Gaussian mixture PHD filter for space object tracking

Abstract

A Gaussian mixture Probability Hypothesis Density (PHD) filter for multiple space object tracking is presented. The PHD filter is a computationally tractable approximate Bayesian multi-object filter based on finite set statistics. The intensity of the Gaussian mixture PHD filter is represented by a variable-size Gaussian mixture, which is propagated and updated by a Gaussian mixture filter that accounts for the nonlinear effect of long-term orbit propagation. A numerical example is used to demonstrate the viability of the filter for space object tracking. © 2013 2013 California Institute of Technology.