ESTIMATING AND DETECTING THE OUTPUTS OF LINEAR DYNAMICAL SYSTEMS
Abstract
This investigation considers three closely related problems: the optimum filtering of stationary or near-stationary random processes with unknown parameters from an infinite parameter set; estimation of the state of a linear discrete dynamical system with nongaussian noisy inputs; and applications of state estimation theory to detection. The form of the optimum filter when the parameters are unknown is found to have weights that are averages of simple functions of the signal and noise spectra averaged over the parameter space. Practical methods for implementation are given. The key problem is nonlinear state variable estimation is obtaining the joint density of the states and the observations in a convenient form.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1964
- Accession Number
- AD0464023
Entities
People
- C. S. Weaver
Organizations
- Stanford University