Random Functions with Nonlinear Regression and their Application
V. V. Gubarev
Novosibirsk State Technical University
gubarev@vt.cs.nstu.ru
Keywords: nonlinear signals, random functions, two-dimensional distributions, regression functions, concorrelation functions, spectral densities
Pages: 39-50
Abstract
This paper presents a mathematical description of the dynamic properties of nonlinear random physical signals, i.e., signals modeled by random functions of time (processes and sequences) with a nonlinear the regression. It is shown that traditional methods of correlation and spectral analyses are poorly suitable or not at all for representing their dynamics. As an alternative, the paper gives a description and analysis of such signals using time (concorrelation functions) and frequency (conspectral density) concors. They exist for all random functions and the modulus of their values re invariant to any instantaneous one-to-one monotonic transformations of signals.
|
