High performance grid structured MIMO antenna with regression machine learning for high-speed sub THz and THz 6G IoT...

Area of Science:

• Electrical Engineering
• Antenna Design
• Machine Learning Applications

Background:

• Next-generation wireless systems require antennas capable of operating in the Terahertz (THz) spectrum.
• Optimization of antenna gain and isolation is crucial for enhancing performance in wireless communication and Internet of Things (IoT) systems.
• Machine learning offers potential for improving the design and prediction of antenna characteristics.

Purpose of the Study:

• To introduce an innovative machine learning approach for optimizing antenna gain in the THz frequency spectrum.
• To design and analyze a compact polyimide-based antenna for THz communication.
• To validate the antenna model using simulation and circuit modeling, and to forecast MIMO antenna gain using regression algorithms.

Main Methods:

• Antenna design and analysis using CST-2018 simulations and RLC circuit modeling.
• Validation through an analogous RLC equivalent model constructed via ADS.
• Application of supervised regression machine learning algorithms (including XGB Regression) to forecast MIMO antenna gain.

Main Results:

• The proposed antenna achieved high peak gains of 11.91 dB and 12.21 dB across two operational bands.
• Exceptional isolation (31.43 dB, 36.1 dB) and high radiation efficiency (92.42%, 86.93%) were demonstrated.
• XGB Regression achieved over 96% dependability in forecasting MIMO antenna gain, validating the machine learning approach.

Conclusions:

• The designed THz antenna exhibits excellent performance metrics, making it suitable for high-speed 6G applications.
• The integration of machine learning models enhances design efficiency and capacity prediction for MIMO antenna systems.
• This research provides a novel solution for next-generation wireless communication systems operating in the THz spectrum.