Feature-Aided RF/IR Track Correlation

Missile tracking sensors such as radar and EO/IR are often called upon to track constellations of multiple closely spaced objects, and recognize pairs of matching tracks from each sensor"s view. The outcome of this track correlation process is crucial for missile defense interceptors, as any failure to identify matches with certainty, for example because of an excessive number of tracks associated with debris, will likely mean that information gathered prior to the interceptor"s own acquisition of the scene can"t be fused reliably and will effectively be rendered useless. The current practice in the Aegis BMD system is for the radar to deliver a target object map (TOM) message to the interceptor, which the interceptor matches against the EO/IR tracks in its own field of view using a distance-based likelihood metric. However, both the RF and EO/IR sensors also have time series signature data available from the objects they are tracking. Physical Sciences Inc. (PSI) and Johns Hopkins University Applied Physics Laboratory (JHU/APL) propose to incorporate this additional information by reducing the signature data to a small set of characteristic features, and then combine the resulting feature values with existing distance-based metrics to produce an augmented track correlation solution.