It's a mighty goal and you're probably more than capable developer but even bigger organizations have failed to produce working phone sw stacks while leveraging open source components.
Therefore while it's a great tinkering project I'm really sceptical about the actual value this could have for "end users" unless the value is the tinkering itself ;-)
+1 for this. A GSM/3G/LTE stack isn't something you can write in an afternoon. I'm on such a project for my company for several months now and I'm probably still a year away from just being able to push a single IP packet through it.
That's why I won't be doing it myself - I'm just going to provide the platform, there are already people working on an open-source baseband, and they're much smarter than me =)
Therefore while it's a great tinkering project I'm really sceptical about the actual value this could have for "end users" unless the value is the tinkering itself ;-)