AI-Driven System-by-Design Synthesis of Static-Passive EM Skins Meta-Atoms

In this work, pixel-based Static-Passive Electromagnetic Skins (SP-EMSs) meta-atoms are designed exploiting a novel artificial intelligence (AI)-driven technique. More specifically, the synthesis problem at hand is effectively and efficiently solved within the System-by-Design (SbD) framework, by leveraging (i) a smart definition/coding of the degrees-offreedom (DoFs), (ii) a deep learning (DL) strategy for building an accurate-but-fast surrogate model (SM) of the forward solver, and (iii) an effective evolutionary-based global search engine. An illustrative example is provided to assess the effectiveness, reliability, and computational efficiency of the proposed SbD method.