A Robust Minimum Distance Detection Rule in the Neyman-Pearson Setting

In practice, it is often that noise distributions are not Gaussian and may vary in a wide range from short-tailed to heavy-tailed forms. To provide stable and high quality detection of a known signal, a robust (asymptotically minimax in the Huber sense) Neyman-Pearson (NP) minimum distance detection rule is designed. Explicit formulas for the false-alarm probability and detection power are derived. For several distribution classes, the least favorable distributions and the corresponding robust minimax detectors are obtained. Some numerical results on their performance are given