A novel method for training and classification of ballistic and quasi-ballistic missiles in real-time

Abstract

In this paper we outline a novel method for classifying ballistic as well as quasi-ballistic missiles using realtime neural network. Fast classification time plays a stellar role for early and prompt action in air-defense scenario. In-order to get the trajectory information of the missile we initially use simulated radar measurements and for final validation real-world radar track is used. Trajectories are segmented to allow small as well as large trajectories to be trained and classified by the same architecture of the neural network. This is needed because ballistic missiles can follow nominal, lofted or depressed trajectory to reach to its target points even when launched from the same point. © 2013 IEEE.