Optimized Deep Graph Shallow Attention Neural Network Based Four-port Multiple-input-multiple-output Antenna Design for Sub-6 GHz 5G Applications

This paper proposes a novel, four-port multiple-input-multiple-output antenna system that is considered for sub-6 GHz 5G applications. A compact multi-band circular printed monopole antenna (MCPMA) is designed for determining the appropriate dimensions of size 110.16×360 mm. The device operates from 0.6 to 1 GHz. To improve cell isolation, four slits are positioned at an angle on the common ground. Deep graph shallow attention neural network with adaptive gold rush optimization algorithm (DGSANN-AGROA) is employed to create a model establishing the relationship among transmission coefficients and antenna geometric parameters. Following this, an adaptive gold rush optimization algorithm (AGROA) is utilized to enhance the antenna array’s decoupling. It showcases pattern diversity, a valuable characteristic for multiple-input-multiple-output implementation. Simulations were conducted using HFSS19 software versions, followed by an evaluation of the introduced antenna in MATLAB. The multiple-input-multiple-output antenna demonstrates favorable diversity characteristics with acceptable diversity gain (>9.5 dB) and envelope correlation coefficient (ECC) (≤0.009).