Machine Learning-Assisted Design of a Wideband Compact Corporate-Fed Quad-Branch Pentagon Microstrip Patch Antenna for mmWave 2X4-Element MIMO 6G Applications
Year: 2026
Type: conference-paper
Abstract: A wideband compact corporate-fed quad-branch pentagon microstrip patch antenna incorporated into a 2x4-element MIMO configuration for mmWave 6G applications is designed with machine learning assistance. In order to improve radiation characteristics and impedance matching, a data-driven technique is used to develop the suggested single-element antenna, which has a compact size of $\mathbf{1 0} \times \mathbf{1 0} \mathbf{~ m m}^{\mathbf{2}}$. With a bandwidth of 7.5 GHz that is appropriate for high-data-rate communication, broad frequency range on 24.5-32 GHz. The concept is expanded to an 8-element (2x4) MIMO configuration with an overall size of $\mathbf{4 0} \times \mathbf{4 0} \mathbf{~ m m}^{\mathbf{2}}$ in order to improve system performance. The MIMO antenna achieves near-ideal VSWR (~1.1), good impedance matching $(\mathbf{S} 11 \approx \mathbf{- 4 8} \mathbf{~ d B})$, peak gain of 11.5 dB, and directivity of 23. Reduced mutual coupling and increased channel capacity are also confirmed by important MIMO performance metrics, low envelope correlation coefficient (ECC < 0.005), high diversity gain (~10 dB), low total active reflection coefficient (TARC < -20 dB), and minimal channel capacity loss (CCL < 0.4 bits/s/Hz). A corporate feeding network in conjunction with the suggested quad-branch pentagon design guarantees effective beamforming and consistent power distribution. Unlike typical corporate-fed pentagon patch antennas, this study presents a machine learning-assisted multi-objective optimization methodology that simultaneously enhances impedance matching, isolation, and gain in a compact MIMO configuration.
Language: English
Topic: Antenna Design and Analysis
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7203945075