Deinterleaving Radar Pulse Trains Using Neural Networks

All known time of arrival (TOA) deinterleaving algorithms have great diiculty separating interleaved radar pulse trains when two or more of the trains are significantly jittered. Such scenarios tend to lead to overwhelming ambiguity and fragmentation diiculties when using traditional techniques. In this report we propose a modiied sequence search technique combined with a novel recursive neural network approach. The sequence search forms the hypothesis that a given chain of pulses is periodic and the neural network is used to intelligently test this hypothesis. This allows the satisfactory dein-terleaving of multiple jittered radar pulse trains.