Work

Photonic band Gap-Based different slot patch antenna with Machine learning Verification for Wi-Fi 5/6

Year: 2026

Type: article

Abstract: This paper introduces a compact photonic band gap (PBG)-based different-slot microstrip patch antenna (MPA) designed for Wi-Fi 5/6 applications at 5.5 GHz. The design has three electromagnetic features all in one structure: a pentagonal PBG built into the substrate, a defected ground structure (DGS) in the ground plane, and a circular slot in the radiating patch. The goal is to make the impedance matching, gain, directivity, and radiation efficiency better while keeping the size at 30 × 30 mm 2 . The 5.5 GHz frequency is chosen because it is important for WLAN operation in the 5 GHz Wi-Fi band, where routers, access points, and small Internet of Things (IoT) devices need low-profile antennas with stable broadside radiation. The last antenna has a resonance frequency of 5.5 GHz, a scattering parameter (S 11 ) value of −31.70 dB, a voltage standing wave ratio (VSWR) of 1.03, a gain of 6.99 dB, a directivity of 7.44 dBi, and a radiation efficiency of 90.17 %. A machine learning (ML) stage was then used as a quick way to check the validity of the design instead of replacing the electromagnetic design. The input features were the antenna’s main geometric variables, such as the dimensions of the patch, feed, PBG arm, DGS, and slot radius. The target variables were gain and radiation efficiency. There were 626 CST-generated samples and 20 % test samples in the dataset. The predictive results are viewed as high-fidelity surrogate validation rather than proof of universal generalisation because the data were based on simulations and were small. XGBoost gave the best efficiency prediction (R 2 = 0.880, RMSE = 0.062, MAE = 0.021), and Random Forest Regression gave the best gain prediction (R 2 = 0.766, RMSE = 0.058, and MAE = 0.019). For the k-fold cross-validation of the dataset, five separate subsets (5-fold) were created to confirm that results were reliable and reproducible. The results together show that the proposed antenna works well with Wi-Fi 5/6 and that using machine learning to check designs can cut down on the number of times they need to be redesigned in future optimisation workflows.

Source: Engineering Science and Technology an International Journal

DOI: 10.1016/j.jestch.2026.102456

Authors: Rayhanur Rahaman, Bashir Ahammed, Mohammad Shuhrawardy, Mohammad Abdul Alim

Institution: University of Chittagong

Language: English

Topic: Antenna Design and Analysis

Subfield: Aerospace Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $1,800