Browsing by Author "Zavodovski, Aleksandr"

Transition to the Digital Television (DTV) has freed up large spectrum bands, known as a digital dividend. These frequencies are now available for opportunistic use and referred to as Television White Space (TVWS). The usage of the TVWS is regulated by licensing, and there are primary users, mostly TV broadcasters, that have bought the license to use certain channels, and secondary users, who can use channels that primary users are not currently utilizing. The coexistence can be facilitated either by spectrum sensing or White Space Databases (WSDBs) and in this thesis, we are concentrating on the latter. Technically, WSDB is a geolocational database that stores location and other relevant transmitter characteristics of primary users, such as antenna height and transmission power. WSDB calculates safety zone of the primary user by applying radio wave propagation model to the stored information. The secondary user sends a request to WSDB containing its location and receives a list of available channels. The main problem we are going to concentrate on is specific challenges that mobile devices face in using WSDBs. Current regulations demand that after moving each 100 meters, the mobile device has to query WSDB, consequently increasing device's energy consumption and network load. Fast moving devices confront the even more severe problem: there is always some delay in communications with WSDB, and it is possible that while waiting for the response the device moves another 100 meters. In that case, instead of using the reply the device has to query the WSDB again. For fast moving devices (e.g. contained inside vehicles) the vicious loop can continue indefinitely long, resulting in an inability to use TVWS at all. A. Majid has proposed predictive optimization algorithm called Nuna to deal with the problem. Our approach is different, we investigate spatiotemporal variations of the spectrum and basing on over than six months of observations we suggest the spectrum caching technique. According to our data, there are minimal temporal variations in TVWS spectrum, and that makes caching very appealing. We also sketch technical details for a possible spectrum caching solution.