DIstributed Kalman Filtering Using The Internal Model Average Consensus Estimator

We apply the internal model average consensus estimator in [1] to distributed Kalman filtering. The resulting distributed Kalman filter and the embedded average consensus estimator update at the same frequency. We show that if the internal model average consensus estimator is stable, the estimation error of the distributed Kalman filter is zero mean in steady state and has bounded covariance even when the dynamical system to be estimated is neutrally stable or unstable.