Design and optimization of dielectric resonator based MIMO antenna with beam tilting at mm-wave for enhanced 5G communication using machine learning

Abstract In this paper, a dual-port ceramic-based multiple-input multiple-output (MIMO) antenna is designed for operation at mm-wave frequencies. The design incorporates a partially reflecting surface (PRS) positioned above the dual-port antenna to facilitate beam tilting. Additionally, various machine learning (ML) algorithms are utilized to optimize the unit cell length and the distance between the Dielectric Resonator Antenna (DRA) and the PRS, thereby achieving the desired beam tilting characteristics. The ceramic element is excited through an aperture, which generates the HEM 11 δ mode in the cylindrical dielectric resonator antenna (CDRA). The proposed antenna demonstrates excellent port isolation, exceeding 35 dB, and operates effectively within the 25.9–27.4 GHz frequency range. The PRS enables a broadside beam tilt of ±30 degrees, incorporating pattern diversity into the design. These characteristics make the proposed antenna an ideal candidate for mm-wave vehicle-to-vehicle and 5G communication systems.