Globalized Knowledge-Based, Simulation-Driven Antenna Miniaturization Using Domain-Confined Surrogates and Dimensionality Reduction

The design of contemporary antenna systems encounters multifold challenges, one which is a limited size. Compact antennas are indispensable for new fields application such as the Internet Things or 5G/6G mobile communication. Still, miniaturization generally undermines electrical and field performance. When attempted using numerical optimization, it turns into constrained problem with costly constraints requiring electromagnetic (EM) simulations. At same time, due to parameter redundancy compact antennas, size reduction poses multimodal task. In particular, achievable rate heavily depends on starting point, while identifying suitable point challenge its own. These issues indicate that should be addressed global optimization methods. Unfortunately, most popular nature-inspired algorithms cannot applied solving tasks because their inferior computational efficacy difficulties in handling constraints. This work proposes novel methodology globalized structures. Our multi-stage knowledge-based procedure, initialized detection approximate location feasible region boundary, followed by construction dimensionality-reduced metamodel thereof; last stage miniaturization-oriented local refinement geometry parameters. For cost reduction, first stages procedure realized use low-fidelity EM model. approach verified four broadband microstrip benchmarked against multi-start search well Superior rates demonstrated all considered cases maintaining reasonably low costs.