A high-gain THz microstrip patch antenna designed for IoT and 6G communications with predicted efficiency using machine learning approaches

Abstract

The integration of Terahertz (THz) technology into 6 G networks represents a significant advancement in wireless communication, particularly within the Internet of Things (IoT) sector. Terahertz’s frequencies offer wider bandwidths and faster data transmission, crucial for applications such as high-definition video streaming, IoT security systems, and healthcare devices. This work introduces a high-performance THz microstrip patch antenna engineered for IoT and 6 G applications, utilizing Graphene-based patches and polyimide substrates. We demonstrate the antenna's performance through machine learning (ML)–enhanced design optimization, achieving a gain of 14.3 dB, an efficiency of 97.7 %, and over 31 dB of isolation across an extensive bandwidth (1 THz to 5.4 THz). To validate the regression machine learning model for THz MIMO antenna design, a comprehensive dataset was generated using full-wave electromagnetic simulations. This dataset comprises six features based on the geometric and material parameters of the antenna. The implementation of various machine-learning techniques, including Extreme Gradient Boosting (XGB) regression, yielded outstanding outcomes. XGB achieved an R-squared value and variance scores of 98 %, demonstrating exceptional accuracy. It also showed minimal error rates in efficiency prediction, with a reassuringly low Mean Absolute Error (MAE) of 1.62 %, a Mean Squared Error (MSE) of 0.37 %, and a Root Mean Squared Error (RMSE) of 2.78 %. The antenna design is rigorously tested using CST and ADS simulation tools, confirming its superior performance compared to existing systems. The study explores multi-objective optimization, covering efficiency, bandwidth, and compactness, which are crucial for future wireless communication systems. This study highlights the potential of integrating THz technology with machine learning to enhance antenna design, presenting a novel framework for the evolution of future wireless networks with improved performance and energy efficiency.