Beyond merely the frequencies themselves, there's a ton of logistics involved in things like harmonic or other interference and propagation, all of this related to power. However, it is increasingly such that a lot of devices are software-based radios, so changes to a lot of existing technology could be just a firmware upgrade, if possible.

Unfortunately, we are not yet at the point where a range of frequencies anywhere near as wide as discussed in the article can be supported by a simple firmware upgrade. RF receivers typically include a very narrow bandpass filter as the first component after the antenna. As spectrum becomes more crowded, these filters are even more critical to prevent saturation of the low-noise amplifier and unwanted products from the downconverter/mixer.

In my experience, the ADC is far from the limiting factor for linearity in an SDR receiver. The analog components up front (LNA and mixer, especially) are the limiting components. Broadband front ends make it even worse, because it's more likely you'll have an interferer passed through the filtering, even if it's nowhere near (in frequency) your signal.

Don't forget the antenna. In many cases, the antennas themselves have specific polarization and beamwidth/radiation patterns along with SWR that is directly tied to the frequency range of the antenna itself. SDR doesn't change any of this.

Yes, yes! Sorry to over simplify it. I guess my main point was mostly the first one, there's a lot to it, heh.

and the bp filter isn't tunable?

A cheap one isn't. An expensive one may be, but you could trade away a lot of performance to get that feature.

That's a pretty good idea: it's like public roads: normally, they're used by the citizens, but if there's a VIP guest or a military situation, the government can clear and use them for themselves only. Makes sense to do the same with frequencies (although it will be a bit harder to manage - like the EBS test in 1971 has shown)...

I worked on the first USA test of White Space WiFi up till last year. Pretty cool stuff, smart engineers. Right now it's being utilized for public services (cops, firemen, highway cams... a few other things that were a little more interesting but... NDA'd)

It's great to see more initiative on rolling out this technology.

Anyone know what kind of speeds could go over these frequencies?

The Shannon limit would be at most 2Gbps...

The channel capacity depends on the amount of noise present, not just the bandwidth. Specifically for a AWGN channel, bits per second <= W*log_2(1+SNR), where W is the bandwidth in Hz and SNR is the signal-to-noise ratio. For example, 5 Gbps has been demonstrated using only 100 MHz of bandwidth: http://en.wikipedia.org/wiki/LTE_Advanced

That's not right. it should be 2 Giga-samples-per-second. You can pack a large number of bits into each symbol.

Based on what?! OP's question wasn't very well defined and your answer is completely arbitrary.

2.2GBPS