Development of a Method for Kinematic GPS Carrier-Phase Ambiguity Resolution Using Multiple Reference Receivers

To perform the most precise relative positioning using GPS, it is necessary to resolve the carrier-phase integer cycle ambiguities. This process becomes increasingly difficult as the distance between the mobile and reference GPS receivers increases, due to the decorrelation of the GPS errors with distance, resulting in a practical limit on the distance over which ambiguity resolution can be performed when using a single reference receiver. This thesis proposes a novel method, called NetAdjust, which uses multiple reference receivers to reduce code and carrier-phase differential measurement errors and improve the ability to resolve carrier-phase ambiguities. The NetAdjust method is based upon an optimal linear minimum error variance estimator, and it “encapsulates” all of the network information into the measurements of a single reference receiver, so standard single-reference differential GPS processing algorithms can be used. The geometry of the reference receiver network is embedded within the error covariance matrix, and a functional form of this covariance matrix is described. The NetAdjust method was tested using two different GPS networks—an 11-receiver network covering a 400 km 600 km region in southern Norway, and a 4-receiver network covering a 50 km 150 km region at Holloman AFB in New Mexico. The results for L1 code, L1 phase, and widelane phase measurements are analyzed in the measurement domain and the position domain, showing improvements in RMS errors of up to 50% when using NetAdjust. Significant improvements in the ability to resolve carrier-phase ambiguities are also demonstrated for the Holloman and Norway test networks. Issues relating to development of an operational, real-time NetAdjust system are discussed. Also, a covariance analysis method is developed which can be used to predict NetAdjust effectiveness under various conditions and network configurations. This covariance analysis demonstrates that, for the conditions present during the Norway test, the network of eleven reference receivers is sufficient to perform widelane ambiguity resolution, but it is not sufficiently accurate for L1 ambiguity resolution throughout the network.