Autocorrelation Functionsin GPS DataProcessing: Modeling Aspects

Kai Borre, Gilbert Strang

Abstract:	Today the processing of GPS data has reaehed the limit of modeling with systematic non-random parameters. In or-der to go further one has to model the random nature of observation and state equation errors. The autocomelation functions from the residuals deserve study. Such functions may describe correlation both in time and space. Covari-ance functions and covarianee matrices derived from auto-correlation functions play a key role in this analysis. Based on real-world examples we discuss autoeorrelation functions to be used in sequential filters, to model the ran-dom nature of the parameters. Three random models fre-quently used are white noise, random walk and exponen-tially correlated (Gauss-Markov). These studies are all made by means of a series of MAT-LAB files that primarily are intended for educational and further software development. ‘Ihose files are linked to our new textbook (Strang & Borre, 1997).
Published in:	Proceedings of the 10th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 1997) September 16 - 19, 1997 Kansas City, MO
Pages:	1143 - 1150
Cite this article:	Borre, Kai, Strang, Gilbert, "Autocorrelation Functionsin GPS DataProcessing: Modeling Aspects," Proceedings of the 10th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 1997), Kansas City, MO, September 1997, pp. 1143-1150.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In