Interacting multiple model filter for tactical ballistic missile tracking

Abstract

This paper proposes a new method for tracking the entire trajectory of a ballistic missile from launch to impact on the ground. Multiple state models are used to represent the different ballistic missile dynamics in three flight phases: boost, coast and reentry. In particular, the transition probabilities between state models are represented in a state-dependent way by utilising domain knowledge. Based on this modelling system and radar measurements, a state-dependent interacting multiple model approach based on Gaussian particle filtering is developed to accurately estimate information describing the ballistic missile such as the phase of flight, position, velocity and relevant missile parameters. Comprehensive numerical simulation studies show that the proposed method outperforms the traditional multiple model approaches for ballistic missile tracking.

Launch point prediction employing the smoothing IPDA algorithm in 3-D cluttered environments

“…Actually, [17] defines two new parameters, the net axial acceleration and the mass depletion rate, but in this study, we ignored mass depletion rate, because there is no thrust force, so the mass of artillery projectile is constant. The dynamic equation of CAF model is given by…”

“…In addition, it is helpful to find the launch point when the track is confirmed rapidly. In this paper, we apply the sIPDA algorithm, which has better target state estimation performance and FTD capabilities than the IPDA algorithm, to track artillery projectiles in the cluttered environments, and LPP is carried out by numerically integrating the constant axial force (CAF) model [3,17]. The net axial acceleration of a projectile is estimated from the CAF model by using the acceleration estimate of the sIPDA algorithm employing the nearly constant acceleration (NCA) model [9].…”

Closed form 3d polynomial solution from multiple 2d sensors

“…In [5] a 3D ballistic missile tracker is initialized using weighted least squares from radar measurements. The algorithm presented here can do the same job as well as utilize both active and passive sensors and not be limited to active sensors only.…”

“…Nonlinear problems require an iterative solution to achieve optimal or near optimal estimates [1][2][3]. Some proposed methods for initializing iterative estimators for passive and active sensors require an iterative solution for the initialization [4,5]. In [4], with passive-only satellite sensors, an altitude is assumed for the first measurement to get the iterative procedure started.…”

A novel suboptimal algorithm for state estimation of Markov jump linear systems

“…Among these algorithms, the IMM algorithm is the most popular, which obtains good performance at modest computational load. It is effective in applications such as target tracking [19].…”