Supervised-Learning-Based Development of Multibit RCS-Reduced Coding Metasurfaces

Abstract

Coding metasurfaces have been introduced as efficient tools allowing meticulous control over the electromagnetic (EM) scattering. One of their relevant application areas is radar cross section (RCS) reduction, which principally relies on the diffusion of impinging EM waves. Despite its significance, careful control of the scattering properties poses a serious challenge at the level of practical realization. This article is concerned with (global) design optimization of coding metasurfaces featuring broadband RCS reduction. We adopt a two-stage optimization procedure involving data-driven supervised-learning, sequential-search strategy, and direct EM-based design closure of the entire metasurface oriented toward maximizing the RCS reduction bandwidth. Our framework is then used to develop a two-bit coding metasurface. To handle the combinatorial explosion at the concurrent meta-atom optimization stage, a sequential-search strategy has been developed that enables global search capability at low computational cost. Finally, EM-based optimization is executed to maximize RCS reduction bandwidth at the level of entire metasurface. The properties of the coding metasurface are demonstrated using monostatic and bistatic RCS performance. The 10-dB RCS reduction can be obtained in the frequency range of 14.8–37.2 GHz, in a monostatic configuration. Also, 15-dB RCS reduction can be maintained in the frequency range of 16.7–37 GHz. Simulations are validated using physical measurements of the fabricated prototypes. Finally, the performance of the structure is benchmarked against recently reported designs.