Radar Target Recognition Based on Micro-Doppler Signatures Using Recurrent Neural Network

Abstract

The micro-Doppler effect focuses on describing the detailed characteristics of moving targets and also plays a key role in the field of radar target recognition. In this paper, recurrent neural network (RNN) is used to classify the micro-Doppler signatures of different targets. RNN models are sensitive to temporal signals and thus can learn the necessary temporal dependence of the micro-Doppler signatures. This paper first constructs two-dimensional time-frequency distribution matrices by using short-time Fourier transformation (STFT). Then four types of RNN model are used in radar target classification, including standard RNN, long short-term memory (LSTM), attention-based RNN and attention-based LSTM. Experimental results based on L-band radar measured data show that those RNN models can capture the underlying features of micro-Doppler signatures and have good performance in the target classification experiments.