Spectrum Sharing in White Spaces

Thesis (Ph.D.)--University of Washington, 2015Demand for wireless Internet traffic has been increasing exponentially over the last decade, due to widespread usage of smart-phones along with new multimedia applications. The need for higher wireless network throughput has been pushing engineers to expand network capacities in order to keep pace with growing user demands. The improvement has been multi-dimensional, including optimizations in MAC/Physical layer for boosting spectral efficiency, expanding network infrastructure with reduced cell sizes, and utilizing additional RF spectrum. Nevertheless, traffic demand has been increasing at a much faster pace than network throughput and our current networks will not be able to handle customer needs in near future. While assigning additional spectrum for cellular communication has been a major element of network capacity increase, the natural scarcity of RF spectrum limits the extend of this solution. On the other hand, researchers have shown that licensed spectrum that is owned and held by a primary user is heavily underutilized. Examples are TV channels in the VHF/UHF band as well as radar spectrum in the SHF band. Hence, a more efficient use of this spectrum is to permit unlicensed users to coexist with the primary owner, i.e. to share the same spectrum if it is not utilized at the current time/location. Spectrum sharing has received considerable attention in recent years for its potential in improving network capacities. Especially, with formal opening of TV-band frequencies by FCC to unlicensed operation as well as proposals for radar-bands to be opened in near future, wireless industry is also showing a great deal of interest in these unlicensed bands. The main challenge behind spectrum sharing is detection of spectrum opportunities, known as white spaces, by the secondaries. Classic methods are based on spectrum sensing which requires highly sensitive receivers. Newer methods, that are currently proposed by FCC for TV white space spectrum, are the so-called DBA approach in which a centralized database determines availability of shared spectrum at any location and time. This work is focused on the latter method. In this work, we have focused on major challenges in spectrum sharing in the white space spectrum. First, the available capacity that is opened through TV white space spectrum is not clearly understood. We define a mathematical framework to evaluate achievable white space capacity in the TV band as a function of location, FCC regulation and secondary network parameters. We use this framework to simulate available TV white space channels and capacity over the entire United States and explore its dependency on various parameters. Second, unlike licensed spectrum, available TV white space spectrum is significantly location-dependent. The number of channels as well as their quality (noise and interference floor) can severely change from place to place. Therefore, designing a cellular network that is based on spectrum sharing requires special channel allocation algorithms to consider these variations in to account. We define the problem of channel allocation in a spectrum sharing scenarios and explore various solutions. Third, spectrum sharing rules in dynamic scenarios such as radar bands are not defined. Due to rotation of radar antennas, the available spectrum is time-dependent and coexistence scenario depends on how much information about the time-varying primary user (radar transmitter) is available to the secondary user. We introduce a spectrum sharing paradigm with rotating radar transmitters that models radar target detection operation as well as random distribution of secondary WiFi transmitters in the environment. We use this model to calculate protection region for the radar as well as achievable throughput. Fourth, the lack of suitable SDR hardware has made evaluation and prototyping of available white space spectrum very challenging. We develop a SDR platform for operation of WiFi devices in the UHF spectrum from 300-MHz to 3.8-GHz band. This platform is then utilized for development of fully functional WiFi-like networks in UW campuses to evaluate white space opportunities in the UHF spectrum and to provide Internet connectivity to end users