A Class of Signal Processing Procedures Suggested by Statistically Optimum Procedures

摘要

This paper investigates a class of data-smoothing procedures for use in problems where the objective is to estimate a set of parameters, on the basis of observed data which depend on these parameters and are also corrupted by noise. These procedures are in part motivated by possible application to orbit estimation problems, which the author believes may provide instances where the suggested procedures are computationally more convenient than those more commonly employed. However, potential applications exist to many other parameter estimation problems. Some of the results are of general interest to parameter estimation theory.