Towards a characterization of white spaces databases errors
Ahmed Saeed, Khaled A. Harras, Moustafa A. Youssef · 2014
Spectrum regulators consider geo-location databases as the most reliable source of spectrum information for White Space Devices (WSDs). Geo-location databases protect TV band incumbents by keeping track of TV transmitters, and their protected service areas based on their location, transmission parameters, and sophisticated propagation models. However, propagation models inaccuracies can cause an overestimation of the protected area of TV transmitters leading to the inefficient usage of white spaces. In this paper, we present a large scale study, spanning an area of around 3000 km^2 over a driving path of around 190 km, showing that one of the most accurate propagation models, the Irregular Terrain Model (IMT), overestimates the signal power by up to 97% of the time. Based on this study, we provide a characterization of spectrum sensory readings that can be used to amend the prediction of propagation models. This characterization allows spectrum sensors to detect the absence of white spaces with a fairly high threshold of -84 dbm, which enables low cost and accurate spectrum sensing. Furthermore, we present the initial design of SPOC, a system that combines spectrum sensing and propagation modeling in order to better detect white spaces.