A neural network approach to Doppler-based target classification

This paper describes a target classification system which uses the measured Doppler signatures to excite a neural network. The radar target is assigned to one of four possible classes: airplanes and ground vehicles, helicopters, groups of small moving targets (e.g. persons), and clutter. A larger number of classes would result in poorer performance of the system, provided that the only available sources of information are the phase and amplitude of the radar signal.