Related Experiment Video
Updated: May 7, 2025
10:54
Design, Fabrication, and Experimental Characterization of Plasmonic Photoconductive Terahertz Emitters
Published on: July 8, 2013
14.8K
Machine learning-based novel-shaped THz MIMO antenna with a slotted ground plane for future 6G applications
Md Ashraful Haque1, Kamal Hossain Nahin1, Jamal Hossain Nirob1
1Department of Electrical and Electronic Engineering, Daffodil International University, Dhaka, 1207, Bangladesh.
Scientific Reports
|December 31, 2024
Summary
This study enhances 6G terahertz (THz) applications by using machine learning (ML) to predict the performance of multiple-input multiple-output (MIMO) antennas. The Extra Tree Regression model accurately predicted antenna gain, improving application efficiency.
Area of Science:
- Electrical Engineering
- Antenna Design
- Machine Learning Applications
Background:
- Next-generation wireless communication systems, such as 6G, require advanced antenna technologies operating at terahertz (THz) frequencies.
- Multiple-input multiple-output (MIMO) antennas are crucial for enhancing spectral efficiency and data rates in wireless communications.
- Optimizing MIMO antenna performance, especially in the THz band, presents significant design and simulation challenges.
Purpose of the Study:
- To investigate the application of machine learning (ML) techniques for improving the performance prediction of MIMO antennas in the THz frequency band.
- To evaluate the accuracy of a supervised regression ML approach in predicting antenna gain.
- To identify the most effective ML model for optimizing antenna performance for 6G applications.
Main Methods:
- Antenna performance was evaluated using simulation and RLC equivalent circuit models.
- The accuracy of simulation results was confirmed by comparing data from CST and ADS simulators.
- A supervised regression machine learning approach was employed to predict antenna gain, with six models analyzed.
Main Results:
- The designed antenna exhibits a broad bandwidth (2.5 THz from 6.2-8.7 GHz), high gain (14.59 dB), and excellent isolation (> -31 dB) with 96% efficiency.
- Simulation results from CST and ADS showed similar reflection coefficients, validating the RLC circuit model.
- The Extra Tree Regression model achieved the highest accuracy and lowest error among the six ML models analyzed for predicting antenna gain.
Conclusions:
- Machine learning, specifically the Extra Tree Regression model, can accurately predict the gain of THz MIMO antennas.
- The developed antenna design meets key performance indicators for advanced wireless communication systems.
- Integrating ML into antenna design workflows offers a promising approach to optimize performance for future 6G applications.
