Title :
Robust noncausal filtering
Author_Institution :
Exploration & Min., CSIRO, Kenmore, Qld., Australia
Abstract :
Filter asymptotes are developed for output estimation and input estimation problems which lead to bounds on the spectrum of the estimation error and a priori estimates for the scalar γ in the H∞ design. State-space formulations for the optimal minimum-mean-square-error and suboptimal noncausal filters are presented. It is demonstrated that an H∞ approach can improve on conventional maximum-likelihood and conditional-mean techniques in a speech filtering application.
Keywords :
H∞ control; filtering theory; maximum likelihood estimation; mean square error methods; parameter estimation; state-space methods; H∞ design; conditional mean techniques; conventional maximum likelihood; estimation error; input estimation; optimal minimum mean square error; output estimation; parameter estimation; robust noncausal filtering; speech filtering; state-space formulations; suboptimal noncausal filters; Australia; Estimation error; Filtering; Maximum likelihood estimation; Nonlinear filters; Riccati equations; Robustness; Speech; State estimation; Uncertainty;
Conference_Titel :
Decision and Control, 2003. Proceedings. 42nd IEEE Conference on
Print_ISBN :
0-7803-7924-1
DOI :
10.1109/CDC.2003.1272592
