Satellite Phase and Code Bias Estimation with Cascaded Kalman Filter

This paper suggests a new method for the estimation of satellite phase and code biases with undifferenced, uncombined phase and pseudorange measurements from a global network of reference stations. The method is based on a generalized measurement model with individual phase and pseudorange biases for each satellite, and uses a generalized Kalman filter for coloured measurement noise. The estimation of phase and code biases is performed with a cascade of two Kalman filters: First, the nondispersive parameters of each receiver-satellite link are combined into a single geometry term, and a conventional Kalman filter is used to estimate the geometry term, ionospheric delays, and individual ambiguities. The mapping of all non-dispersive parameters to single geometry terms improves the conditioning of the system of equations and, thereby, enables a faster ambiguity fixing. A second filter is then used to refine the geometry term, i.e. to estimate orbital errors, code biases and tropospheric delays. As the a posteriori estimates of the first Kalman filter are correlated over time and correspond to the measurement of the second filter, the generalized Kalman filter of Bryson and Henrikson is used to whiten the measurement noise and to keep the measurement and process noises independent. One of the largest challenges for reliable precise point positioning with integer ambiguity resolution is code multipath, which has to be considered carefully at both the reference stations and mobile receiver. For the reference station, a sidereal filtering can be applied to the pseudorange residuals to efficiently obtain multipath corrections. It is shown that these corrections whiten the measurement noise and result in substantially improved stability of the bias estimates.      «

This paper suggests a new method for the estimation of satellite phase and code biases with undifferenced, uncombined phase and pseudorange measurements from a global network of reference stations. The method is based on a generalized measurement model with individual phase and pseudorange biases for each satellite, and uses a generalized Kalman filter for coloured measurement noise. The estimation of phase and code biases is performed with a cascade of two Kalman filters: First, the nondispersi...     »