The radio spectrum is quite broad from a few MHz up to hundreds of GHz but in practice the usable spectrum is constrained by two (main) factors:
- The lower the frequency used the longer the antenna needed to capture the signal (remember the antennas of the old TV we had in the living room?) but also the better the propagation (the signal can cover a broader area and is less disturbed by external phenomena, like rain, foliage...).
- The higher the frequency the smaller the size of the antenna, so you can stuck it into small things, but the less effective the propagation. Rain drops can stop the signal and you can only manage coverage of small areas (down to few metres if you go really up in frequency).
In addition the higher the frequency the faster the chip processing the radio signal should be. Thanks to the amazing progress of electronics (courtesy of Moore's law) we have chips that can process higher and higher frequencies and we have seen wireless systems moving from around 800MHz in the first generation to over 2.5GHz in the fourth generation (and even higher with WIMAX).
Now with 5G the capability to move further up, from the processing point of view, is there. And indeed, although there is not a standard yet, there are talks on using double digits GHz, even though massive penetration of 5G requires frequencies below 6GHz.
So far the highest frequencies being discussed were 81-86GHz by China and US. Now, the FCC Chairman in a recent talk proposed to free the spectrum above 95GHz for 5G letting the industry come up with proposals on how to use it.
Clearly, at these frequencies you cannot imagine to cover a city, the propagation range is just no there. However, one could use these frequencies to provide very high bandwidth in very small areas. An example? A car may have a halo made of 95+GHz. Such a communication bubble can support bandwidth in the order of several GBps, enough to download a movie as it coast by a toll boot. Or, several cars in close proximity (which is not unusual in our cities) can exchange plenty of date with one another.
These examples are just from the top of my head, and may be they do make any sense. I am using them just to give and idea of possible applications. Actually, in his speech the Chair of FCC explicitly said that he does not know what industry could do with that spectrum but since no one is using it why not opening it up and let researchers and industry find useful applications?
If you think about it this way of reasoning is quite a change from the past, when regulators where imposing strict barriers on the use of spectrum enforcing the specific services that could be delivered (e.g. for television broadcast and not for telecommunications). With 5G the possibility of having the terminals deciding on what spectrum to use and adopting a dynamical allocation of spectrum we are in for a revolution.