On-the-fly ambiguity resolution for real-time kinematic differential GPS positioning

To achieve centimeter level accuracy real-time kinematic differential GPS positioning, the inherent integer cycle ambiguity of carrier-phase observations must be correctly resolved. It is the objective of this study to develop a fast, reliably and sophisticated algorithm for on-the-fly ambiguity resolution. The first part of this study investigated three major components of the AROTF techniques, namely, the ambiguity estimation process, the ambiguity search process, and the ambiguity identification process. Emphasis was placed on the characteristics of unmodeled double-difference residual biases and the necessary processes to remedy the inadequate modeling of the simplified observation model. The second part of this study is devoted to the implementation of a new AROTF technique, which is referred to as the Multi-Stage Least-Squares Search (MSLSS) algorithm. The minimum variance estimation with stochastic a priori covariance is used to generate the float solution. The resulting ambiguity search space is reduced significantly, and the high correlation between estimated parameters will be decorrelated considerably when the a priori information is given. Special strategies to optimize the ambiguity search speed were provided. Various hypothesis tests based on rigorous statistical background were employed to identify the correct integer set of ambiguities. The concepts and applications of global ambiguity resolution to enhance the robustness and reliability of AROTF technique were discussed. Test results for both static and kinematic positioning campaigns were also analyzed and presented. The centimeter level accuracy of real-time kinematic differential GPS positioning is achievable through the demonstrations. Five verification methods were applied to verify the correctness, reliability and repeatability of the MSLSS algorithm. Test results indicates the proposed AROTF technique is remarkably promising.