High-Performance Quatrefoil-Slotted THz MIMO Antenna for 6G Applications with Regression-Based Machine Learning Validation

Abstract The advent of 6G wireless communication demands ultra-high-speed, low-latency, and spectrally efficient solutions, necessitating advanced antenna technologies operating in the terahertz (THz) spectrum. This work presents a compact twoelement Multiple Input Multiple Output (MIMO) microstrip antenna using a graphene radiating patch on a 5 µm polyimide substrate and a copper ground for THz applications. The proposed antenna resonates at 4.78 THz with an exceptionally low return loss of − 52 dB and operates across a wide frequency band of 3.81–5.13 THz, achieving a bandwidth of 1.32 THz. It delivers a peak gain of 11.97 dB, a high radiation efficiency of 90%, and outstanding isolation of − 40 dB between ports. The diversity performance is validated by an envelope correlation coefficient (ECC) of 0.0000175 and a diversity gain (DG) of 9.999, making it highly suitable for next-generation high-data-rate applications. An equivalent RLC circuit is developed and validated against full-wave simulations to model the antenna's impedance characteristics accurately. Furthermore, machine learning (ML) regression algorithms, including Extra Trees Regression, are integrated into the design process, significantly reducing evaluation time and enabling multi-parameter optimization. Comparative studies confirm that the proposed antenna surpasses state-of-the-art designs in bandwidth, gain, isolation, and adaptability, establishing it as a strong candidate for 6G communication, biomedical imaging, and high-resolution sensing. The suggested antenna, characterized by its compact dimensions, superior isolation, and remarkable efficiency, demonstrates significant potential for high-speed 6G applications, offering a robust solution for next-generation wireless communication systems