Communications Insights: Trends and Cool Stuff

Spectrum band plan created by price-guided mechanisms

Winston Churchill famously said, “democracy is the worst form of government except all the others that have been tried.”  Perhaps the same can be said of spectrum auctions.  Auction mechanisms have been used starting in New Zealand in 1994 to award spectrum licenses to those who have the highest monetary value. Spectrum auctions have generally been highly effective, with the occasional failure.

Despite their success, auctions have some notable drawbacks such as the so-called winners curse and the fact the up-front license fees require spectrum users to raise capital beyond the princely sums necessary to build a wireless network – a barrier to entry.  However, auctions are far better than the administrative processes which have been used for nearly a century to determine spectrum assignments.  Administrative decisions tend not to be economically efficient because the regulator has limited access to information which market participants would be more able to amass and utilize. There are also problems of political independence and of regulatory capture.

While auctions have been used to determine who gets spectrum rights, they have not really been used to determine the contours of those rights.  These contours are still determined through administrative decisions.

I have just completed a major study on next generation spectrum regulation which can serve as the basis for removing certain barriers to spectrum access, allowing more effective sharing and efficient allocations.

I can think of no reason why a properly designed auction could not determine not only who gets the spectrum rights, but what those rights are.  (Think of it this way: an auction on eBay for a car could determine not just who gets the car, but the color of the car and whether it comes with, say, leather seats or alloy wheels.)  I built a mathematical model of a next-generation spectrum auction using the Shannon-Hartley Theorem as a means modeling behavior by valuing the spectrum when considering the actions of other would-be users.  In my model bidders could express their demands for not just bandwidth, but power, modulation, underlay/interference, and other characteristics.  When I ran an MS Excel-based version of the model, the result was a mix of high and low power uses in the winning bids.  The low power bidders (similar to UWB spectral densities) could in a second round be aggregated into some form of licensed commons with the coordination protocol determined in that part of the auction.  The outcome would resemble a shared use or common arrangement where no one party controlled the spectrum.  However, the most interesting thing was that because bidders could obtain spectrum allocations that more closely fit their needs, more than 40% of the spectrum bandwidth available in the auction was left unsold.  This spectrum was valued by the market to be best allocated to either public sector use or even low- to mid-power unlicensed use.

Insight:  You cannot see, touch, taste, smell, or hear radio spectrum.  Spectrum is not a thing; it is an idea – a legal and engineering construct that explains a physical phenomenon and helps us arrange our behavior accordingly.  That fundamental physical phenomenon is the fact that when electromagnetic waves are: (1) harmonic in frequency; (2) incident in time; and (3) alight on the same reception device, the ability of those waves to be used as information carriers is degraded.  This deleterious effect is known to us as interference.  Without some form of intervention, it is impossible to exclude or limit the use of a common resource such as spectrum. Without exclusion, users consume the spectrum without regard to fact that their usage causes the deleterious effect of interference for other would-be users.  Policies which help to mitigate inference with the least amount of effort will be the most socially beneficial.