scientific article; zbMATH DE number 4018920

From MaRDI portal
Publication:3030700





[For the entire collection see Zbl 0614.00003.] On one hand the authors consider a ``physical signal process \(z_ P'(t)\) and its noise corrupted observation \(y_ P(t)\) which may be regarded as the solutions of linear differential equations with wide bandwidth noise excitation and on the other hand their ``idealized counterparts \(z_ i(t)\) and \(y_ i(t)\) which are the solutions of the corresponding linear Itô stochastic differential equations resulting from the assumption of ideal white noise excitation. As pointed out by an example, the optimal Kalman filter reconstructing \(z_ i(t)\) from \(y_ i(t)\) needs not to be the optimal filter reconstructing \(z_ P(t)\) from \(y_ P(t)\). The authors prove that the Kalman filter is ``nearly optimal with respect to a quite broad class of alternative filters, however. They treat the asymptotic problems in time and bandwidth, as well. The paper is an outline of a more detailed report to appear in IEEE-AC.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3030700)