A Novel Dual-Band Printed SIW Antenna Design Based on Fishnet & CCRR DGS Using Machine Learning for Ku-Band Applications

Abstract

—This paper analyzes and solves the complexity to determine the optimum positions of the Fishnet & Complementary Circular Ring Resonator (CCRR) based Defected Ground Structures (DGS) for Substrate Integrated Waveguide (SIW) based antennas. A new state-of-art technique based on Artificial Neural Network (ANN)-Machine Learning (ML) is proposed for overcoming the lack of solid and standard formulations for the computation of this parameter related to a targeted frequency. As a proof of concept and to test the performance of our approach, the algorithm is applied for the determination of the CCRR and Fishnet-DGS’s optimal positions for a SIW based antenna. The SIW technique provides the advantages of low cost, small size, and convenient integration with planar circuits. The ANN-ML based technique is optimized to attain dual-band resonances with optimal gain and radiation efficiency. The simulation results of the first Fishnet-DGS based antenna show good minimum return losses at two center frequencies, namely, 16.6 GHz (with gain of 6 dB and radiation efficiency of 95%) and 17.7 GHz (with gain and radiation efficiency of 9 dB and 96%, respectively). The second CCRR-DGS based antenna shows about 8 dB gain and a radiation efficiency of 87% at 17.3 GHz, and gain and efficiency of about 8.5 dB and 85% are observed at 17.8 GHz. The proposed CCRR and Fishnet-DGS based antenna are low profiles, low costs, with good gains and radiation efficiencies, making both designs very suitable for Ku-band applications. There is a fair agreement between the measured and simulated results. The achieved dual-band resonances act as a proof of concept that the proposed ANN-ML techniques can be employed for the determination of the optimal positions for CCRR and Fishnet thereby attaining any target dual-bands in the Ku-band with good accuracy of about 98% and a save of 99% in the overall the computational time. © 2021, Electromagnetics Academy. All rights reserved.