Machine learning-assisted validation of a high-isolation THz MIMO antenna for 6G communication and IoT application

The terahertz (THz) frequency spectrum, which enables ultra-high data rates, low-latency communication (URLLC), and vast connectivity, is poised to be a key component in advancing wireless technology for 6G and beyond. But creating THz antennas that can overcome obstacles such as high propagation loss and complex construction is still quite difficult. A compact 1 × 2 MIMO patch antenna with dimensions of 130.7 μm × 288 μm is shown in this study. It is optimized for dual-frequency bands at 8.20-9.63 THz and 9.95-10.86 THz, making it ideal for secure communication and high-resolution photography. The suggested antenna outperforms current designs, achieving an astounding 14.07 dB of gain, 87.7% efficiency. Moreover, the MIMO antenna's capability is confirmed by an isolation of - 44.64 dB, an ECC of 1.07 × 10^-10, a DG of 9.99 dB, TARC and CCL values less than - 10 dB and 0.4 bps/Hz, respectively. Machine learning (ML) techniques, specifically regression models such as Extra Trees, were used to optimize design parameters to improve performance. The result was remarkable prediction accuracy, with an R² value of 98.13%, a mean absolute error (MAE) of 1.31%, and a root mean square error (RMSE) of 1.76%.