hard handover. If you treat everything as if it were a vertical handover, I think you can even wash that requirement out. And you can fudge the perceived voice call continuity.
HARQ is a soft real time requirement, and 4ms is within what you can get in Linux interrupt response. That is, it's not going to force you to do hard real time hacks in your AP.
Anyway, the real question is: Can you build a 4G radio you can trust because it does little enough that it can have open firmware and driver and protocol software that takes less than 10 years to write? I think the answer is "Yes. Maybe simpler than 2G." Or at least "Much more likely than with a 3G radio."
Alternatively you can go the Neo900 route, where the baseband processor is in fact a separate USB module which can be separately turned off - just placed on the same PCB.
Why do people talk about the Neo900 as though it's a real product that exists or can be bought? Has anyone seen any evidence it's anything but vaporware?
Well, I see the work being made on it by myself and already watched similar project (GTA04) reaching its goals. Plus that's me who you can blame personally for any delay with the Neo900 website content or newsletters... ;P (and if you find linked talk hard to listen, you can blame me for that as well ;])
it's been well over a year since the neo900 project was announced, and I've never seen any goldelico project actually ship more than a handful of prototypes.
I really hope the neo900 succeeds, but sending money to goldelico is generally a bad bet.
Neo900 has been announced at the moment when all interested parties agreed on "it seems feasible, let's do it". There was no prior investment, the development could start only after raising funds. It simply takes time - especially when organizational burden turned out to be a bit more challenging than initially assumed.
All of Goldelico customers who paid for GTA04 already received one long time ago. I don't count latest fundraiser for its new revision[1], since it will be fully refunded if not enough people will preorder to actually produce a new batch.
Also, Goldelico is "just" a contractor doing work for Neo900 UG, which is the actual entity running the Neo900 project.
But I think we're getting a bit off-topic here. #neo900 on freenode might be a better place for that ;)
Are there actually any open gsm capable chips? I'd expect there to be other transmitted codes for certain functionality which isn't as readily documented but still abuseable.
TI Calypso, with OsmocomBB[1]. Haven't you seen the link you're commenting under? :P
[1] of course all feature and legal restrictions apply, so don't get too excited about "yay free baseband finally!". It's a amazing research project nevertheless, which together with OpenBSC already gave us incredible insights on GSM security. All hail Harald Welte! \o/
All the various unbranded/generic/clone handsets coming out of China are likely illegal to use on any networks too since they haven't been given the certification, but AFAIK they work fine since they adhere to the protocols and would probably pass the certification anyway. I don't know how far along OsmocomBB is, but if it is indistinguishable from a "real" phone as far as the network is concerned, then it'd be a similar scenario.
What the networks don't want - and the motivation for the certification - is for some device to misbehave (e.g. emit a continuous transmission) and interfere with the operation of other devices on the network.
Of the compatible phones, only Sony Ericsson J100a appears to support North American GSM bands, and eBay doesn't have any of it on sale, at least not at the moment.
