High-performance terahertz MIMO antenna for 6G: Graphene integration and machine learning prediction

The rapid evolution of wireless communication systems and the increasing demand for advanced terahertz (THz) sensing technologies have catalyzed the need for high-performance, compact antennas capable of meeting the stringent requirements of 6G networks. This paper presents a novel triple-band multiple-input multiple-output (MIMO) antenna specifically engineered for THz operation. The antenna exhibits five distinct resonances at 2.8096 THz, 3.328 THz, 3.8752 THz, 4.3488 THz, and 4.8928 THz, corresponding to three effective operating bands with bandwidths of 0.4388 THz, 0.422 THz, and 1.5162 THz. These operating bands provide wide spectral coverage for multi-functional applications such as high-data-rate wireless communication, sensing, and imaging. The proposed MIMO antenna achieves a peak gain of 13.479 dB and a high radiation efficiency of 94.3 %, ensuring minimal signal loss and reliable performance in high-frequency environments. Furthermore, excellent port isolation of −36.867 dB confirms reduced mutual coupling between antenna elements, while the envelope correlation coefficient (ECC) of 0.00012446 and Diversity Gain (DG) of 9.999938 highlight outstanding diversity performance and low signal correlation. To enhance design optimization and performance prediction, machine learning (ML)-based regression models were incorporated and trained using simulated data. The ML models demonstrated strong predictive accuracy, achieving R 2 = 93.34 %, explained variance score (EVS) = 93.91 %, mean square error (MSE) = 1.94 %, root mean square error (RMSE) = 13.93 %, and mean absolute error (MAE) = 4.73 %. These results validate the capability of ML to accelerate antenna design by minimizing iterative simulations. The combination of multi-band operation, high gain, strong isolation, and ML-assisted optimization makes the proposed antenna a promising candidate for next-generation THz systems, including short-range secure communications, low-power sensor networks, non-destructive testing, biomedical imaging, and ultra-fast 6G backhaul links.