Machine Learning-Driven Inspired MTM and Parasitic Ring Optimization for Enhanced Isolation and Gain in 26 GHz MIMO...

Area of Science:

• Electrical Engineering
• Antenna Theory
• Metamaterials

Background:

• 5G millimeter-wave (mmWave) communication demands high-performance antenna arrays.
• Designing compact, efficient MIMO antennas for 26 GHz presents significant challenges.
• Metamaterials and AI offer potential solutions for antenna optimization.

Purpose of the Study:

• To develop an intelligent design framework for a 26 GHz MIMO antenna array for 5G.
• To optimize antenna performance using machine learning (ML) models.
• To validate the AI-assisted design methodology through simulations and measurements.

Main Methods:

• A compact microstrip patch antenna was designed for 26 GHz operation.
• A Multi-Layer Perceptron (MLP) model optimized metamaterial unit cell dimensions for gain enhancement.
• A Random Forest (RF) model with Bayesian optimization tuned parasitic ring parameters to minimize mutual coupling.

Main Results:

• The single-element patch achieved S11 < -40 dB and a wide bandwidth (~26.2%) at 26 GHz.
• MLP optimization increased antenna gain by +2 dB.
• RF optimization reduced mutual coupling (S12) from -25 dB to -58 dB at 26 GHz.
• Simulated and measured results closely matched ML predictions, validating the design framework.

Conclusions:

• The AI-assisted design framework provides a rapid and reliable method for developing next-generation mmWave MIMO antenna systems.
• The proposed intelligent design methodology effectively optimizes antenna parameters for 5G applications.
• The validated framework accelerates the realization of advanced wireless communication technologies.