AI-Assisted Metasurface Antennas Design/Optimization and Performance Enhancement Techniques: A Comprehensive Survey

Abstract

Metasurface optimization employing evolutionary algorithms (EAs) for forward design and machine learning (ML) for inverse design has emerged as a rapidly developing field that uses artificial intelligence (AI) to design and optimize metasurface antenna systems. Metasurface antenna design encompasses the manipulation of electromagnetic waves by controlling their amplitude, phase, and polarization. The optimization process focuses on enhancing the metasurface antenna characteristics to achieve desirable performance metrics, including gain, bandwidth, radiation pattern, and directivity. The advancement of metasurface antennas aims to improve their performance, functionality, and adaptability to satisfy contemporary communication and sensing requirements. This study aims to provide novel recommendations for designing and optimizing metasurface antennas using AI techniques such as intelligent optimization algorithms, ML, and deep learning (DL). Therefore, this paper presents a comprehensive survey of recent literature on metasurface antenna technologies from forward-looking perspectives. Specifically, this study examines recent progress in evolutionary algorithms and AI-enabled approaches, develops and validates a multi-objective fuzzy logic genetic algorithm (MOFLGA) for complementary split-ring resonators (CSRR), and establishes a machine-learning-based classification framework for frequency-selective surfaces through comparative analysis of multiple ML models. Currently, Numerous metasurface designs have been developed by researchers worldwide. However, a comprehensive study integrating different types of electromagnetic metasurfaces, metasurface antennas, and their applications with AI and performance enhancement techniques remains lacking.