This report studies two aspects of tracking ballistic missiles during boost phase. The first part compares the performance of several nonlinear filtering algorithms in tracking a single target: the extended Kalman filter (EKF); the unscented Kalman filter (UKF); the particle filter (PF); and the particle filter with UFK update (UPF). Measurements are range, azimuth and elevation. In the absence of measurement error, all algorithms work well except for the PF, which does not converge. With measurement noise (standard deviations of 10 meters and 1 degree) the EFK also performs poorly, while the UPF is the top performer (although it is also the most computationally intensive). The second part compares the extended information filter (EIF) with earlier work on track scoring to perform sensor/data fusion in a multi-hypothesis framework. Here we find that the EIF handily outperformed other fusion algorithms based on track scoring that we tested
展开▼
