Friday, August 15, 2008

Use it or lose it

I’ve been thinking about how a trademark-inspired approach might change how radios are regulated. (See my February post on De-situating Spectrum; a conference paper will be available soon).

One of the interesting aspects of trademark is you own it as long as you use it. Wireless rights in a trademark-inspired regime would work the same way.

Under the current regime, many radio licenses are subject to build-out conditions that can lead to the retraction of a license if a network isn’t constructed to use it. (In practice, these conditions don’t really have teeth.) However, unlicensed operations don’t have build-out conditions. They should – though they will differ from licensed ones, since there isn’t a license holder to go after.

The idea is that the regulator (the FCC in the US, for example) would make clear that an unlicensed allocation is “use it or lose it”: if the band is not used in the way promised by manufacturers and contemplated by the regulatory decision, the rules will lapse. The unlicensed allocation may be replaced by something else.

Here’s a straw man.

“Use it” tests: At least three devices should have obtained certification by Year Three after allocation; at least 500,000 devices/year should be sold in Year Five. If not, the allocation is deemed not be used. If device sales fall below 50,000/year for three successive years at any time after that, the allocation will also be deemed to be unused.

Define the “trigger date” for losing the allocation as the year in which the allocation fails the “use it” test.

“Lose it” consequences: At the trigger date, the FCC will cease certifying devices; this means that no new devices will be sold. However, the (few) intrepid souls who bought devices should be able to continue using them, at least for a while. Let’s assume devices have a useful life of five years; unlicensed operation will continue to be legal for that period. At that point (five years after the trigger date) the allocation will lapse, and the FCC may choose to issue new unlicensed rules, auction the spectrum, or do nothing.

The parameters proposed above were by way of example. One would have to be sure that the “use it” test doesn’t discourage innovation and investment, and that the “lose it” consequences don’t drain the bathwater too soon; also, both should be as resistant as possible to manipulation by opponents.

Rather than an all-or-nothing loss, the FCC might reduce the size of the allocation as an interim step, as it did with UPCS. This is particularly easy if the unlicensed devices are reprogrammable over the air.

A regular review of operating parameters should be built into all unlicensed rules, even if the allocation is being used by the above criteria. If the rules change, there will be a two-step phase in that parallels the “lose it” consequences. In the first five years, legacy devices will be grandfathered in, but no new certifications will be issued for legacy operation. This should allow people to extract economic value from their investment. Five years after the rule change, though, operation of legacy devices will no longer be legal.


One of the most compelling arguments against unlicensed white spaces is that allocating them would preclude ever cleaning up the “UHF mess”: valuable spectrum is under-used because cheap receivers require very large frequency gaps between transmissions to avoid one TV broadcast from interfering with another. For only a few cents in additional receiver hardware, transmitters could be packed more closely, allowing additional systems to operate. The other part of the “mess” is that fewer than 15% of Americans get their TV over the air; most of it is piped into homes via cable. High power transmitters are running night and day – precluding other uses – for the benefit of a small number of people.

Unlicensed advocates contend that Wi-Fi like devices should be allowed to operate in the frequency gaps. These devices will be smart enough not to transmit in the same channel as TV broadcasts, and their own transmissions will be so weak that they don’t cause interference to TV stations in adjacent channels.

Opponents say that this will set the mess in concrete. If a way is ever found to persuade broadcasters to cease transmission – a big IF, given that broadcasters have immense political power, and their imminent demise has been predicted for decades – then the new owners of the ex-broadcasting channels will have to contend with unlicensed “neighbors”. Once the technical parameters are set, the secondary unlicensed users become squatters and can’t easily be evicted. (Military transmissions from time to time interfere with garage door openers, e.g. in Denver and Quantico. The military has priority, but the manufacturers just ignored this. . . the military backed off, it seems ) Rather than have a pristine post-broadcast band to work with, new licensees will have to co-exist with unlicensed devices.

I don’t have much of a problem with this, provided the unlicensed devices are actually widely deployed. The problem arises when there are just a few scattered users that are blocking a new rights allocation (as a handful of over-the-air TV viewers are doing right now). Given the FCC’s administrative proceedings, it is exceedingly hard to get allocations changed. The 20 MHz unlicensed PCS (UPCS) band was allocated by the FCC in May 1995 but never really used. It took almost ten years for the FCC to re-allocate the lower half of the band to licensed use, and tweak the rules in the upper half to allow more technologies to operate.

There is also a more general objection to unlicensed: it’s very hard to back out of a bad decision, as the UPCS experience shows. Licensees that have flexible-use rights can change their mind about technology and rebuild their systems, if necessary by negotiating variances in spectrum rights with those in adjacent parcels; Tom Hazlett gives the example of the conversion from analog to digital in the cellular bands. In contrast, unlicensed allocations once made are slow to change. Built-in sunsets such as those I propose above could fix that problem.

No comments: