Suboptimal State Estimation under Communication Delays

We present a suboptimal state estimation method under communication delays. The missing measurements are replaced with model predictions and these predictions are put into memory. When the delayed measurements arrive, the effect of using prediction as fictitious measurements is removed. The optimality is recovered, which this is the key property of the method. The algorithm complexity is low compared to the optimal solution. The method requires the measurements to be time stamped. The plant model must be linear, time invariant.