Design, RLC equivalent circuit modeling, and ML/DL-based performance prediction of a graphene MIMO antenna for 6G THz systems

Intense interest in ultra-high-speed, low-latency, and large-capacity wireless communication has accelerated the development of terahertz (THz) antenna systems for sixth-generation (6G) networks. This work presents a compact graphene-based 1 × 2 multiple-input, multiple-output (MIMO) slotted microstrip patch antenna on a low-loss polyimide substrate for wideband terahertz applications. To improve impedance matching, mutual coupling, and radiation performance, the proposed configuration comprises designed slots, a defective ground structure (DGS), and a vertical copper decoupling strip. With a super-wideband from 0.62 to 3.46 THz, the antenna resonates at three frequencies: 1.2928 THz, 2.177 THz, and 3.1717 THz, covering 2.84 THz of bandwidth. The simulation results show that it has a peak gain of 12.2 dB, a radiation efficiency of 96.47%, a mutual coupling of less than − 30.12 dB, an envelope correlation coefficient of 0.000167, and a diversity gain of 9.9992 dB. A similar lumped-element RLC circuit model is built in ADS to confirm the electromagnetic behavior at the circuit level using the R, L, and C elements. Therefore, the RLC-based reflection coefficient is in perfect agreement with the CST full-wave simulation results, corroborating the validity of the proposed antenna design. Moreover, ML and DL regression models are employed to predict antenna performance based on geometric design parameters. We evaluate ten classic ML models and two DL models, which are a small multilayer perceptron and a residual multilayer perceptron architecture, in a leakage-safe grouped cross-validation procedure. The Gradient Boosting regressor outperforms all other models in predictive accuracy, with an R² of 0.9932, an MAE of 0.0529, and an RMSE of 0.0859. The result indicates that the proposed graphene-based MIMO antenna, validated by RLC circuit analysis and predicted by comparative ML and DL modeling, is a promising candidate for future 6G THz communication, high-speed sensing, biomedical diagnostics, and intelligent IoT applications.