It is well known that the type and quality of measurements play a key role in determining the quality of the results in an estimation application. It often occurs that the measurement subsystem is specified except for certain parameters. This research presents a method whereby these parameters may be selected so as to enhance the resulting estimates. The problem is posed as the deterministic optimization of a performance index embodying the desired results with respect to the estimator error covariance and the costly measurement parameters. The error covariance differential equation of the assumed first order conditional mean estimator enters as an equality constraint on the optimization. Necessary conditions for the solution result from an application of the matrix minimum principle. Both linear and nonlinear continuous system and measurement models with additive white noise are considered. Several examples illustrate how the method is applied. Numerical results show how the optimized parameters respond to various specifications in the performance index and in the system and measurement models. (Author).

Project 7904. This report was originally prepared as the author's thesis. Also available as AD759175.