Information distribution in equivalent measurements

When carrying out track-to-track fusion, the use of equivalent measurements from non-collocated trackers is a novel alternative because it gives optimal results while providing bandwidth savings of more that 50 percent over the traditional track-to-track fusion algorithm based on the processing of target state estimates and their predictions. Furthermore, the use of equivalent measurements provides an optimal solution to the problem of joint registration and track-to-track fusion when sensor-level tracks from non-collocated trackers contain registration errors. Equivalent measurements are generated by inverse filters and are initially presented in target state space irrespective of the sensor(s) used to generate the original raw measurements. They represent new information at stage k that was not available at stage k - 1. We demonstrate that the distribution of useful information among the components of an equivalent measurement depends on the sensor(s) that generated the original raw measurements. When the sensors generate location measurements, the resulting equivalent measurements in target state space will not contain velocity information. This implies that when the sensors responsible for generating the original raw measurements are known, it is possible to obtain additional bandwidth savings without loss of information by converting the equivalent measurements from the target state space to the measurement space of the sensors. This is important because communication is generally more expensive than processing for each bit of information handled by the network.