A Novel Decision Directed Kalman Filter to Improve RTK Performance

G. Van Meerbergen, A. Simsky, F. Boon

Abstract:	It is well known that GNSS Kalman _lters solely based on a white noise assumption produce underestimated variances of position and ambiguity states, which pre- vent the development of statistically consistent ambi- guity validation methods. In this paper we address this problem. Our approach is based on the colored- noise form of the Kalman _lter, which signi_cantly im- proves _xing performance and allows for a Decision Directed Mode, a more _exible manner of di_eren- tial phase processing than a traditional combination of _oat and _xed modes. We begin with a development of more realistic noise models, which take into account time correlation of measurements (colored noise). To handle such models, a more sophisticated form of the Kalman _lter is used, which produces more realistic variances of position and ambiguity states. With better knowledge of ambiguity variances, more rigorous methods of ambiguity _xing can be employed. Our ambiguity _xing algorithm uses a so-called A-Posteriori Probability (APP) test, based on Bayesian considerations and purely statistical by nature. A signi_cant advantage of APP is in its ability to assess the individual (i.e. per satellite) reliability of ambiguity candidates. Instead of the binary choice of RTK modes (_oat or _xed) a more _exible Decision- Directed Mode (DDM) becomes possible, where any amount of ambiguities can be integer or _oat, and only the estimation of position accuracy tells the user how precise is the RTK solution. The results of time to _rst _x tests have demon- strated the signi_cant improvement of the _xing per- formance with the use of the new RTK engine, in par- ticular for longer baselines.
Published in:	Proceedings of the 23rd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2010) September 21 - 24, 2010 Oregon Convention Center, Portland, Oregon Portland, OR
Pages:	3268 - 3276
Cite this article:	Van Meerbergen, G., Simsky, A., Boon, F., "A Novel Decision Directed Kalman Filter to Improve RTK Performance," Proceedings of the 23rd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2010), Portland, OR, September 2010, pp. 3268-3276.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In