PGMP: A Device-Free Moving Object Counting and Localization Approach in the Varying Environment

Targets to be localized in the Device-Free Localization (DFL) scheme are generally moving, making the location vectors hold different supports. However, the existing Multiple Measurement Vectors (MMV) algorithms need the estimated vectors to own the same support, which is invalid for the moving objects. In this letter, we propose a novel algorithm called Potential Greedy Matching Pursuit (PGMP), in which the moving object counting and localization issue is investigated under the Compressive Sensing (CS) framework for the first time. Firstly, we exploit the mobility characteristics and design the potential dictionary to store the relationship between the target positions, which can significantly reduce the domain of the possible target positions. Then we build the index dictionary to collect all the possible target traces, among which the real target positions can be selected through the iteration. Finally, we create a novel criterion to recover the real target traces, which has low computation complexity. Simulations show that PGMP can significantly improve target counting and localization performance.