Wireless spectrum regulation is an area of increasing interest, complexity, and importance. After decades of single-purpose, exclusive spectrum allocations, the Federal Communications Commission (FCC) brought about the era of dynamically shared spectrum with their landmark ruling in 2008. Although unlicensed spectrum—“free for all” spectrum such as the 2.4 GHz band—is universally agreed to be critical to economic development, there is scarcely enough to meet current demands. WiFi could never have succeeded without sufficient unlicensed spectrum, yet its success has “filled up” much of the available spectrum. In an effort to address the shortage, the FCC ruled that unlicensed devices are now allowed to operate in portions of the television bands.
These “unused” portions are called the television whitespaces (TVWS). Due to the excellent propagation characteristics of the TV band frequencies, a commonly discussed application for TVWS is rural broadband, i.e. using wireless to provide last-mile access to high-speed Internet. Typical TVWS trial deployments have demonstrated the use of TVWS spectrum for backhaul, e.g. to connect school buildings in South Africa or for remote video feeds at a zoo. Others envision Internet-of-things (IoT) applications such as garbage can sensors. The critical piece, though, is that TVWS regulations have to have sufficient flexibility to encourage a wide variety of applications and innovation.
This thesis largely approaches spectrum regulations from a quantitative perspective. It begins by quantifying the current TVWS opportunity, finding that rural spectrum is abundant while urban spectrum is somewhat scarce. But spectrum alone does not tell the whole story so we next look at the data rates achievable with TVWS under various scenarios. The results of this exploration highlighted a shortcoming of the current regulations: lack of sufficient transmit power in rural areas. This led us to explore dynamic power limits for TVWS, showing that intelligent choices for these limits can dramatically improve the utility of whitespaces.
Next, we compare single-use spectrum allocation with whitespaces in the TV band context. We show that allowing the use of whitespaces is superior to repurposing the band from the incumbent’s perspective. We also use a novel data set from the FCC to further compare various spectrum allocation schemes.
In addition to dynamic power limits, we explore other changes to the whitespace ecosystem that we believe will increase utility and promote innovation among whitespace devices. In particular, we carefully decompose the current whitespace architecture into its constituent parts. The de- composition makes testing and certification easier while at the same time allowing more flexible composition of the components. For example, our proposed architecture allows a smartphone to provide a whitespace device with the information necessary to begin transmission in the TVWS; under the current paradigm, only another whitespace device can provide this support. We also demonstrate how the proposed architecture can enable the use of whitespace devices indoors by allowing a variety and combination of location information sources rather than depending solely on GPS which typically does not work indoors.
Whitespaces are not unique to the TV bands. In fact, much of our spectrum lays dormant, particularly in rural areas. While the FCC is currently looking into using whitespace in the 3.5 and 5 GHz bands, we explore the use of whitespaces in the cellular GSM bands. This work dovetails with existing work on community cellular networks (CCNs), community-run cell phone networks in hyper-rural areas, which chiefly lack spectrum in which to legally operate. Our work on GSM whitespaces details how allowing CCNs to use carrier-owned spectrum in rural areas is a win-win-win for regulators, wireless carriers, and citizens alike. We use an existing CCN implementation to demonstrate and evaluate our ideas.
One of the most important commonalities in the aforementioned work has been the quantitative approach. Prior work principally explores the space either in a purely theoretical way or with limited real-world data. In contrast, the work in this thesis is almost entirely based on analysis of real-world data, complementing the theoretical work. In an effort to encourage others to explore spectrum regulation from a quantitative perspective, I have made my source code publicly available. Part VI discusses the overall architecture of the Matlab and Python code, as well as some of the design decisions that were made.
Since whitespaces present both technical and political problems, we have had the opportunity to submit various official comments to both the FCC and Ofcom. In the spirit of transparency and completeness, we include these comments in Appendix A.
