System identification using Measurements Subject to Stochastic Time Jitter
| Authors: | Eng Frida, Linkopings universitet, Sweden
Gustafsson Fredrik, Linkopings universitet, Sweden |
|---|
| Topic: | 1.1 Modelling, Identification & Signal Processing |
|---|
| Session: | Frequency Domain Estimation Methods |
|---|
| Keywords: | signal processing, sampling, stochastic properties, spectrum analysis,system identification, noise characteristics |
|---|
Abstract
When the sensor readings are perturbed by an unknown stochastic timejitter, classical system identification algorithms based on additiveamplitude perturbations will give biased estimates. We here outline the maximum likelihood procedure, for the case of bothtime and amplitude noise, in the frequency domain, based on themeasurement Discrete Time Fourier Transform (DTFT). The method directly applies to output error continuous time models,while a simple sinusoid in noise example is used to illustrate the biasremoval of the proposed method.
