On Detection-Estimation Schemes for Signals with Uncertain Models.
Abstract
A new approach to the problem of estimating a random signal whose statistical description contains unknown parameters is presented. The uncertainty in the parameters is described by means of several possible parameter sets corresponding to various levels of uncertainty about the signal model. A joint detection-estimation scheme is introduced for the estimation of this class of signals. The joint detection-estimation scheme consists of a bank of estimators, one for each parameter set, and a detector to select the most appropriate estimator for each realization of the signal. As standard minimax or adaptive approaches are not appropriate for the given description of uncertainty, a weighted minimax criterion is proposed. This performance criterion can be thought of as resulting from a multi-objective optimization approach where the maximum of each component of the objective function is to be minimized. The optimization of the joint detection estimation scheme with respect to the weighted minimax criterion is reformulated as a standard minimax problem in an augmented parameter space. Results of existence and characterization of solutions are obtained.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1976
- Accession Number
- ADA029906
Entities
People
- Rafael Antonio Padilla Lovera
Organizations
- University of Illinois Urbana–Champaign