Automated Radar Signal Analysis Based on Deep Learning

Radar emitter analysis is achieved by step by step process which starts with removing noise from a received radar signal, then train the deep learning neural network to generate a real-time prediction; so those important parameters can be abstracted. The first part is denoising a radar signal using a wavelet transform (WT). Consists of a few steps such as power envelope, Hilbert Transform, and wavelet transform to remove the noisy data. The second part includes training a long short-term memory (LSTM) model with the radar dataset, to trace radar signal. Three cases were studied in this paper. In the first case, the radar dataset was transformed using HAAR. The DB3 WT was used in the second case. In the third case, the dataset was used without WT. It is observed that the wavelet transform significantly improves the accuracy of the LSTM model prediction. It is also observed that using HAAR, the predictions are more accurate than that of DB3.