Federated filter for computer-efficient, near-optimal GPS integration

A simple, add-in implementation of the federated filter has been developed that is well-suited to integrating off-the-shelf GPS navigation equipment with new or existing navigation systems. This implementation provides near-optimal estimation accuracy for the integrated system, with a minimal increase in memory and throughput requirements for the existing navigation processor. This method can readily replace the common, loosely-coupled cascaded (LCC) approach to integrating off-the-shelf GPS navigators with existing systems. The LCC approach processes the GPS output solution in a cascaded Kalman filter, and ignores that fact that successive GPS filter outputs actually contain both old and new information. The federated filter solves the old/new information problem by maintaining separate partitions for information derived from GPS and other sources. These partitions are periodically combined by a fusion algorithm that provides near-globally optimal estimation accuracy. The processing algorithms, logic, and associated computer burden are very similar to those of the LCC filter. Performance simulation results are shown for the federated filter, LCC filter, and a globally optimal centralized filter.