The design of ternary perturbation signals for linear system identification

Authors:	Barker Tony, University of Wales Swansea, United Kingdom Tan Ai Hui, Multimedia University, Malaysia Godfrey Keith, University of Warwick, United Kingdom
Topic:	1.1 Modelling, Identification & Signal Processing
Session:	Experiment Design
Keywords:	Linear systems; Maximum-length sequences; Multilevel signals; Perturbation signals; Pseudorandom signals; System identification

Abstract

A method is described for designing ternary perturbation signals to reduce the effects of both nonlinear distortion and noise in the identification of linear systems. This is achieved by suppressing harmonic multiples of 2 and 3 in the signals and optimizing three performance indices. Methods are given for designing primitive ternary signals to meet these criteria, and results are tabulated for signals with periods up to 78. It is then shown that these signals can be used to design pseudorandom ternary signals with longer periods that also meet the criteria. Results are tabulated for these signals and illustrated by an example.