Parameter Estimation of Rotary Drones in Far Distance using Long-Time Spectral Processing

We investigate the micro-Doppler effect generated by drone rotors in this work, with the aim of estimating relevant parameters and motion information of rotors from the received signal. The commonly used method, short time Fourier transform (STFT), may fail due to the deficiency of the signal accumulation gain, especially when the drone is far away from the observing radar. To counteract this issue, we focus on the rotor parameter estimation of drones in far distance using Long-Time Spectral Processing (LTSP). Although LTSP is capable of preserving a large processing gain, it cannot present instantaneous spectral information of rotors. Instead, a series of harmonics is produced by LTSP. Different from existing works, we provide a theoretical analysis on the LTSP, which explained the generation principle of harmonics with the rotational frequency as the fundamental frequency. Furthermore, we point out the limitation of cepstrum to distinguish frequencies of a multi-rotor drone and propose a multi-harmonic separation method using peak frequency sub-traction to break through the limitation. Both simulations and experiments have been conducted to validate the effectiveness of the theoretical analysis and proposed methods.