Radar Determination of Elevation of Two Nearby Targets Using Neural Network

Monopulse methods have been successfully used in radar to determine the direction of arrival of a reflected signal, giving good results when a single isolated target is involved. However, when two nearby targets need to be detected and their angular position determined, this method fails. We show that it is possible to train and use a neural network to recognize two nearby targets. This neural network is then used to determine the angle of arrival of the reflected signal with sufficient accuracy, even when the distance of the targets is less than the beamwidth. We show the results on an eight-beam antenna model by measuring the elevation using an amplitude monopulse and neural network method. While using monopulse, we can reliably discriminate and determine the angular position of two targets only when their distance is larger than one beamwidth. The neural network method works reliably even for targets at distances much smaller.