Author(s)
Christopher Wahl and Bertrand Hochwald
Abstract
The wireless spectrum is a scarce and valuable resource underpinning many modern technologies we rely on regularly. Despite its importance, there are few ways to effectively monitor it. One approach to monitor the spectrum is radio map estimation (RME). In RME, spectrum sensors are placed around a region of interest to take measurements of the radio environment. The sensors, locations, and measurements are then aggregated to an RME system to estimate the radio environment across the entire region of interest. Naturally, the locations of the spectrum sensors have a significant impact on the accuracy of the radio map estimates. In this work, we propose a sensor placement optimization scheme that finds high-performing sets of sensor locations based on a minimum error objective function. We then validate our approach by comparing our optimized sets of sensor locations with other sets of sensor locations with a practical neural network-based radio occupancy map estimation system in three different regions.