Wednesday, February 4, 2009

Introducing the White Spaces Database Group



Since November's big vote at the FCC, some have begun asking when we'll start seeing consumer mobile devices take advantage of TV white spaces spectrum.

As the Commission made clear in its ruling, a working white spaces database must be deployed in order for consumer devices to be available in the market. Before sending or receiving data, devices will be required to access this database to determine available channels in the vicinity. Combined with spectrum sensing technologies, use of a geo-location database will offer complete protection to licensed signals from harmful interference.

With this mandate in mind, this morning we joined Comsearch, Dell, HP, Microsoft, Motorola, and Neustar to launch the White Spaces Database Group.

In the coming weeks and months, members of the group will be offering to the Commission their perspectives, and some specific recommendations, about the technical requirements we would like to see adopted for the database. Many of these specifications ultimately will be heavily technical; put simply, we'll advocate for data formats and protocols that are open and non-proprietary, with database administration that is also open and non-exclusive.

We don't plan to become a database administrator ourselves, but do want to work with the FCC to make sure that a white spaces database gets up and running. We hope that this will unfold in a matter of months, not years.

Stay tuned to this blog for further updates on the group's work.

6 comments:

reinharden said...

Google should reconsider being the database administrator. Assuming that the whitespaces work is allowed to run its full course, eventually we'll require a centralized database capable of dealing with ephemeral spectrum auctions. And, at first glance, there seems to be relatively little difference between the technology behind those leases and the technology behind AdWords (well perhaps some location-specific issues).

And frankly, I've a lot more confidence in private industry putting something together quickly and efficiently than the government. Especially if we're trying to scale the auction database internationally.

reinharden

Stephen said...

I'm not sure why there has to be one place that has *the* definite db. I'm sure that at least some kind of distributed decentralized aspect would be very helpful. At least as distributed as dns -- most queries are location sensitive so most places just having a db that has part of the db that answers from cache most of the time would work wonderfully.

Also since the FCC Universal Licensing System already tracks licenses, they seem to be a very good place to host such information. Also nbc, cbs, abc, etc seem to have a vested interest in not having this interfere with their signals. krem and other local people as well.

I do agree with google : data formats and protocols that are open and non-proprietary, with database administration that is also open and non-exclusive.

I'd like to see an rfc or similar that is free from patent concerns. I'd also like to see reference implementations or two under a floss license.

One place it seems that it might make sense to have this information is in conjunction with the routing daemon/wireless neighbor discovery place. A wireless system might work better if it knew not only the frequencies in use in it's own area but within a hop or two as well. I'm assuming omnidirectional antennas and bidirectional traffic on a used frequency and some kind of ad hoc wireless routing would be desired.

Also I don't know why you mention auctions, but you are right that more than one government entity in the internation scene would want to run their own databases.

reinharden said...

First, there's nothing that keeps "the definite database" from being distributed and decentralized. There are, by way of analogy, root DNS servers. Most ephemeral spectrum access would be quite locality specific and would thus be must efficiently dispatched via a relatively local server.


I mention auctions because the end-state of Dynamic Spectrum Access (DSA) is that entities that "own" spectrum in particular localities or at particular times will eventually choose to "lease" that spectrum. Spectrum leases are a part of commercial spectrum access already; however, they often take months to negotiate and formalize. DSA leases will eventually be micro-leases, for very short durations or for very small physical locations (aka low power only).

I will grant that this is only tangentially related to white space devices; however, industry would be better served if there were a single mechanism that had to be checked before initiating spectrum access rather than having to build lots of "smarts" into the devices that figured out on their own which access mechanism was appropriate for the current time, access duration, location, power level, waveform, polarity, proposed frequency, etc.

Also, having a "centralized" point somewhere in the system allows for dynamic spectrum access to bands otherwise reserved for emergency or government services. In the event of an emergency, a higher priority "policy" can be propagated into the spectrum access servers temporarily rescinding non-ES utilization of the spectrum of interest in the specific areas of need.

Finally, with regards to Stephen's comments, please don't presuppose omnidirectional antennas, bidirectional traffic, or a universal wireless mesh running similar protocols. The end goal of open spectrum access is to create a universe of new technologies not all of which will interoperate with one another, but none of which will interfere with one another. So all devices possibly need to be capable of communicating with a spectrum policy manager, but they don't need to be capable of communicating with one another.

[The end goal of truly open spectrum access of course obviates the need for a spectrum manager; however, as we won't reach that end state for many years, if ever, we're best served coming up with as lightweight of a spectrum manager as possible).


DARPA's XG program, in additional to producing functional dynamic spectrum access radio prototypes, touched upon a lot of the above concepts and included initial work in developing a machine actionable (aka "cognitive radios") spectrum policy ontology. Much of the resulting technology was deeply vetted by the FCC and the NTIA (and several of their international equivalents) and received at least some level of review by Google, Intel, and Microsoft (among others). Arguably, DARPA XG laid the groundwork for Google's and Microsoft's White Spaces efforts.

reinharden

bogner said...

In establishing a data base, how
will you meet Para. 15.712(f)(2)?
(2)(A) may be an error, but (2)(B)
does not define "occupied". What
about partial occupation eg a TV
station or LPTV that covers part of
a city? This is meant to clear a
channel for wireless mikes, but it
can be totally restrictive in big
cities like NY, where no channel
seems to clear. Richard D. Bogner

Mimi said...

Is this White Spaces Database Group still open? We, as a small company working in this area would very much like to join and contribute to any work necessary to make this happen.

Anyone knew if this group still exists and how we can join the effort?

Thanks very much in advance...Mimi

forcefactor said...

Gigantic marketing machine to drum-up their stock price, they failed to deliver on opensocial and gphone, now they have something new to talk about.
Force Factor