Performance Analysis of Sequence-based Deep Learning Model for LPI Radar Waveform Recognition in Fading Channel

Abstract

Although many studies have been conducted for the recognition of low probability of intercept (LPI) radar waveforms, only a few consider fading channels because the recognition in the fading channel is more challenging than that in the additive white Gaussian noise channel. In this paper, we investigate the recognition performance of the sequence-based deep learning model for LPI radar waveforms in a fading channel. As inputs of the model, we consider the received radar waveform, its discrete Fourier transform, and its autocorrelation, respectively. Simulation results show that it is advantageous to exploit the discrete Fourier transform of the received radar waveform as the input of the sequence-based recognition model in the fading channel.