Scholarlink

<p>In recent years, the flex power of the global positioning system (GPS) Block IIR-M and Block IIF satellites has frequently been activated and deactivated. Ground-based GPS flex power monitoring faces several challenges, including uneven receiver distribution, multipath error arising from complex environments, and significant fluctuations in low-elevation carrier-to-noise ratio data. To overcome these challenges, we propose a novel GPS flex power detection method by introducing a detection metric, CN0<sub>el</sub>. Spaceborne receiver data from 15 low earth orbit (LEO) satellites and historical data from February 2022 to December 2023 are utilized to train an extreme gradient boosting (XGBoost) machine learning model. The detection model achieves an accuracy exceeding 99%, as validated by the measured data from January and February 2024. These findings demonstrate that integrating XGBoost machine learning with LEO satellite data offers a highly effective solution for GPS flex power detection.</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

GPS flex power detection leveraging XGBoost machine learning based on LEO satellites

  • 期刊论文

  • 发表日期：

    2025年5月2日

查看全文
  • Yanjun Du,
  • Yuanxi Yang,
  • Xiaolin Jia,
  • Wanqiang Yao,
  • Qin Li

摘要

In recent years, the flex power of the global positioning system (GPS) Block IIR-M and Block IIF satellites has frequently been activated and deactivated. Ground-based GPS flex power monitoring faces several challenges, including uneven receiver distribution, multipath error arising from complex environments, and significant fluctuations in low-elevation carrier-to-noise ratio data. To overcome these challenges, we propose a novel GPS flex power detection method by introducing a detection metric, CN0el. Spaceborne receiver data from 15 low earth orbit (LEO) satellites and historical data from February 2022 to December 2023 are utilized to train an extreme gradient boosting (XGBoost) machine learning model. The detection model achieves an accuracy exceeding 99%, as validated by the measured data from January and February 2024. These findings demonstrate that integrating XGBoost machine learning with LEO satellite data offers a highly effective solution for GPS flex power detection.