AI-Driven Design of a Quasi-Digitally-Coded Wideband Microstrip Patch Antenna Array

Artificial intelligence (AI) is enabling the automated design of contemporary antennas for numerous applications. Specifically, the use of machine learning (ML)-assisted global optimization techniques for the efficient design of modern antennas is now fast becoming a popular method. In this work, we demonstrate for the first time, the ML-assisted global optimization of a high-dimensional non-uniform overlapping quasi-digitally coded microstrip patch antenna array using a new AI-driven antenna design technique, called TR-SADEA (the training cost-reduced surrogate model-assisted hybrid differential evolution for complex antenna optimization). The TR-SADEA-generated array showed very promising simulated frequency responses for potential wideband applications with a -10 dB impedance bandwidth of 5.75 GHz to 10 GHz, a minimum in-band realized gain of 5.82 dBi, and a minimum in-band total radiation efficiency of 87.84%.