Research on UAV Detection and Classification Based on Time-Frequency Analysis and Support Vector Machine

In recent years, unmanned aerial vehicles (UAVs) have been widely used in various fields, bringing many conveniences to our lives but also posing a threat. Therefore, the control of UAVs is imminent. In this paper, we propose a method based on short-time Fourier transform (STFT) and support vector machine (SVM) to identify UAVs in complex environments. First, the remote control signal of the UAV is collected by the self-made signal acquisition equipment, then the two-stage haar wavelet transform is used to de-noise and compress the data, and the signal is processed by STFT transform. The interference of the fixed frequency signal is eliminated by the time-frequency spectrum cancellation method. Finally, the instantaneous frequency, signal duration point, and signal bandwidth of the frequency hopping signal are extracted to form a feature set and classified by SVM. The method is used to test three UAVs. The algorithm has an average recognition rate of 98.43% for three UAVs under the condition of SNR of 5dB.