Asymptotic Robustness Coefficient for Signal Detection Algorithms

Abstract

Both the form of the noise distribution and a number of signal and interference parameters can be a priori uncertain in many practical cases of signal detection. Therefore, the ability to maintain detection algorithm performance in the case of noise environment changing is one of the key problems. In the most cases at the stage of synthesis, it is not possible to ensure the independence of the average losses of an algorithm on the noise distribution, which makes it necessary to estimate the stability of the average losses of algorithms in the case of nonparametric a priori uncertainty. This property is usually called robustness. In this paper, we propose the asymptotic robustness coefficient which can be used for estimating the robustness of signal detection algorithms.