COMPARITIVE PERFORMANCE ANALYSIS OF DUAL-BAND DUAL-POLARIZED RECONFIGURABLE ANTENNA WITH FR4 AND RT DUROID 6002 SUBSTRATES

Abstract

Modern This thesis presents the design, simulation and performance analysis of a Circular Reconfigurable Dual-Band Dual-Polarized Microstrip Patch Antenna integrated with a PIN diode switching mechanism. The antenna was simulated using CST Studio Suite 2024 Learning Edition. The proposed antenna structure is a circular copper patch of radius 12.7 mm on a 40 $\times$ 40 mm FR4 lossy substrate ($\varepsilon_r$ = 4.4, $\tan \delta$ = 0.02) with copper ground plane. The dual-polarization capability is achieved by introducing a U-slot on the radiating patch and using two orthogonal feed ports. A PIN diode model (MADP-00935) is utilised for frequency reconfiguration. The antenna has been analysed with two substrates namely FR4 (Case 1) and RT Duroid 6002 (Case 2). For the FR4 case, the antenna exhibits dual-band resonances at 3.1 GHz and GHz with S11 values of −8.8 dB and −12.5 dB, respectively with a directivity of 6.432 dBi. The RT Duroid 6002 substrate showed better performance with a directivity of 6.909 dBi and a better radiation efficiency of 0.5 dB. This demonstrates the effect of low loss substrate materials on antenna performance. The antenna is designed for WiFi/WiMAX (3.3 GHz) and WiFi/WLAN (5.8GHz) frequency bands. Moreover, a machine learning (ML) integration study was conducted using Python in Google Colab on a synthetic dataset of 200 samples. Five different regression algorithms were used to predict the resonant frequency based on the antenna parameters, Decision Tree, Random Forest, K-Nearest Neighbours, XGBoost and Gradient Boosting. The Gradient Boosting algorithm performed the best with R2 of 0.9934, MSE of 0.4022, and cross-validation R2 of 0.9320, indicating its suitability for antenna performance prediction and design optimisation, thus considerably reducing the iterative full wave electromagnetic simulations.