Scholarlink

<p>In this research, some key issues, such as performance enhancement, impedance matching, and miniaturization, are tackled to design and optimize a compact 38&#xa0;GHz millimetre wave (mmWave) microstrip patch antenna (MPA) for fifth generation (5G) wireless systems. A Rogers RT5880 substrate (ε<sub>r</sub> = 2.2, tanδ = 0.0009) was utilized to design and simulate the designed antenna with an octagonal-shaped patch structure in CST Microwave Studio. Several design turns were applied in order to optimize the electromagnetic performance, including rectangular and hexagonal photonic band gap (PBG) on the substrate as well as patch and ground plane modifications. The design step further comprised machine Learning (ML) algorithms, including Gradient Boosting Model Gain (GBMG) optimization and Support Vector Regression (SVR), trained on 150 and 118 sample datasets, respectively. At 38&#xa0;GHz, the record antenna stipulated had a return loss of −34.80 dB, a gain of 7.979 dB, a directivity of 8.263 dBi, and a radiation efficiency of 93.65%. The reliability of ML-based optimization can be verified by the SVR model that achieved R2 (prediction efficiency) = 0.987 and by GBM, which achieved an R2 of 0.9667 with Root Mean Squared Error (RMSE) = 0.2896 for gain prediction. Due to the overall stiff structure, high efficiency, and stable radiation property of the proposed ML-assisted antenna prediction, it is more suitable for 5G mmWave applications, including small-cell communication systems and device-integrated communication systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Compact and high-efficiency mmWave microstrip antenna for 5G with machine learning-based performance prediction

  • 期刊论文

  • 发表日期：

    2025年12月12日

查看全文
  • Rayhanur Rahaman,
  • Bashir Ahammed,
  • Tofayel Karim,
  • Mohammad Abdul Alim

摘要

In this research, some key issues, such as performance enhancement, impedance matching, and miniaturization, are tackled to design and optimize a compact 38 GHz millimetre wave (mmWave) microstrip patch antenna (MPA) for fifth generation (5G) wireless systems. A Rogers RT5880 substrate (εr = 2.2, tanδ = 0.0009) was utilized to design and simulate the designed antenna with an octagonal-shaped patch structure in CST Microwave Studio. Several design turns were applied in order to optimize the electromagnetic performance, including rectangular and hexagonal photonic band gap (PBG) on the substrate as well as patch and ground plane modifications. The design step further comprised machine Learning (ML) algorithms, including Gradient Boosting Model Gain (GBMG) optimization and Support Vector Regression (SVR), trained on 150 and 118 sample datasets, respectively. At 38 GHz, the record antenna stipulated had a return loss of −34.80 dB, a gain of 7.979 dB, a directivity of 8.263 dBi, and a radiation efficiency of 93.65%. The reliability of ML-based optimization can be verified by the SVR model that achieved R2 (prediction efficiency) = 0.987 and by GBM, which achieved an R2 of 0.9667 with Root Mean Squared Error (RMSE) = 0.2896 for gain prediction. Due to the overall stiff structure, high efficiency, and stable radiation property of the proposed ML-assisted antenna prediction, it is more suitable for 5G mmWave applications, including small-cell communication systems and device-integrated communication systems.