RCS Based Target Classification Using Deep Learning Methods

In this paper, RCS-based target classification using the deep learning method is proposed. Illumination of targets with narrow-band radar signals results in backscattering of the incident energy from the target. The backscattered signal is a function of the target's geometry and its material. The reflected signal carries useful information and can be utilized to identify and classify the target. The RCS is a measure of this property of the target and has been exploited as an extracted feature in our work. The required labeled data is simulated using the SBR method in HFSS. RNN/LSTM is proposed for training and testing the deep learning model. Various models are trained and the best classification accuracy achieved is 98.1%.