A Novel Two-Stage Optimization Framework for Designing Active Metasurfaces Based on Multiport Microwave Network Theory

Reconfigurable and programmable metasurfaces have shown promising potential in a variety of fields because of their remarkable ability to control electromagnetic (EM) waves. However, the design of metasurface elements requires extensive high-cost EM simulations to obtain certain functionalities, limiting the metasurface developments and applications. In this work, we propose a novel two-stage optimization framework to design active metasurface elements based on microwave network theory and optimization methods. In the proposed framework, a low-cost multiport network model is first established to achieve the rapid prediction of the EM responses of the meta-elements containing multiple loads. Based on the low-cost multiport network model, a genetic algorithm (GA) is then used to choose suitable port loadings, such as tunable components, short-circuit, and open-circuit. For the tunable components, we adopt a gradient-based optimization method to determine their working states. To verify the performance of the proposed method, a reflection phase-modulated metasurface is designed and fabricated. Simulation and experimental results are in good agreement with design goals, demonstrating the effectiveness and efficiency of the proposed method.