Scholarlink

<p>This article introduces a compact four-port MIMO antenna specifically developed for use in 5G midband and ultra-wideband communication systems. The proposed antenna features a compact structure with four symmetrical monopole radiating elements arranged in a closely spaced configuration (less than λmax/2), enabling efficient space utilization without using any decoupling structures in between the radiating elements. Machine learning (ML) algorithms namely random forest, decision tree and K-nearest neighbour (KNN) were employed to predict S-parameters from antenna design features, with KNN achieving the best accuracy and lowest mean squared error. Unlike traditional HFSS-based parametric sweeps, this work integrates ML regression with grid search to optimize element spacing, minimizing total S-parameter error across the frequency range. This approach significantly reduces simulation effort while ensuring optimal performance for compact, high-isolation MIMO antennas. The antenna is fabricated using FR4 substrate (ε<sub><i>r</i></sub> = 4.4) and measures only 60 × 60 × 1.6 mm<sup>3</sup>. An operational bandwidth of 9.9&#xa0;GHz (2.1–12&#xa0;GHz) and strong isolation of 15&#xa0;dB are offered. The MIMO characteristics are validated using diversity parameters such as envelope correlation coefficient (ECC), diversity gain&#xa0;(DG), mean effective gain (MEG) and total active reflection coefficient (TARC). The designed antenna demonstrates excellent MIMO performance exhibiting an average ECC of 0.0304, DG close to 9.99 and MEG<sub>ij</sub> difference below 0.94&#xa0;dB. It offers TARC under − 10&#xa0;dB and exhibits a signal group delay below 0.25&#xa0;ns confirming strong diversity and MIMO performance across the intended wide frequency band.</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

ML-Driven Spatial Optimization of a Quad-Port MIMO Antenna with Superior Diversity for 5G Midband and UWB Systems

  • 期刊论文

  • 发表日期：

    2026年5月9日

查看全文
  • G. Srivatsun,
  • R. Karthikeyan,
  • S. Arun Kumar

摘要

This article introduces a compact four-port MIMO antenna specifically developed for use in 5G midband and ultra-wideband communication systems. The proposed antenna features a compact structure with four symmetrical monopole radiating elements arranged in a closely spaced configuration (less than λmax/2), enabling efficient space utilization without using any decoupling structures in between the radiating elements. Machine learning (ML) algorithms namely random forest, decision tree and K-nearest neighbour (KNN) were employed to predict S-parameters from antenna design features, with KNN achieving the best accuracy and lowest mean squared error. Unlike traditional HFSS-based parametric sweeps, this work integrates ML regression with grid search to optimize element spacing, minimizing total S-parameter error across the frequency range. This approach significantly reduces simulation effort while ensuring optimal performance for compact, high-isolation MIMO antennas. The antenna is fabricated using FR4 substrate (εr = 4.4) and measures only 60 × 60 × 1.6 mm3. An operational bandwidth of 9.9 GHz (2.1–12 GHz) and strong isolation of 15 dB are offered. The MIMO characteristics are validated using diversity parameters such as envelope correlation coefficient (ECC), diversity gain (DG), mean effective gain (MEG) and total active reflection coefficient (TARC). The designed antenna demonstrates excellent MIMO performance exhibiting an average ECC of 0.0304, DG close to 9.99 and MEGij difference below 0.94 dB. It offers TARC under − 10 dB and exhibits a signal group delay below 0.25 ns confirming strong diversity and MIMO performance across the intended wide frequency band.