Reliable Field Strength Prediction through an Adaptive Total-Variation CS Technique

The prediction of the 2-D electric field strength distribution from a limited set of measurements and without any prior information on the source is addressed in this letter. Towards this aim, an innovative sparseness-promoting approach is presented based on the profitable integration of a total-variation compressive sensing (TV-CS) recovery technique with the LOLA Voronoi adaptive sampling strategy. From the one hand, sparsity priors are enforced on the discrete gradient of the unknown field strength allowing to exploit physical knowledge on the addressed problem. On the other hand, the LOLA Voronoi approach enables a reduction of the number of field samples and to properly build the TV-CS observation operator. Representative numerical results are discussed to assess, also comparatively, the potentialities and features of the proposed method.