Advanced Optimization based on Bayesian Learning to Design Nonlocal Metasurfaces Deflectors for Satellite Communication - Laboratoire Jean-Alexandre Dieudonné
Pré-Publication, Document De Travail Année : 2024

Advanced Optimization based on Bayesian Learning to Design Nonlocal Metasurfaces Deflectors for Satellite Communication

Résumé

An advanced optimization strategy is deployed to design high performance nonlocal metasurface deflectors, which are used as building blocks of a low-profile 2D-beam-steering-antenna in the Ka-band for SatCom. This new methodology combines two important ingredients. The first one is the nonlocal near-field effect generated by the interactions among adjacent sub-wavelength dielectric pillars. In this paper, we exploit this phenomenon, which was first investigated in the field of photonics, in the microwave domain. Moreover, we have advanced our approach by introducing non-uniformity in the lengths of the all-dielectric pillars. This enhancement has increased the degrees of freedom, offering a broader spectrum of design choices. These expanded choices lay the groundwork for the second ingredient: global optimization algorithm based on statistical learning to derive the geometrical characteristics of the constituting dielectric pillars, driven by a design objective. This optimization strategy delivers superior performance compared to the traditional global optimization algorithms and requires fewer solver calls. In this scenario, the deflector is specifically designed for Kaband operation, achieving a 30°deflection angle at 30 GHz. The proposed solution demonstrates exceptionally high performance across the entire band of interest. The comparison between the optimized nonlocal metasurface deflector and its conventional counterpart, i.e. a phase gradient metasurface, shows numerous advantages, mainly in enhancing the diffraction efficiency, reducing the side lobes levels (SLL) and decriminalize the unwanted polarization. The experimental prototype confirms the results predicted by the simulation, highlighting the potential of this advanced optimization strategy for the design of high-performance beam-deflection systems.
Fichier principal
Vignette du fichier
Advanced_science_v1.pdf (4.91 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
licence
Domaine public

Dates et versions

hal-04850081 , version 1 (19-12-2024)

Licence

Domaine public

Identifiants

  • HAL Id : hal-04850081 , version 1

Citer

Ayoub Bellouch, Mahmoud Elsawy, Stéphane Lanteri, Erika Vandelle, Thi Quynh Van Hoang. Advanced Optimization based on Bayesian Learning to Design Nonlocal Metasurfaces Deflectors for Satellite Communication. 2024. ⟨hal-04850081⟩
0 Consultations
0 Téléchargements

Partager

More