RNN-Aided Zoom-In Matched Filter Algorithm for Motion Parameter Estimation in Forward Scatter Radar for SSA

The growing number of space objects has highlighted the need for new de-tection methods, with forward scattering (FS) showing great potential. However, the shadow-inverse synthetic aperture radar (SISAR) algorithm used in FS scenarios is computationally demanding. Recent advances in deep learning offer a solution: in this study, a recurrent neural network (RNN) was trained to classify FS signals, reducing the motion parameter search space. A zoom-in matched-filter (MF) approach was then used to estimate the motion parameters of the target.