Spatial Sampling Points Selection for 3D REM Construciton

Radio environment map (REM) play an important role in managing frequency resources. In this paper, we exploit the sparsity of the signal source in spatial domain to construct a three-dimensional radio environment map(3D-REM) by compressed sensing (CS). Firstly, we proposed a spatial sampling points selection algorithm based on minimizing the column correlation values of sensing matrix. It is proved that the proposed algorithm can improve the recovery accuracy of REM under the same sampling rate. In practical scenarios, there may be some limitations to place sampling receivers in some specific locations, e.g., obstacles. Therefore, we consider a limted set of points which can not be sampled. Simulated annealing (SA) algorithm is used to find candidates when they are not available. Simulation results show that the proposed scheme can achieve better performance of recovery accuracy of REM than other algorithms.