Jacob Wekalao, Amuthakkannan Rajakannu, Tobias Topisia
Introduction The rollout of 6G wireless systems demands compact, frequency-agile antennas capable of operating across multiple millimeter-wave bands to support diverse use cases such as wearable modules, reconfigurable intelligent surfaces, and small-cell infrastructure. Conventional fixed-band antennas struggle to meet these multi-band requirements without increasing device footprint or complexity, motivating the need for reconfigurable architectures that can dynamically switch between frequency states while maintaining high gain and efficiency on a low-loss, flexible substrate suited to sub-terahertz operation. Methods A quad-band frequency-reconfigurable slot antenna was designed on a liquid crystal polymer substrate, incorporating four RF MEMS cantilever switches loaded across a complementary split-ring slot resonator. Five independent switching states were defined to control current-path routing across the resonator, enabling discrete reconfiguration among four millimeter-wave bands spanning 26.5–27.8, 37.2–38.6, 47.4–48.9, and 55.1–57.3 GHz. Full-wave electromagnetic simulations were used to characterize gain, return loss, and radiation efficiency for each switching configuration, and results were validated against experimental measurements. A machine learning surrogate model was additionally trained on physics-informed synthetic data to predict S-parameters and gain, intended to reduce the computational burden of iterative full-wave simulation during early-stage design. Results The proposed antenna achieved realized gains of 8.74–12.86 dBi across all four bands, with return loss consistently exceeding 15 dB and radiation efficiency above 86% in every configuration. Simulated and measured results showed close agreement, with a mean absolute error of 0.41 dB and a maximum gain deviation of 0.44 dBi. The antenna's area-normalized gain efficiency exceeded that of contemporary reconfigurable designs evaluated for comparison. The machine learning surrogate model predicted S-parameters and gain with a mean absolute error below 0.5 dB. Discussion The strong agreement between simulated and measured performance, combined with the antenna's compact footprint and high gain-to-area ratio, supports its viability for 6G applications requiring multi-band agility within tight space constraints, including wearable modules, reconfigurable intelligent surfaces, and small-cell infrastructure. The accuracy of the physics-informed ML surrogate model further suggests a practical pathway for accelerating iterative antenna design by reducing reliance on computationally intensive full-wave simulations, offering a complementary tool for rapid performance estimation in early design stages.