Wide band THz MIMO antenna with isolation prediction using machine learning algorithms for IoT applications

In this paper, a compact wideband two-port multiple-input-multiple-output (MIMO) antenna is presented which operates in the Terahertz (THz) range specifically at 1.58 THz and 2.6 THz. The proposed design is based on a polyimide substrate (ε r = 3.5, tan δ = 0.003) with a graphene patch radiator and copper ground plane to be used in Internet of Things (IoT) applications. The single-element antenna is developed in four progressive design steps, rectangular patch, corner-chamfered octagonal patch, ring-frame with central aperture and finally, cross-slot (plus-sign) loaded octagonal structure, which has two resonances analyzed at 1.6 THz (S11 = − 32 dB) and 2.35 THz (S11 = − 53 dB). Systematic optimization of the MIMO configuration is done by three placement configurations, and the optimized configuration exhibits an inter-port isolation of up to − 18 dB in the range of 2.6 THz and − 25 dB to − 45 dB in the range of 1–1.5 THz, gain of 6–6.5 dBi over the operating bands. A comprehensive analysis of MIMO diversity has shown in terms of envelope correlation coefficient (ECC) of less than 0.005, a diversity gain (DG) of 9.9–10 dB, a channel capacity loss (CCL) of less than 0.5 bits/s/Hz and a total active reflection coefficient (TARC) of less than − 10 dB. In order to overcome the computational cost of large-scale parametric sweeps, the supervised machine learning (ML) surrogate model is built to predict inter-port isolation (S21) as a function of frequency and inter-element spacing, which helps to minimize the computational load of repeated parametric full-wave HFSS analysis. The hyperparameter optimization of gradient boosting regressor obtains R 2 of 99.99%, MAE of 0.26 dB, and RMSE of 0.37 dB for the held-out test data, outperforming all baseline regression models, making the Gradient Boosting Regressor with Optuna hyperparameter optimization a reliable surrogate model for the isolation prediction of THz MIMO antenna analysis.

Authors

Institutions

Publication Details

Journal
Scientific Reports
Published
2026-08-26
DOI
https://doi.org/10.1038/s41598-026-67362-8
Primary Topic
Antenna Design and Analysis
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Wide band THz MIMO antenna with isolation prediction using machine learning algorithms for IoT applications

Mouaaz Nahas, Gifty Gupta, Jun-Jiat Tiang, Isha Kansal et al.
Scientific Reports
Antenna Design and Analysis
article

Wide band THz MIMO antenna with isolation prediction using machine learning algorithms for IoT applications

Mouaaz Nahas, Gifty Gupta, Jun-Jiat Tiang, Isha Kansal, Ashish Kumar, Md. Ashraful Haque
article en

Abstract

In this paper, a compact wideband two-port multiple-input-multiple-output (MIMO) antenna is presented which operates in the Terahertz (THz) range specifically at 1.58 THz and 2.6 THz. The proposed design is based on a polyimide substrate (ε r = 3.5, tan δ = 0.003) with a graphene patch radiator and copper ground plane to be used in Internet of Things (IoT) applications. The single-element antenna is developed in four progressive design steps, rectangular patch, corner-chamfered octagonal patch, ring-frame with central aperture and finally, cross-slot (plus-sign) loaded octagonal structure, which has two resonances analyzed at 1.6 THz (S11 = − 32 dB) and 2.35 THz (S11 = − 53 dB). Systematic optimization of the MIMO configuration is done by three placement configurations, and the optimized configuration exhibits an inter-port isolation of up to − 18 dB in the range of 2.6 THz and − 25 dB to − 45 dB in the range of 1–1.5 THz, gain of 6–6.5 dBi over the operating bands. A comprehensive analysis of MIMO diversity has shown in terms of envelope correlation coefficient (ECC) of less than 0.005, a diversity gain (DG) of 9.9–10 dB, a channel capacity loss (CCL) of less than 0.5 bits/s/Hz and a total active reflection coefficient (TARC) of less than − 10 dB. In order to overcome the computational cost of large-scale parametric sweeps, the supervised machine learning (ML) surrogate model is built to predict inter-port isolation (S21) as a function of frequency and inter-element spacing, which helps to minimize the computational load of repeated parametric full-wave HFSS analysis. The hyperparameter optimization of gradient boosting regressor obtains R 2 of 99.99%, MAE of 0.26 dB, and RMSE of 0.37 dB for the held-out test data, outperforming all baseline regression models, making the Gradient Boosting Regressor with Optuna hyperparameter optimization a reliable surrogate model for the isolation prediction of THz MIMO antenna analysis.

Scientific Reports
Multimedia University (MY), Umm al-Qura University (SA), University of Liberal Arts Bangladesh (BD), Chitkara University (IN)
Umm Al-Qura University
Affordable and clean energy
Openalex Percentile: Top 6%
Antenna Design and Analysis
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.