Improved float ambiguities for real-time GNSS attitude determination via UKF

The accuracy of float ambiguities affects the performance of integer ambiguity resolution. This paper proposes an improved method for float ambiguity estimation in real-time attitude determination using single-frequency Global Navigation Satellite System carrier phase measurements. The functional relationship between double-differenced phase observations and attitude quaternion is formulated to fully exploit all the available information of the multi-antenna geometry configuration. Considering the nonlinearity of the system, the unscented Kalman filter is developed to simultaneously estimate float ambiguities and attitude quaternions. The least-squares ambiguity decorrelation adjustment method is further used to fix the float solutions and modify the attitudes. A static ultra-short baseline (0.55 m) BeiDou test is carried out to evaluate the performance of the proposed method. Results show that the accuracy of float ambiguities is improved compared to conventional baseline-solving techniques and higher success rate and less Time-to-Fix are achieved for integer ambiguity resolution. In addition, the robustness to reduction of the number of available satellites is also demonstrated.