3G load-balancing, driven by the end user. WiFi as SDR radio middleware?

My last point on "shared" 3G dongle access, hooked up via WiFi , has got me thinking in a bit more blue-sky fashion.

Looking foward a few years, there will be lots of 3G connections, of varying speeds, prices and coverage. Connections and QoS will likely be patchy, especially indoors. But it will also be increasingly easy to share these connections, whether they're hooked up via tethered phones, 3G modems or other devices. Sharing could be via WiFi, Bluetooth, USB, or even "local 3G" via femtos.

I wonder whether we'll start to see devices being smart enough to load-balance between multiple connections, looking for the best speed / price / latency or whatever. We could even see "managed" load-balancing - for example a phone that uses simultaneous routing diversity through its own 3G, and via WiFi, someone else's in the same room. Good for resilience, and could also probably increase the aggregate throughput through the wireless spectrum as a whole. (Are you paying attention, regulators?)

This also hooks into the type of concept that Google hinted at a few weeks ago with its connection-auctioning patent.

This also fits with my general theme that handset & PC connection managers will get much, much smarter - and might even become the source of control, value and identity for the end user.

Apart from anything else, the idea of 3G connection-sharing, meshing or load-balancing (especially across multiple operators) would finally kill the notion that 1 SIM = 1 Access = Person = 1 identity. And it would render all these charts of "subscriptions" for mobile broadband even more vague than the current situation with GSM users with multiple SIMs.

It would also of course mean that whichever service provider had the best network at any given location would also have to contend with all the other operators' users acting as (effectively) parasites on the connection. Sounds like some sort of dynamic-MVNO billing and pricing model is needed - a sort of new 2-sided business model perhaps.....

[Edit - essentially I guess this is a user-driven model for software defined radio (SDR). The traditional view of SDR has a software-reconfigurable radio chip which can morph between different profiles and radio technologies. But if silicon's cheap enough and small enough, why not just have multiple radios and the device software deciding which to switch on/off on a case-by-case basis? A bit like the dual-WWAN notebooks I mentioned yesterday. Clearly there's a lot of constraints here - device size, cost, limited # of radios that could be supported. But using WiFi as licence-exempt "radio middleware" linking to other devices could partly solve that.]