On sensitivity of detector structures to contaminants in non-white mixture noise model

In this paper, we study the structure of the locally optimal detector in non-white transformation noise with -contaminated marginal density. We show that each of the memoryless nonlinearities in the detector can be at least approximately represented as a term for the nominal density plus a correction term in closed form that is directly proportional to . The approximations are found, by simulation, to be good in most cases except when is large and the tail weight of the contaminant is large. We also show that the eigenfunctions of the bivariate noise density can be put in a similar form. This paper was presented in part in [5]. Oscar C. Au is with the Department of Electrical and Electronic Engineering, Hong Kong University of Science and Technology.