Optimal field measurement design for radio environment mapping

White spaces are locally or temporally available frequencies that are principally reserved for primary use such as TV broadcasting. The currently prevalent view to the utilization of such spectrum resources is that white space devices must query a geolocation database to obtain information about available frequencies and related maximum transmission powers. A geolocation database is fundamentally based on field strength estimates of the primary service obtained using radio propagation models. However, field strength estimates for geographically extensive broadcast networks are typically computed for spatial resolutions of hundreds of square meters. Therefore, even with the most sophisticated propagation models currently available, the predicted values always contain errors due to the limited geographical information. To overcome such deficiency, we propose a geostatistical approach for estimating the radio environment based on universal kriging interpolation. In addition, we optimize the locations of field measurements using spatial simulated annealing, and show that it improves significantly local field strength estimates. Index Terms-radio