Estimation of Electrical Characteristics of Complex Walls Using Deep Neural Networks

TL;DR

Through-wall radar signatures suffer from wall-induced distortions. The paper tackles this by estimating the wall's dielectric profile $\epsilon_r(x,y)$ and conductivity profile $\sigma(x,y)$ from scattered fields using electromagnetic inverse scattering with three DNN configurations: FC-NN, CNN, and GAN. Networks are trained on FDTD-simulated data and validated on real measurements, achieving wall-thickness estimation and NMSEs in the low 0.04–0.15 range for the profiles, with GAN offering the best robustness under limited data. The results enable wall-effect deconvolution in practical through-wall radar and demonstrate transfer to real-world walls with diverse materials and structures.

Abstract

Electromagnetic wave propagation through complex inhomogeneous walls introduces significant distortions to through-wall radar signatures. Estimation of wall thickness, dielectric, and conductivity profiles may enable wall effects to be deconvolved from target scattering. We propose to use deep neural networks (DNNs) to estimate wall characteristics from broadband scattered electric fields on the same side of the wall as the transmitter. We demonstrate that both single deep artificial and convolutional neural networks and dual networks involving generative adversarial networks are capable of performing the highly nonlinear regression operation of electromagnetic inverse scattering for wall characterization. These networks are trained with simulation data generated from full wave solvers and validated on both simulated and real wall data with approximately 95% accuracy.

Figures (7)

• Figure 1: (a) FDTD simulation setup for single transmitter and multiple element receiver array before (b) wall type-1: single layer homogeneous lossy dielectric wall, (c) wall type-2: single layer dielectric wall with periodic lossy regions, and (d) wall type-3 multiple layered dielectric walls with periodic lossy regions.
• Figure 2: System diagram depicting electromagnetic inversion operation using FC-NN/CNN for obtaining dielectric and conductivity profiles of walls.
• Figure 3: System diagram depicting electromagnetic inversion operation using GAN for obtaining dielectric and conductivity profiles of walls.
• Figure 4: Training and validation loss of (a) FC-NN, (b) CNN, and (c) generator and critic loss of GAN network.
• Figure 5: Estimation of dielectric profiles (a)-(d) and conductivity profiles (e)-(h) of (i) wall type-1 in top row, (ii) wall type-2 in second row, and (iii) wall type-3 in bottom row. Profiles shown in (a),(e) correspond to ground truth; (b) and (f) correspond to FC-NN, (c) and (g) correspond to CNN, and (d) and (h) correspond to GAN.