Robust adaptive systems identifications by using median smoothing

The authors present a robust adaptive ARMA (autoregressive moving average) systems identification algorithm for systems having additive noise which is composed of Gaussian noise and a number of wild-points. The influence of adaptive systems identification caused by wild-points is cancelled perfectly. Thus, the robust performance of adaptive systems identification is improved, and the systems estimate is consistent and asymptotically normal. Simulation results on the robust adaptive equation error ARMA system identification are presented