Google wants permission to build an experimental wireless 2.5GHz network to test secret technology using 50 base stations and 200 devices.
The network will squeeze into two 2MHz-wide slots, starting at 2524MHz and 2567MHz, and will use both directional and omni-directional antennas mounted on walls and ceilings across the Google campus over the next two years - but what kit or protocol it will be using remains a mystery.
A filing to the FCC requesting permission to operate the experimental radio network was spotted by one Steven Crowley, who promptly pointed the Wall Street Journal in the direction of the much-redacted documents (one in HTML, the other in PDF, less interesting than one might hope).
So we don't know who's supplying the kit nor what Google will be using it for, only the bands in which it will operate and the location where it will be deployed, which leaves us in the comfortable realm of idle speculation.
The bands fall within the "3G Expansion" band, generally referred to as 2.6GHz, which are being auctioned off around the world for 4G (LTE) deployments. LTE can be squeezed into 2GHz, but the band being used by Google is paired with one much higher up the spectrum that's almost certainly a red herring.
What's more likely is that the band was a convenient one for testing, and the testing is most likely to be in White Space database techniques which we know Google is actively involved in developing.
White Space databases keep a list of locally available radio frequencies, supplied by the FCC and doled out to enquiring devices in exchange for GPS coordinates; short-range gadgets, such as wireless mics and small-scale networking kit, can therefore use the databases to pick unused frequencies in the area in which to operate.
Google is one of the companies planning to run a White Space database in the US, but most of the companies involved are planning to add some intelligence to the process in the cause of differentiation.
Spectrum Bridge, for example, has already suggested polling devices for reports on how successful their White Space communication is with a view to refining its own data on-the-fly. Most companies are also planning to log issued frequencies in order to avoid giving the same band to two devices in the same place (though if they're using different databases that will still be possible).
Google will want to be doing all that, and more, for which it will need to refine techniques for detecting interference, logging usage and coping with mobile devices, all of which can usefully be developed with a pair of 2MHz channels at 2.5GHz.
This is just guessing of course - perhaps the Laser Tag team at Mountain view just fancy having their own comms network; with Google one can never be sure, but White Space databases are a long-term investment and it would be nice to think Google is doing something beyond working out how to make us click on adverts more often. ®