ADAPTIVE MACHINE LEARNING METHODS FOR UAV RECOGNITION IN A VARIABLE RADIO FREQUENCY ENVIRONMENT
DOI:
https://doi.org/10.34132/mspc2025.01.14.19Keywords:
UAV, recognition, radio frequency analysis, domain adaptation, machine learning, dynamic environment, RF fingerprinting.Abstract
The paper addresses the challenge of recognizing unmanned aerial vehicles (UAVs) based on radio frequency signals in variable operating conditions. It highlights that conventional machine learning models, trained in static environments, often suffer performance degradation when exposed to changes in channel characteristics, antenna positioning, noise, and multipath effects. The need for adaptive approaches is emphasized, particularly through domain adaptation techniques, which help preserve recognition accuracy without requiring full retraining. An architecture is proposed that incorporates a feature extraction module, a domain discriminator, and a simulated environment generator. This approach offers promising prospects for the development of robust UAV monitoring systems in both military and civilian applications.
References
Jafar N., Paeiz A., Farzaneh A. Automatic modulation classification using modulation fingerprint extraction. Journal of Systems Engineering and Electronics. 2021. Vol. 32, no. 4. P. 799–810.
Optimized Radio Frequency Footprint Identification Based on UAV Telemetry Radios / Y. Tian et al. Sensors. 2024. Vol. 24, no. 16. P. 5099.
Zhan, Y., et al. (2021). "UAV Detection Based on RF Fingerprints with CNN-LSTM Hybrid Model." Sensors.
Shi, Y., et al. (2020). "Deep Learning for RF Fingerprinting: A ResNet-Based Approach." IEEE Transactions on Vehicular Technology.
Tzeng, E., et al. (2017). "Adversarial Discriminative Domain Adaptation." CVPR.
