Scholarlink

This work introduces a robust Alternative Position, Navigation, and Timing (APNT) system for aircraft navigation under GNSS denial or jamming. The system uses powerful Extended Kalman Filter (EKF) and Long Short-Term Memory (LSTM) neural networks to fuse ground-based Distance Measuring Equipment (DME) stations, onboard IMUs, and magnetometers. The hybrid architecture changes between GNSS and DME/IMU modes, and the LSTM learns from prior navigation data to fix mistakes. Experimental findings show less than 5-m positioning accuracy under jamming, 90% error reduction versus GNSS-only systems, and robustness in urban, hilly, and adversarial situations. MSR, PSNR, SNR, and latency certify the system's real-time capabilities. The suggested approach aligns with FAA AC 20-138D GNSS backup system requirements and provides a scalable framework for military, commercial, and autonomous applications in disputed situations, addressing significant aviation safety gaps. Real-time filtering and adaptive error correction make the hybrid EKF + LSTM architecture better than the solo EKF or LSTM. Conventional DME/DME range limits are solved by multi-DME optimization using binary integer linear programming. Compliance satisfies FAA 2030 backup navigation system criteria. Integrating ARAIM for integrity monitoring, using lightweight LSTM models to decrease computational overhead, and field testing in real-world flying conditions are future objectives.

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

Multi-DMS Navigation System with EKF and Machine Learning for Alternative Position, Navigation, and Timing (APNT) in On-Route Aircraft Aircraft with Blocking GNSS

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Hasan Mohammed Khalil,
  • Ekbal Hussein Ali,
  • Hatam Kareem Kadhom

摘要

This work introduces a robust Alternative Position, Navigation, and Timing (APNT) system for aircraft navigation under GNSS denial or jamming. The system uses powerful Extended Kalman Filter (EKF) and Long Short-Term Memory (LSTM) neural networks to fuse ground-based Distance Measuring Equipment (DME) stations, onboard IMUs, and magnetometers. The hybrid architecture changes between GNSS and DME/IMU modes, and the LSTM learns from prior navigation data to fix mistakes. Experimental findings show less than 5-m positioning accuracy under jamming, 90% error reduction versus GNSS-only systems, and robustness in urban, hilly, and adversarial situations. MSR, PSNR, SNR, and latency certify the system's real-time capabilities. The suggested approach aligns with FAA AC 20-138D GNSS backup system requirements and provides a scalable framework for military, commercial, and autonomous applications in disputed situations, addressing significant aviation safety gaps. Real-time filtering and adaptive error correction make the hybrid EKF + LSTM architecture better than the solo EKF or LSTM. Conventional DME/DME range limits are solved by multi-DME optimization using binary integer linear programming. Compliance satisfies FAA 2030 backup navigation system criteria. Integrating ARAIM for integrity monitoring, using lightweight LSTM models to decrease computational overhead, and field testing in real-world flying conditions are future objectives.