Mean square robustness of linear game theoretic estimators

Robust signal processing methods deal with design of estimators that exhibit resilience to uncertainties in the user specified inputs regarding the operating conditions. The H∞ estimator was proposed as a possible replacement to the Kalman filter while operating in the presence of uncertainties in the statistical and parametric inputs to the filter. The results in this paper compare the Mean Squared Error (MSE) performance (or MSE robustness) of the H∞ estimator and the Kalman filter under conditions of misspecified statistics. This study would facilitate the user to choose the better estimator under the given operating conditions. The results presented here show that under different operating conditions both the filters conditionally dominate one over the other.