Towards “A Theory of Spectrum Zoning”

Spectrum zoning is the problem faced by regulators: how to allocate different bands to different rules for radio operation. This problem has largely fallen through the cracks since to the extent that they even recognize the problem’s existence, traditional communities have viewed this as “somebody else’s problem.” Engineers/information-theorists assume that economists/lawyers will handle this and economists/lawyers put their trust in some varying combination of engineers and “the market” viewed as a pseudo-religious entity. This preliminary investigation puts forth a semi-Rawlsian perspective on zoning: viewing it as a robust optimization problem (max minmax for now) in which zoning decisions are made at a slow time-scale in the face of both natural variation in the suitability of different bands for different uses and significant uncertainty regarding user preferences. Consequently, it can make sense to choose band-plans that are not on the Pareto frontier: but only if their runtime flexibility is more valuable than the performance overhead they impose. We examine the issues in the context of a toy model of TVs and wireless ISPs.