Radar Fingerprint Feature Extraction Based on VMD
Fuyan Lin, Songlin Sun, Xiaoying Hou, Yuhao Liu · 2019
Extracting effective radar fingerprint features is crucial for accurate radar fingerprint recognition which is an important part of radar reconnaissance . The received radar signals are nonstationary and nonlinear time series. Traditional feature extraction methods such as bispectral estimation and time-frequency analysis have high complexity and are easy to lose useful information. Based on the advantages of variational mode decomposition in signal decomposition, this paper proposes an optimized radar fingerprint feature extraction method. First, the radar signal is decomposed by the variational mode decomposition method. The signals of different radars are decomposed into different subsignals. The multi-domain features of each modal component are then extracted to characterize the radar fingerprint from multiple angles. Experiments on real radar signals show that the proposed method can effectively extract radar fingerprint features based on partial intrapulse data and realize high accuracy radar fingerprint recognition.