article · Results in Engineering
Researchers have designed a compact multiple-input multiple-output antenna intended for terahertz-band wireless communications and Internet of Things devices. The system features a one-by-two configuration built on a polyimide substrate, achieving dual resonances at 7.52 and 8.2 terahertz. Simulation across a 2.6 terahertz bandwidth demonstrated a peak gain of 12.116 dB, high radiation efficiency of 88.86 percent, and isolation exceeding 36 dB between antenna elements. To support design verification, an equivalent circuit model was created and shown to align closely with electromagnetic simulations. Furthermore, six supervised regression machine learning models were assessed to forecast antenna isolation characteristics. Among these, the gradient boosting regression model achieved the highest precision, predicting isolation with an accuracy above 94 percent based on statistical evaluation metrics.
Next-generation wireless networks, including 6G and advanced Internet of Things devices, demand ultra-fast data transfer and compact component layouts. Operating in the terahertz spectrum allows for massive bandwidth, but maintaining high signal isolation in micro-scale antennas is challenging. Using machine learning to accurately predict performance speeds up antenna development and lowers the computational cost of testing complex high-frequency communications hardware.
This work is oriented towards future 6G telecommunications and terahertz-frequency Internet of Things applications. The primary potential users include hardware engineers and radio frequency designers seeking rapid ways to model high-isolation micro-antennas. The research remains at an early stage, as findings are based entirely on software simulations, equivalent circuit analysis, and computational predictive models without physical fabrication or experimental over-the-air validation.
AI-generated from the published abstract. Always read the original work before citing.
• A 1 × 2 MIMO antenna was designed for IoT applications, demonstrating dual resonances at 7.52 THz and 8.2 THz. • The antenna delivers a wide 2.6 THz bandwidth and an impressive gain of 12.11 dB, ideal for 6 G. • Features a compact size of 100 × 200 µm² with −36.27 dB isolation and low ECC, ensuring high performance for 6 G. • An equivalent RLC circuit was modeled in ADS, demonstrating close alignment with CST simulation results. • Advanced machine learning techniques were employed to optimize and predict isolation characteristics. This article presents unique research on the application of machine learning techniques to enhance the efficiency of antennas for wireless communication and Internet of Things (IoT) applications in the Terahertz (THz) frequency band. This work utilizes Computer Simulation Technology (CST) Microwave Studio modelling techniques considering the compact dimensions of 120 × 200 μm 2 and a polyimide substrate. The design attains a peak gain of 12.116 dB, isolation exceeding 36 dB, and an efficiency of 88.86 %, covering a broad frequency range of 2.6 THz (7.2438–9.84 THz). The outcomes from the CST were verified by designing and simulating a similar RLC circuit in ADS. Both CST and advanced design system (ADS) simulators produced comparable reflection coefficients. The supervised regression machine learning technique accurately predicted the antenna's isolation. The performance of machine learning (ML) models can be assessed using criteria such as variance score, R squared, mean square error (MSE), mean absolute error (MAE), and root mean square error (RMSE). Gradient Boosting Regression demonstrated the smallest error and highest accuracy among the six ML models tested. The isolation prediction accuracy exceeds 94 %, as indicated by the R-squared and variance scores. The proposed antenna utilizing simulations, multiple regression machine learning models, and an equivalent Resistance-Inductance-Capacitance (RLC) circuit model are strong contenders for THz band applications.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1016/j.rineng.2024.103507
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
