Machine Learning-Assisted Synthesis of Millimeter-Wave Circularly Polarized Arrays with 0°/180° Phase Control for Low Sidelobes

Millimeter wave (mm-Wave) frequency band is susceptible to multipath effects and polarization mismatch. Circular polarization technology can effectively suppress these interferences, and the low lobe characteristic is particularly critical for information leakage prevention in high-security scenarios such as military communication. Therefore, this paper proposes a design for a 16×16 mm-Wave low sidelobe circularly polarized array antenna operating in the 37-39 GHz frequency band. This design employs a unit layout method based on neural network, achieving low sidelobe characteristics through precise control of the distribution of units in the 180° phase state. In terms of feed network design, an offset-coupled microstrip line structure is used to achieve precise control of 0° and 180° phases, thereby avoiding the design challenges posed by traditional complex feed networks.