Design and optimization of a deep learning algorithm assisted stub-loaded dual band four-port MIMO antenna for Sub-6 GHz 5G and X band satellite communication applications

Abstract

In this article, C-type stubs loaded hexagonal shaped four-element MIMO antenna is proposed with improved isolation for sub-6 GHz 5G and X-band satellite wireless communication applications. Initially, the design and simulation studies of the MIMO antenna are performed using Computer Simulation Software (CST) software but its design parameter estimation and optimization consume higher time using conventional EM simulation tools. In order to solve this issue, Deep Learning algorithm-based approach has been proposed to determine the optimal physical parameters along with best possible performance characteristics. This suggested DL based approach is resource and time efficient for parameter estimation and optimization during the design process. To reduce design space and generation of an effective dataset, the feature reduction method (FRM) is implied during the design process. The S-parameters are predicted through Dual-Channel (DC-DNN) Deep Neural Network model. The dual-band MIMO antenna operates with −10 dB impedance Bandwidths of 40 % (2.8 GHz – 4.2 GHz), 38.41 % (6.1 GHz – 9.0 GHz) and shows excellent diversity performance. The proposed MIMO antenna is suitable for Sub-6 GHz n77 (3.3–4.2 GHz), n78 (3.3–3.8 GHz) 5G NR bands and also supports X band satellite (7.25–7.75 GHz and 7.9–8.4 GHz) communication applications.