@cognitive radio is coming
Yes, it is coming, but there are serious issues in making it work acceptably in the bands that have other legacy systems. The key problems that I see (and have already mentioned) are:
1) How sensitive do you need to be?
This of course depend on the type of antenna / placement / system that is currently in use. Basically you need to be sure that you can detect the same signal before broadcasting something that will wipe out the other users.
This, of course, is the primary reason why band planning is such a good idea. Unfortunately that goes against the 'free market' religion of the USA and big business. They do not care about minority users or overall quality, only about monetising the spectrum.
2) How do you revoke the use of any frequency ranges that are causing problems one the devices are deployed?
The key point here is that once they are being sold, you can say goodbye to a return to previous use. The ‘deregulation’ of a given spectrum is essentially a one-way process, as even with stopping the sale of a class of device, once they are in use by millions of non-professional users, it will take years, if not decades, for the usage to drop to near zero. If you made a mistake, then really there is no going back.
Now there are good arguments that TV is crap and not that worthwhile usage, but its there and tens of millions of users have an expectation of it, as do the networks who support it.
The ideal cognitive radio use is in a band where it is already deregulated, so there are no expectations of a given quality or access level. If you do plan on deploying it in a band where there are expected issues of band access for intermittent uses, or high gain antennas and sensitive receiver and so on (e.g. wireless mics and rural TV as currently deployed) then you should have devices that operate both cognitively and with some centralised permission system.
In essence, each radio should be required to have a method of regularly obtaining a ‘certificate’ to use given frequencies (perhaps specific to its serial number), and be aware of its geographic location (may be tied to the certificate generation). If it can’t obtain the valid certificate, or if it is revoked due to complaints over certain frequencies, then the radio must stop using some or all of the band frequencies.
This would mean having some sort of low rate network (or side channel) that manages the system centrally, which again is something the ‘free market’ mob would vehemently oppose, but it would allow both spectrum efficient re-use where it is genuinely unused, but with a guarantee of revoking troublesome devices or protecting users against poor detection capabilities.
However, I seriously doubt that the key proponents here (MS & Google) or the USA government (i.e. FCC) are likely to listen to technical sense where it is going to cost money to implement. Even if some of this (e.g. geographic / cell location) could be very useful for some types of service, advertisement targeting, etc.
Again, if I see these point answered, I would support it. But I have my doubts...