Smoothing and prediction filters have a wide range of applications, e.g., communications, signal processing, radar tracking, orbital parameter estimation and spacecraft attitude determination, to name just a few.   Over the years, a trend has developed that equates smoothing and prediction with the Kalman filter.   However, the field has a history going back to Gauss, and so there are many other approaches that also deserve our attention.   This talk will present four classes of filter techniques: Polynomial filters, Gauss-Newton differential correction, the Bayes-Swerling filter and the Kalman filter, and will identify areas of commonality and difference between them.   As a specific example, the performance of these filters will be contrasted when used for tracking near-earth artificial satellites.

Dr. Norman Morrison of the University of Cape Town, South Africa, has a 40 year history in smoothing and prediction.   He received his BSc from the University of the Witwatersrand, Johannesburg in 1952, MSc from the University of Akron in 1961 and PhD from Case Western Reserve University in 1965.   He has worked with several companies in the US defense industry, as well as Bell Labs, on issues relating to smoothing and prediction, and has written a textbook entitled "Introduction to Sequential Smoothing and Prediction", McGraw Hill, 1969.