Circular MIMO antenna with ML-based bandwidth and isolation prediction for 6G communications

Area of Science:

• Electrical Engineering
• Antenna Theory
• Wireless Communications

Background:

• Next-generation 6G wireless systems require compact, wideband, high-isolation antennas operating at terahertz (THz) frequencies.
• Traditional electromagnetic optimization methods are computationally expensive for complex multi-parameter antenna designs.

Purpose of the Study:

• To introduce a compact circular MIMO antenna for THz communication systems.
• To develop a machine learning (ML)-based framework for efficient performance prediction and design optimization of THz MIMO antennas.

Main Methods:

• Design of a compact circular MIMO antenna with radiating elements, concentric ring, side stubs, isolation walls, and partial ground structure.
• Development of an ML-based predictive model using five supervised regression algorithms trained on key geometrical parameters.
• Evaluation of antenna performance metrics including bandwidth, gain, radiation efficiency, diversity gain (DG), and total active reflection coefficient (TARC).

Main Results:

• The proposed antenna operates at 5.55 THz with a wide bandwidth (4.56-5.86 THz), peak gain of 8.04 dB, and 85.64% radiation efficiency.
• High inter-element isolation was achieved, with DG of 9.985 dB and TARC below -35 dB.
• Cat Boost regression model demonstrated high prediction accuracy (R² of 97.05% for bandwidth, 92.56% for isolation).

Conclusions:

• The integrated ML framework enables efficient prediction and optimization of compact THz MIMO antennas.
• The proposed antenna design offers a promising solution for the demanding requirements of 6G wireless systems.
• ML-based predictive modeling accelerates the development cycle for advanced antenna technologies.