Deep Learning-based Interference Detection and Classification for LPI/LPD Radar Systems

In this paper, we propose an interference detection and classification (IDC) algorithm for low probability of intercept/detection (LPI/LPD) radar communications. Specifically, considering LPI/LPD radar systems with frequency-modulated continuous-wave (FMCW) waveforms, an IDC algorithm based on anomaly detection (AD) and multi-class signal classification is proposed in this paper. First, a mathematical model of the received signal is constructed, when the radar signal reflected by a target interferes a Gaussian jammer and illegitimate in-band FMCW waveform. Next, the received signal is represented in time and frequency domains using time-frequency distribution (TFD). Then, the IDC is trained using the TFD and In-phase and Quadrature (I/Q) representations of the received signal as features. Using the proposed IDC algorithm, we can classify four distinct interference signal types with an accuracy greater than 97%. In terms of true positive ratio (TPR) and false positive ratio (FPR), simulation results demonstrate that the proposed algorithm outperforms other existing algorithms.