On Zero Memory Nonlinear Transformations of Gaussian Processes

Abstract

This paper treats the second moment properties of a zero memory nonlinearity, given that the input is a stationary Gaussian process. The output autocorrelation function is shown to be expressed conveniently in terms of the input autocorrelation function and a set of coefficients describing the ZNL. Two theorems are proved concerning the output process bandwidth. The first shows that the output bandwidth is generally greater than the input bandwidth. The second gives necessary and sufficient conditions for the output process to be strictly bandlimited.