Building a Smart EM Environment -- AI-Enhanced Aperiodic Micro-Scale Design of Passive EM Skins

TL;DR

This paper introduces an AI-enhanced design method for passive smart skins that accounts for electromagnetic interactions at micro and macro scales, improving the reliability and efficiency of the design process.

Contribution

It combines inverse source formulation, System-by-Design, and an AI-based digital twin to optimize aperiodic micro-scale passive EM skins with validated results.

Findings

Reliable prediction of EM interactions in complex skins

Enhanced design efficiency with AI-based digital twin

Validated layouts with full-wave simulation accuracy

Abstract

An innovative process for the design of static passive smart skins (SPSSs) is proposed to take into account, within the synthesis, the electromagnetic (EM) interactions due to their finite (macro-level) size and aperiodic (micro-scale) layouts. Such an approach leverages on the combination of an inverse source (IS) formulation, to define the SPSS surface currents, and of an instance of the System-by-Design paradigm, to synthesize the unit cell (UC) descriptors suitable for supporting these currents. As for this latter step, an enhanced Artificial Intelligence (IA)-based digital twin (DT) is built to efficiently and reliably predict the relationships among the UCs and the non-uniform coupling effects arising when the UCs are irregularly assembled to build the corresponding SPSS. Towards this end and unlike state-of-the-art approaches, an aperiodic finite small-scale model of the SPSS is derived to generate the training database for the DT implementation. A set of representative numerical experiments, dealing with different radiation objectives and smart skin apertures, is reported to assess the reliability of the conceived design process and to illustrate the radiation features of the resulting layouts, validated with accurate full-wave simulations, as well.