Modelling of the floor effects in device-free radio localization applications

Device-free localization (DFL) systems exploit the human-induced perturbations of the electromagnetic (EM) fields as a privacy-preserving sensing tool for passive detection, recognition, localization, and tracking. Without wearing any electronic device, the monitored subjects (targets) modify the EM field (e.g., the Received Signal Strength - RSS) in a way that depends on their location relative to the wireless devices. Thus, DFL systems exploit specific radio maps to reconstruct the body-induced alterations of the EM field and enable motion tracking. These maps can be learned from training data or obtained from a physical/EM model. Practical EM models are based on the scalar diffraction theory and predict the impact of subject motions on the radio propagation without requiring time-consuming computations. However, they are often limited by free-space propagation assumptions that are unsuitable for complex environments characterized by significant multipath effects. This paper discusses and extends the generic diffraction-based models by considering also the floor influence in indoor scenarios. The proposed model is validated by EM simulations and experiments. The impact of this model on the statistical characterization of the RSS is also analyzed for selected target locations.

I. Introduction

Device-free radio localization (DFL) employs stray ambient signals from Radio Frequency (RF) devices to detect, locate, and track human subjects without requiring them to carry any electronic device [1], [2]. DFL systems re-use the wireless communication signals for sounding the environment to extract 2D/3D views of the traversed objects and bodies [2]. Typical signals adopted are those employed in wireless communications (WiFi or machine-type communications), as they are sensitive to human body presence and motion.