(A) study of tracking, classification, and parameter estimation for the macro/micro-motion target in radar system

Abstract

In the first part, we propose macro/micro-motion target tracking and classification algorithm. We consider an environment in which multiple targets occur instantaneously, such as ballistic missile (BM) stage separation. In this environment, not only the warhead to be tracked, but also confusing objects such as fuel tank and joint are generated. If target tracking is conducted for all targets, it may be very inefficient as tracking down to confusing objects is meaningless. This could result in eliminating the opportunity to perform other important tasks in radar system. To solve this, the target classification should be performed simultaneously with the tracking process. In the case of short-range ballistic missiles where stage separation occurs at low altitudes, air resistance is caused by the atmosphere, which causes micromovement of the moving targets. Since the targets generated after the stage separation have no thrust, it is difficult to control the attitude, so the micro-motion affects the macro-motion. This movement characteristics are very different from the motion in the case where the micro motion occurs like an aircraft, because the aircraft can control the attitude. Therefore, for the targets considered in this study, it is difficult to obtain good tracking performance through existing target tracking techniques. In this case, the six degree of freedom equation (6DOF) that considers macro/micro-motions at the same time is considered for more accurate target tracking. However, in order to use 6DOF equation, we need to know the aerodynamic force and moment information of the target. In this study, it is assumed that information related to physical characteristics such as warhead and fuel tank can be obtained from literatures such as military reports and technical reports. Through this, the aerodynamic forces and moments for each target are calculated by using fluid dynamics analysis program such as FloEFD in advance and used for position, velocity, angular velocity, and attitude calculation. Furthermore, to improve the attitude tracking and target classification performance, radar cross section (RCS) sequences of the targets obtained from long-range radar are used. As the relationship between state variables and RCS measurement is difficult to be represented in closed form, particle filter (PF) is applied. Furthermore, we conduct Rao-Blackwellized PF (RBPF) that tracks the linear state variables among state variables with extended Kalman filter (EKF) and tracks nonlinear state variables with PF in the reduced dimension space that particles cover to improve tracking performance with the same number of particles. The proposed method tracks and classifies the targets simultaneously using different aerodynamic characteristics and RCS information for each target, and improves tracking performance compared to the target tracking method without considering the aerodynamic characteristics. Meanwhile, the size and rotation vector of a target are very important information in determining the characteristics of the rotating target. In order to estimate this, in the second subtopic of this study, we consider the radar environment using the stepped frequency (SF) modulated waveform which is very useful when extracting accurate range information in a radar environment having a narrow bandwidth antenna because a high resolution range profile (HRRP) can be obtained even with a narrow instantaneous bandwidth. However, as HRRPs are generated by synthesizing several pulses during a long pulse repetition period, distortion of range profiles called the range-Doppler coupling phenomenon occurs when the target moves. We use this distortion for estimating the size of board-shaped rotating target. The size of the target can be successfully estimated assuming that the target position and the direction of the rotation vector are known. However, since it is almost impossible to know the rotation vector of the target in advance, a distributed radar system is considered to estimate the rotation vector. To do this, we analyze the characteristics of HRRPs on time-range images (the sequence of range profiles over time) when the board-shaped rotating target is considered. In addition, by analyzing the influence of the geometry between the target and the radar on the HRRP, the information extracted from the HRRPs on time-range images is used as a measurement to estimate the size and rotation vector of the target. Through the two kinds of nonlinear least-squares methods, desired parameters are estimated. We analyze the target size and rotation vector estimation performance through several simulation results and verify the validity of the results by comparing with Cramer-Rao lower bound (CRLB).