DETECTION OF NONLINEARLY DISTORTED SIGNALS USING MUTUAL INFORMATION (MonPmPO2)

Author(s) : Umut Ozertem (OHSU, United States) Deniz Erdogmus (OHSU, United States) Ignacio Santamaria (University of Cantabria, Spain)

Abstract : The traditional matched filter is the optimal signal detector under quite restrictive conditions, such as linearity and Gaussianity. In this paper, a nonlinear filter topology based on mutual information is proposed to exploit higher order statistics instead of linear second-order measures. Results demonstrate superior performance in nonlinear amplitude and temporal signal distortion situations.

Menu