An Analytical and Mode Mapping Approach Using Machine Learning for Optimizing Hybrid Material-Based THz Antenna With Surface Plasmon Polaritons

Abstract

This article proposes two complementary Yagi-Uda antenna configurations based on hybrid metal-graphene materials with the exploration of surface plasmon polariton (SPP) mechanism for enhanced performance at the THz regime. These antennas are compared with the estimated outcomes in terms of impedance matching, directivity, radiation pattern, and front-to-back ratio. The SPP mechanism plays a crucial role in optimizing the performance by influencing the dispersion properties and mode propagation. The antenna configurations provide −10-dB impedance bandwidths of 10.59% (2.87–3.19 THz) and 22.95% (2.7–3.4 THz). The maximum achieved gain and efficiency are 7.8 dBi and 84.43%, respectively. The performance of the antenna is validated using analytical optimization at the THz regime, incorporating plasma frequency and collision frequency effects. In addition, the equivalent circuit model is analyzed and explored using advanced system design (ADS) to compare the antenna performance with conventional approaches. Moreover, this work provides a significant analysis of modes pertaining to the radiating elements, which efficiently predict the far-field characteristics of the antenna. The existence of higher order modes TM24 and TM44 is analyzed. To address the existence of these modes, a machine learning (ML) technique based on a convolutional neural network (CNN) model is adopted for mode mapping, involving data cleaning, feature extraction, and normalization. Thus, the performance of the antenna is rigorously validated by employing both analytical and ML approaches, with the exploration of SPP mechanism in hybrid material-based nanoantennas at the THz regime.