Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Physics-Informed Deep Neural Network Design of Reactively Loaded Metasurfaces

Malik Almunif, John Le, Anthony Grbic

Abstract

A tandem deep neural network approach is presented for the inverse design of reactively loaded metasurfaces with prescribed far-field radiation characteristics. The proposed approach integrates a deep neural network (DNN) with a physics-based microwave network forward solver. The DNN maps target far-field patterns to distributions of reactive loads across the metasurface unit cells. The predicted distribution of reactive loads is evaluated by the forward solver to compute the resulting radiation pattern and guide the learning process through a cosine-similarity loss function. The forward solver enables a fast evaluation of the metasurface's electromagnetic response, significantly reducing the computational cost required for training. The proposed approach is applied to a metasurface with aperture-coupled unit cells loaded with reactances. Several design examples are presented to demonstrate the accurate synthesis of shaped and steered radiation patterns. Full-wave electromagnetic simulations are performed to validate the accuracy of the designed beamforming metasurfaces.

Physics-Informed Deep Neural Network Design of Reactively Loaded Metasurfaces

Abstract

A tandem deep neural network approach is presented for the inverse design of reactively loaded metasurfaces with prescribed far-field radiation characteristics. The proposed approach integrates a deep neural network (DNN) with a physics-based microwave network forward solver. The DNN maps target far-field patterns to distributions of reactive loads across the metasurface unit cells. The predicted distribution of reactive loads is evaluated by the forward solver to compute the resulting radiation pattern and guide the learning process through a cosine-similarity loss function. The forward solver enables a fast evaluation of the metasurface's electromagnetic response, significantly reducing the computational cost required for training. The proposed approach is applied to a metasurface with aperture-coupled unit cells loaded with reactances. Several design examples are presented to demonstrate the accurate synthesis of shaped and steered radiation patterns. Full-wave electromagnetic simulations are performed to validate the accuracy of the designed beamforming metasurfaces.
Paper Structure (8 sections, 8 equations, 3 figures, 1 table)

This paper contains 8 sections, 8 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Review of the Reactively Loaded Metasurface Design
  3. Deep Neural Network
  4. Inverse Network
  5. Forward Solver
  6. Training Data
  7. Design Examples
  8. Conclusion

Figures (3)

  • Figure 1: Aperture-coupled metasurface composed of a two-dimensional grid of unit cells. All cells are terminated with reactive loads, except for one centrally located unit cell that is directly-fed to excite the metasurface. All dimensions are in mm.
  • Figure 2: Proposed tandem deep neural network architecture. The inverse network maps a target far-field pattern in the $u$-$v$ plane to a distribution of reactive loads connected to the ports of the metasurface. The predicted distribution of reactive loads is then passed to a microwave network forward solver to compute the resulting radiated far-field and reflection coefficient.
  • Figure 3: Three example metasurfaces designed using the proposed inverse network. The first example corresponds to a uniformly excited array steered off the principle planes. The second example radiates a flat-top far-field pattern oriented along the diagonal of the $u$-$v$ plane. The third example radiates a multibeam far-field pattern.