Spatial Sensor Layout Optimization for Radio Environment Map Construction

Radio environment map (REM) can visually display the spectrum situation and play a significant role in spectrum monitoring and management. In this paper, an efficient layout op¬timization of spectrum sensors for REM construction is proposed, which involves the minimum number of sampling sensors and an optimal layout. Firstly, the semi-variogram function, rather than the correlation function, is adopted to derive the correlation distance (or the spatial sampling interval), and then the minimum required number of sensors is determined. Secondly, we propose a Local linear approximation-Voronoi (L-Voronoi)-based sampled position scoring criteria as well as the spatial covariance-based uncertainty metric. By evaluating the effectiveness of sampled positions and the uncertainty of unsampled positions, the posi¬tions of sensors are adjusted online. Finally, the proposed layout scheme is verified and evaluated by the simulated REM data. Numerical results demonstrate that its construction error is 4.26%-41.18% better than those of traditional layout methods.