Robust Identification from Impulse and Step Responses

A robust identification method that is insensitive to anomalous data is presented to estimate system models from the samples of their impulse and step responses. The power and usefulness of the proposed approach is illustrated with the help of simulated and real-life examples.