Faster Kalman trajectory smoothing
Brian Billard · AIAA Journal · 1977
Innovations are described which greatly speed up the operation of Kalman filtering and smoothing when applied to trajectory analysis on a limited main storage computing system. They include the use of a variable lag smoother, and the detection and bypassing of repetitive calculations when the parameters of the filtering system are changing only slowly. The innovations have been implemented in a program which computes a missile trajectory from ground-based instrumentation data. The program has been proven in use on a production basis for some time using both an IBM 7090 and, more recently, an IBM 370/168. Tests of the filtering and smoothing cycle times have been made, and they provide insight into the advantages of the innovations described.