The centralized Kalman filter and the federated Kalman filter
are used respectively to fuse data of this process, the error
curves are shown in Fig 2 and Fig 3. These figures shows the
accuracy of the centralized Kalman filter is higher when the
system model is accurate, but the velocity and position errors
increase significantly if the statistical properties of noise are
not accurate. Because the federal Kalman filter presented in
this paper uses distributed filter in which each sub-filter
using adaptive Kalman filter, even if the statistical properties
of measurement noise changing, model still able to describe
the statistical properties of measurement noise accurately by
the adaptive adjustment, the accuracy and reliability are
better than those of the centralized Kalman filter