A Wideband Polarization Diversified MIMO Antenna with Improved Isolation using Machine Learning Optimized Metasurface Absorber

Abstract This communication presents a model and optimization of a dual-sense circularly polarized (RHCP/LHCP) 2-port MIMO antenna with increased isolation. The antenna operates in a broad frequency range of 1.76–4.0 GHz and is loaded with a metasurface. It has dimensions of 98.5 × 39 mm 2 and an edge-to-edge distance of 55.5 mm. The MIMO antenna is created by replicating a single unit and placing it in a mirrored configuration around the y -axis. The dimensions of the unit antenna are 40 × 39 × 0.8 mm 3 and it is printed on a substrate made of FR4 with a thickness of 0.8 mm. The substrate has a relative permittivity ( ε r ) of 4.4 and a loss tangent (tan δ ) of 0.02. The isolation between the ports may be enhanced by including a metasurface absorber, increasing from 12 dB to 23 dB. To verify the wideband and CP properties, we examine the MIMO antenna unit element for its reflection coefficient, peak gain (4.2 dB), radiation efficiency (about 90%), axial ratio, and surface current distribution. Machine learning is used to improve the design of metasurface unit cells to achieve optimum absorption. The Envelope correlation coefficient (ECC), Diversity gain (DG), and Channel capacity loss (CCL) of the proposed metasurface absorber-enabled antenna are simulated and calculated to prove its MIMO identity. The HFSS simulations are consistent with the experimental results obtained from the developed model.