LSTM-Based Boost-Phase Ballistic Missile Tracking

Abstract

Ballistic missile (BM) tracking during the boost phase is the basis of early missile defense. Traditional BM tracking algorithms either perform dynamic modeling with a certain degree of accuracy on the target movement, or rely on the template information of BM’s trajectory or acceleration. When the dynamic modeling is not accurate enough or the prior template information cannot be obtained, both the tracking accuracy of the two methods will be reduced. In order to solve this problem, we propose a boost-phase BM tracking method based on the long short-term memory (LSTM) network. Specifically, a BM trajectory database is first established to provide sufficient offline data for network training. And then a LSTM-based network suitable for boost-phase BM tracking is developed, which consists of three bidirectional LSTM layers, a fully connected layer and a linear output layer. Simulation results demonstrate that the proposed method has better comprehensive performance in terms of tracking accuracy and computational complexity compared with the existing BM tracking algorithms.