Fixed-interval smoothing is a method of extracting useful information from inaccurate data. It has been applied to problems in engineering, the physical sciences, and the social sciences, in areas such as control, communications, signal processing, acoustics, geophysics, oceanography, statistics, econometrics, and structural analysis. This monograph addresses problems for which a linear stochastic state space model is available, in which case the objective is to compute the linear least-squares estimate of the state vector in a fixed interval, using observations previously collected in that interval. The author uses a geometric approach based on the method of complementary models. Using the simplest possible notation, he presents straightforward derivations of the four types of fixed-interval smoothing algorithms, and compares the algorithms in terms of efficiency and applicability. Results show that the best algorithm has received the least attention in the literature. This volume also includes new material on interpolation, fast square root implementations, and boundary value models.
Features include: an annotated bibliography of smoothing literature; simple notation and clear derivations; compares algorithms from a computational perspective; and identifies a best algorithm. "Fixed Interval Smoothing for State Space Models" is for those wanting to understand and apply fixed-interval smoothing: academics, researchers, and graduate students in control, communications, signal processing, statistics and econometrics.