Comparison of Distributed and Federated Filtering in Multi-Coordinate Systems

Introduce two important equations used in Kalman filter algorithm with coordinates-converting errors. Analyze the characteristic of distributed Kalman filtering algorithm with coordinates converting uncertainties in multi-coordinate sensor systems. Then, an improved federated filtering algorithm is introduced in the same systems. In the new algorithm, local processor can get global optimal estimation by the transformation of measurement equation and coordinate-converting equation. Accordingly, the Kalman filtering algorithm is transformed. Based on these mathematic methods, we only need coordinates converting once to obtain global estimation, which are always needed twice in the distributed algorithms. So, the filtering accuracy is improved. Simulation results also show that the federated algorithm has a better performance in improving the estimation accuracy