AUTOMATIC RECOGNITION OF RADAR SIGNAL TYPES BASED ON CNN-LSTM

摘要

In the field of cognitive electronic warfare, automatic feature learning and recognition of radar signal is an important technology to ensure intelligence reconnaissance. This paper analyses a novel structure of CNN-LSTM and proposes an automatic recognition algorithm for radar signals. The main contributions are as follows: Firstly, the radar signal is transformed into a time-frequency image, and the principal component information of the image is extracted by the proposed image processing method (clipping-marginal frequency interception-binarization-remodeling). Then, the designed network CNN-LSTM is employed to realize self-learning and image category annotation (automatic recognition of signal types). In this network, CNN can extract spatial characteristics, LSTM can extract temporal characteristics, CNN-LSTM can utilize temporal and spatial characteristics at the same time. The simulation results show that the proposed algorithms can effectively identify eight kinds of radar signals in low signal-to-noise ratio (SNR).