Maximum Likelihood Approach to the Estimation and Discrimination of Exoatmospheric Active Phantom Tracks using Motion Features

An optimal kinematics-based discrimination algorithm is presented for the nearly real-time discrimination of exoatmospheric active phantom (deception) tracks against other physical targets (PTs). The new approach uses a batch-processing maximum likelihood estimator (MLE) to precisely estimate the deception range of decoys from raw radar tracking measurements by using motion features. Hence once these parameters are estimated, they can serve as direct statistics to initiate a discrimination. By augmenting the state vector with deception parameters, explicit expressions of motion models of decoys in the radar centered East-North-Up (ENU) coordinate system (CS) and spherical-CS are derived. Based on these models, the theoretical Cramer-Rao lower bound (CRLB) and the observability of the deception parameters are also analyzed. A Levenberg-Marquardt method is employed to obtain more robust estimate of these parameters, and the estimated parameters combined with the CRLB are used for designing discrimination algorithm. The simulations verify the feasibility of the algorithm. Furthermore, the discrimination performance due to the influence of radar position, data rate, and radar measurement error are also covered. The advantage of the algorithm lies in that it can position the warhead precisely as well as discriminating active decoys simultaneously when compared with the traditional 6-dimensional orbit determination methods.