Accession Number : AD0669510

Title :   ON OPTIMAL AND SUBOPTIMAL LINEAR SMOOTHING.

Descriptive Note : Technical rept.,

Corporate Author : HARVARD UNIV CAMBRIDGE MASS DIV OF ENGINEERING AND APPLIED PHYSICS

Personal Author(s) : Mehra,R. K.

Report Date : MAR 1968

Pagination or Media Count : 36

Abstract : Recursive form of results on smoothing for linear dynamic systems were first given by Bryson and Frazier (1962). Alternate formulations of the problem were given by Rauch et al (1965), Mayne (1966), Fraser (1967) and Kailath (1968). The present report shows that the results of Mayne and Fraser can be derived in a more general setting using the Orthogonality Principle of Linear Estimation. The form of the results is particularly useful for the sensitivity analysis of the optimal smoother. Explicit equations are derived for the actual covariance of a suboptimal smoother which uses wrong information about the mean square values of noise inputs. Conditions are established under which the calculated values of the covariances provide upper bounds on the actual covariances of the smoothed estimates. (Author)

Descriptors :   (*INFORMATION THEORY, NUMERICAL ANALYSIS), OPTIMIZATION, VECTOR SPACES, STATISTICAL PROCESSES, SENSITIVITY, RANDOM VARIABLES, ERRORS, THEOREMS

Subject Categories : Cybernetics

Distribution Statement : APPROVED FOR PUBLIC RELEASE