Approximations to Robust Wiener Filters.
Abstract
Several approximations to the robust Wiener filters for two uncertainty classes of signal and noise spectra are considered. The approximations consist of realizable n-th order filters of the Butterworth and Chebyshev types. For each uncertainty class, the worst-case performances of the approximate and ideal robust filters are compared. It is found that, for the epsilon-contaminated uncertainty class, the approximate robust filter using only second-order Butterworth filters gives a very good approximate robust filter using only second-order Butterworth filters gives a very good approximation. For the p-point class, where there is more uncertainty in the knowledge of the true spectra of the signal and noise, it is necessary to use filters of relatively high orders to obtain good approximations. Furthermore for this class, the best approximation is obtained by using Chebyshev filters of high orders and small ripple. Keywords: Thesis; Robust signal estimation; Minimax design; Finite-order filters.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 04, 1984
- Accession Number
- ADA161321
Entities
People
- Myrna Roula Cotran
Organizations
- University of Illinois Urbana–Champaign