Steps toward a national unlicensed spectrum prediction engine.

Sandvig, C. (2007). Bad Neighborhoods of the Electromagnetic Spectrum: A Method for Predicting the Deployment of Unlicensed Devices. A Paper Presented to the 35th Telecommunications Policy Research Conference (TPRC) on on Communication, Information, and Internet Policy, Arlington, Virginia, USA. (Full text online in PDF.)

Abstract:
An uncontroversial prediction about the future of electromagnetic spectrum use is that the developed world will soon be filled with far more license-exempt low powered devices. Today's widespread deployment of Wi-Fi and Bluetooth exemplify this trend. Although in many ways a consumer's Wi-Fi enabled portable device couldn't be more different from a licensed television broadcast tower serving a major city, the mental frameworks used in policymaking to reason about the use of spectrum remain in thrall to high-powered broadcast towers. For example, some proposals for differential power limits divide the nation into two kinds of places: rural and urban. Often policy action about spectrum is never based in any empirical knowledge about use or congestion. This paper demonstrates the problems with these approaches by reporting on a longitudinal empirical study of low-powered unlicensed spectrum use (specifically, Wi-Fi deployment) in Illinois and California over four years.

We combine these data with the US Census to construct a national unlicensed spectrum prediction engine based on the extensive literature on the diffusion of innovations. Given any address in the US, our service -- named the "RED Project" for "Rendering Electromagnetic Distributions" -- predicts Wi-Fi deployment and (using a Macromedia Flash front end) dynamically renders the regression equation predictions on a street map or satellite photo at a variety of zoom levels.

Aside from the possible utility of this prediction service (for example, at predicting congestion), surprisingly, our findings also overturn a number of habits of thought about spectrum. For a variety of reasons, some densely populated city blocks are devoid of Wi-Fi and some rural areas are approaching congestion failure. As the diffusion of innovations literature predicts, adoption cannot be predicted simply by income or by population density. Some "bad" neighborhoods in US cities are "good" neighborhoods of the spectrum--that is, they are devoid of congestion--and vice-versa. Some neighborhoods that communication companies think of as a very desirable urban market are almost devoid of signals. We conclude by trying to come to terms with the variegated new spectrum topography of low-powered consumer devices.

Tags: diffusion, electromagnetic-spectrum, unlicensed, wi-fi, wireless   (See all possible tags)