Going beyond ordinary auctions, an FCC working paper proposes a new approach based on mathematical game theory.

The FCC is looking at a new way of allocating radio frequencies.

One of the FCC’s most difficult jobs is squeezing an ever-growing number of users into a fixed amount of spectrum. The variety of radio-based applications keeps growing, but the amount of spectrum suitable for each application is more or less fixed. True, engineering advances keep opening ever-higher frequencies, but the physics of radio waves limits those very high frequencies to a relatively narrow set of applications. Most new spectrum users must fit in and around the incumbents.

The parts of the spectrum best suited to most applications, very roughly from 30 MHz to 30 GHz, are complexly occupied. A band in this region may have primary and secondary allocations, under which secondary users are obliged to protect the primary users from interference, and must accept interference from the primary users. There may be multiple co-primary and/or co-secondary services, with those at the same level required to protect one another. The band may also have one or more types of unlicensed users, who must protect everybody else except each other, and must accept whatever interference comes their way. Much of the FCC-regulated spectrum is also shared with the U.S. government, which adds further layers of complexity.

Yet new entrants keep knocking at the FCC’s door. There is no one process for accommodating them. Instead the FCC staff conducts a proceeding specific to the individual frequency band and the individual application, guided in part by submissions from the would-be entrant and the incumbents. Rarely, the FCC can auction off a completely empty band. The conversion to digital TV emptied out the 698-806 MHz stretch, for example, creating one such opportunity, and possibly the last. Much more often, the FCC auctions a band encumbered by existing users. Or else it authorizes new types of unlicensed operation under increasingly complex rules, calculated to ensure a very low risk of interference to users of higher rank. The recently adopted “white space” rules illustrate an extreme case.

Just accounting for other users of the band to be shared is complicated enough. But that may be only part of the problem. Equally important are users of the adjacent bands, and sometimes even other bands farther away. The recent problems of LightSquared are an object lesson here. LightSquared sought to use its mobile satellite spectrum for terrestrial communications via towers, much like cell phone towers. But LightSquared’s frequencies were close to the frequencies that GPS satellites use to transmit data to the ground, the data by which a GPS receiver determines its own location. LightSquared did not transmit in the GPS band. But GPS receivers, like many other types, tend to be promiscuous: most are sensitive not only to their intended frequency, but also to other frequencies nearby – including those licensed to LightSquared. GPS users objected that LightSquared’s tower-based transmissions would drown out the satellite signals. The FCC ultimately agreed, and forbade LightSquared from proceeding.

The FCC has long looked for a more systematic way of handling new entrants. Auctions originated as one such idea, on the principle that the parties who put the most value on a piece of spectrum would also put it to best use. The results have been mixed: highly successful for relatively unencumbered spectrum for mobile applications, moderate success for spectrum burdened by same-band incumbents or sensitive users in adjacent bands, and a large-scale failure for the higher frequencies most suited to fixed use. Another early idea would have used the concept of “interference temperature” to maximize transmitter power while monitoring actual interference at receivers. This one looked great on paper, but bogged down so badly in the details of implementation that the FCC abandoned it without a trial.

Now the economists at the FCC have come up with another approach. Like auctions, it is based on the concept of markets, but with a twist. Market theory, at least in our own undergraduate days, made unrealistic assumptions: that everybody in the market had perfect access to all relevant information, and that each participant acted with perfect rationality to maximize his own benefits. Some economists have modified the first assumption by calling on the branch of mathematics called game theory. Here each participant may be lacking key information – and in particular, may not know the strategies in use by the other participants. The goal again is to find a strategy that maximizes one’s benefits, now assuming that the other participants will be picking their own strategies with the same goal for themselves.

A new paper from the FCC tries to apply these principles to the problem of allocating spectrum among various same-band and adjacent-band users. The paper, issued as part of the FCC Staff Working Papers series through a program managed by the FCC’s Office of Strategic Planning and Policy Analysis, uses an approach based in part on the work of mathematician John Nash, the subject of the 2001 film “A Beautiful Mind.” The authors acknowledge that assumptions of rational behavior, and of rational estimates of others’ behavior, may not hold up in the real world. But their approach may come a little closer to real-world conditions than a simple auction to the highest bidder.

We here at CommLawBlog are lawyers, not economists; we will leave an evaluation of the FCC’s paper to those more qualified to judge it. But we are pleased to see the FCC thinking creatively about the problem of working new entrants into increasingly shared and encumbered spectrum.