A Potential Orthogonal Matching Pursuit Algorithm for Device Free Moving Target Localization

Existing Compressive Sensing (CS) based Device-Free Localization (DFL) schemes focus on stationary targets. However, targets to be localized are always movable, whose positions are not the simple combination of the stationary ones. We investigate the moving target localization scheme in this letter and propose a novel approach named as Potential Orthogonal Matching Pursuit (POMP), which can recover the target positions in multiple time intervals simultaneously. Firstly, we transfer the moving characteristic to the relationship of the location vectors under the CS framework, which significantly reduces the possible location domain of the target. Secondly, we seek all the possible target traces using the tree searching operation and design a novel measure to select the most valuable columns of the sensing matrixes through iteration. Finally, the location vectors can be obtained by the Least-Square algorithm. Simulations show that the proposed POMP can significantly improve the recovery performance compared with the existing OMP algorithm.