Research on UAV Spoofed Signal Detection Based on Multipath Effect Modeling

This paper proposes a UAV spoofed signal detection method that integrates multipath effect modeling with machine learning. First, the signal generation stage fully considers the multipath propagation characteristics present in real communication environments, while introducing Gaussian white noise to simulate more realistic normal and spoofed signal samples. Subsequently, various feature images are extracted from these signals, including power spectral density (PSD), spectrogram, and histogram, in order to comprehensively reflect different dimensional characteristics of the signals in the frequency domain, time-frequency domain, and amplitude distribution. For feature classification, a Support Vector Machine (SVM) is employed as the classifier to effectively distinguish between normal and spoofed signals. This method not only takes into account both the physical and statistical properties of the signals but also enhances the generalization and detection accuracy of the model, providing strong technical support for UAV communication security and abnormal signal identification.