Scholarlink

<p>The structural modifications of a two-port multiple input multiple output (MIMO) antenna design have been structured and accomplished using characteristic mode analysis (CMA) for the improved isolation between two radiating elements. The proposed two-port MIMO antenna consists of a pair of rectangular-shaped resonators between the two patches and a rectangular-shaped defected ground structure (DGS) as the isolation enhancement tools. Different design steps are constructed through the modal analysis to identify the root of higher mutual coupling, and accordingly the location and shapes of the resonators and DGSs are finalized for better isolation. This CMA-assisted optimized design provides isolation more than 20.49dB (simulated) at 5.09–5.29&#xa0;GHz band. For further improvement of isolation, different design parameters of this design are optimized through surrogate-based Machine Learning (ML) algorithms which enhance the isolation above 37.38dB. The required datasets are generated from CMA-assisted design using parametric analysis in CST microwave studio. These datasets are used for training different ML algorithms to optimize the S-parameters of the final structure. Out of different ML algorithms used for the design here, Gradient Boosting Regression (GBR) and Random Forest Regression (RFR) exhibit better results. This final ML-assisted optimized structure offers 74.81% (simulated 82.43%, measured 74.81%) improved isolation in comparison to the CMA-assisted design. Further, an equivalent circuit model (ECM) is also developed that shows a good agreement. The measured S-parameters, envelop correlation coefficient (ECC), diversity gain (DG), Total Active Reflection Coefficient (TARC), Mean Effective Gain (MEG), Channel Capacity Loss (CCL) and radiation patterns also shows a good agreement with the simulation results.</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

Machine Learning-Assisted MIMO Design with Improved Isolation for Wireless Application

  • 期刊论文

  • 发表日期：

    2025年7月1日

查看全文
  • Aditi Sharma,
  • Sumit Kumar Gupta,
  • Agnivesh Pandey,
  • Robert Mark,
  • Kaushik Mandal,
  • Soma Das

摘要

The structural modifications of a two-port multiple input multiple output (MIMO) antenna design have been structured and accomplished using characteristic mode analysis (CMA) for the improved isolation between two radiating elements. The proposed two-port MIMO antenna consists of a pair of rectangular-shaped resonators between the two patches and a rectangular-shaped defected ground structure (DGS) as the isolation enhancement tools. Different design steps are constructed through the modal analysis to identify the root of higher mutual coupling, and accordingly the location and shapes of the resonators and DGSs are finalized for better isolation. This CMA-assisted optimized design provides isolation more than 20.49dB (simulated) at 5.09–5.29 GHz band. For further improvement of isolation, different design parameters of this design are optimized through surrogate-based Machine Learning (ML) algorithms which enhance the isolation above 37.38dB. The required datasets are generated from CMA-assisted design using parametric analysis in CST microwave studio. These datasets are used for training different ML algorithms to optimize the S-parameters of the final structure. Out of different ML algorithms used for the design here, Gradient Boosting Regression (GBR) and Random Forest Regression (RFR) exhibit better results. This final ML-assisted optimized structure offers 74.81% (simulated 82.43%, measured 74.81%) improved isolation in comparison to the CMA-assisted design. Further, an equivalent circuit model (ECM) is also developed that shows a good agreement. The measured S-parameters, envelop correlation coefficient (ECC), diversity gain (DG), Total Active Reflection Coefficient (TARC), Mean Effective Gain (MEG), Channel Capacity Loss (CCL) and radiation patterns also shows a good agreement with the simulation results.