Robust DFT based on adaptive censored estimate for FM signal processing in non-Gaussian noise environment

A problem of denoising frequency-modulated signals embedded in non-Gaussian noise by means of robust DFT is considered. The requirements for robust estimation implied within robust DFT denoising framework are formulated. It is shown that several standard robust estimators do not meet these requirements well enough. A new adaptive robust estimator is proposed and tested. The robust DFT form based on the proposed estimator is compared to several known robust DFT-methods. Advantages of using the proposed adaptive censored estimator for the considered application are demonstrated via numerical experiments.