Applied Soft Computing 19 (2014) 280–289 Contents lists available at ScienceDirect Applied Soft Computing j ourna l ho me page: www.elsevier.com/locate /asoc Dynamic classification of ballistic missiles using neural networks and hidden Markov models Upendra Kumar Singh a , Vineet Padmanabhan b,∗ , Arun Agarwal b a PGAD, DRDO, Hyderabad 500059, India b School of Computer & Information Sciences, University of Hyderabad, Hyderabad-500046, India a r t i c l e i n f o Article history: Received 21 February 2013 Received in revised form 18 December 2013 Accepted 17 February 2014 Available online 11 March 2014 Keywords: Artificial neural networks Hidden Markov models Trajectory prediction a b s t r a c t This paper addresses dynamic classification of different ranges of ballistic missiles (BM) for air defense application based on kinematic attributes acquired by radars for taking appropriate measures to inter- cept them. The problem of dynamic classification is formulated using real-time neural network (RTNN) and hidden Markov model (HMM). The idea behind these algorithms is to calculate the output in one pass rather than training and computing over large number of iterations. Besides, to meet the conflicting requirements of classifying small as well as long-range trajectories, we are also proposing a formulation for partitioning the trajectory by using moving window concept. This concept allows us to use parame- ters in localized frame which helps in handling wide-range of trajectories to fit into the same network. These algorithms are evaluated using the simulated data generated from 6 degree-of-freedom (6DOF) mathematical model, which models missile trajectories. Experimental results show that both the net- works are classifying above 95% with real-time neural network outperforming HMM in terms of time of computation on same data. The small classification time enables the use of real-time classification neural network in complex scenario of multi-radar, multi-target engagement by interceptor missiles. To the best of our knowledge this is the first time an attempt is made to classify ballistic missiles using RTNN and HMM. © 2014 Elsevier B.V. All rights reserved. 1. Introduction Classification of tactical and strategic missiles with reasonably accurate probability is of paramount importance for success of any air-defense applications. One of the key elements (challenges) in air defense is ascertaining which element in the threat cloud is the lethal object [15]. Air space during war typically consists of a combination of fighters, bombers, helicopters, transport planes, other air-crafts and ballistic missiles and cruise missiles. Since bal- listic missiles (BM) have the capability to deliver weapons of mass destruction it is important to identify/classify these targets and take suitable measures to neutralise them. Fig. 1 shows one sce- nario where enemy missile is launched from point zero at x-axis and aims to destroy location at the range of 1500 km. From the figure, it can be seen that ballistic missiles have three phases of flight, namely boost phase, mid-course phase and termi- nal phase. Boost-phase and early mid-phase are difficult to destroy ∗ Corresponding author. Tel.: +91 9989897205. E-mail addresses: singh ukin@yahoo.co.in (U.K. Singh), vineetcs@uohyd.ernet.in, vcpnair73@gmail.com (V. Padmanabhan), aruncs@uohyd.ernet.in (A. Agarwal). as detection requires sensors in enemy aero-space or satellite intelligence, which is not possible and not done. This leaves late mid-phase and terminal phase as possible phases of interception of the enemy missile. Though the general policy of ballistic missile defense is to neutralise incoming threat at higher height and longer range so that debris falls away from intended impact zone [11], this is often not realised due to late detection and short response time of launch. Another major element in the air-defense system is the Radar which has the multi-function capability of searching, detecting and tracking multiple targets. The radar provides data with higher rates to enable to compute area of impact and determine the time of launch by fire-control system. Electronically scanned phased array radars are highly reliable and provide multi-target tracking of tactical and strategic missiles [16]. The information provided by these radars can be used to classify the targets which are either direct measurement or derived parameters which includes posi- tion, velocity, accelerations, jerk factor, specific energy, altitude, radar cross section (RCS), etc. Currently, the conventional methods using correlation based logic are able to deliver results up to 90% confidence level. The success of these statistical or template matching methods depends http://dx.doi.org/10.1016/j.asoc.2014.02.015 1568-4946/© 2014 Elsevier B.V. All rights reserved.