Deep learning assisted fractal slotted substrate MIMO antenna with characteristic mode analysis (CMA) for Sub-6 GHz n78 5 G NR applications: design, optimization and experimental validation

Abstract

Abstract This article presents the design and optimization of a Deep Learning assisted MIMO antenna for 5 G New Radio (NR) applications in Sub-6 GHz band. The Prototype of the simulated geometry has been practically modelled and experimentally tested to validate the theoretical results. The optimized two-port MIMO antenna of 48 × 24 mm2 size is designed using fractal slotted substrate geometry. The concept of characteristic mode analysis (CMA) method is performed to characterize the performance of the proposed slotted substrate monopole MIMO Antenna. Usually, Optimization techniques based on traditional electromagnetic simulation tools (CST, HFSS, etc) are time-consuming for the design purpose of complex MIMO antenna geometries. The simulation studies using electromagnetic simulation tools are needed to be performed repeatedly to evaluate the optimum geometrical parameters for achieving better acceptable performance. In order to solve this issue, Deep Learning (DL) method-based design optimization approach is considered as an effective technique for determining the optimum physical parameters. In this article, Deep learning method based Dual-Channel type Deep Neural Network (DC-DNN) optimization technique is suggested to estimate the S-parameters. The generated data is fed to DC-DNN acting as a surrogate model to map the input features with the reflection and transmission coefficients, thereby enabling the search for the optimum design parameters to obtain the best performance. The diversity parameters are observed within acceptable limits and the characteristic parameters related to impedance and radiation performance are best suitable for lower sub-6 GHz n78 (3300–3800 MHz) 5 G NR network applications.