Presentation + Paper
7 June 2024 Improved investigation of electromagnetic compatibility between radar sensors and 5G-NR radios
Anas Amaireh, Yan (Rockee) Zhang, Dexiang (John) Xu, David Bate
Author Affiliations +
Abstract
The emergence of 5G networks in frequency bands close to those used by aviation radar altimeters introduces new interference challenges, necessitating innovative solutions for accurate altitude prediction. This paper introduces a novel approach using machine learning (ML) algorithms to predict aircraft altitude from down sweep signals of frequency-modulated continuous wave (FMCW) radar altimeters, focusing on overcoming 5G interference. It details the implementation of various ML models and the use of down sweep data, which provides unique signal characteristics advantageous for altitude estimation. The methodology involves collecting and processing real 5G signals, emulating radar altimeter operation under different interference levels to create a comprehensive dataset, and rigorously evaluating the ML models with statistical metrics to verify their accuracy in altitude prediction amidst 5G signals. The results show that this ML-based framework markedly enhances altitude estimation accuracy, offering a robust method for radar altimeter operation in the 5G era. This research advances flight safety by providing a solution for reliable altitude measurement despite potential 5G interference.
Conference Presentation
(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.
Anas Amaireh, Yan (Rockee) Zhang, Dexiang (John) Xu, and David Bate "Improved investigation of electromagnetic compatibility between radar sensors and 5G-NR radios", Proc. SPIE 13048, Radar Sensor Technology XXVIII, 130480C (7 June 2024); https://doi.org/10.1117/12.3013749
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Data modeling

Radar signal processing

Education and training

Machine learning

Radar

3D modeling

Neural networks

Back to Top