Knowledge of differential inter-systems phase/code biases are critical to integrate observations from several different Global Navigation Satellite Systems. Many studies on tightly-combined model have mostly focused on the overlapping frequencies and short baselines, and fewer have focused on non-overlapping frequencies and medium-long baselines. Considering differential inter-systems code biases and inter-frequency code biases, we propose a tightly combined real-time kinematic model by a reparameterization process for non-overlapping frequencies among GPS/GLONASS/BDS systems. Compared with the traditional double-difference ambiguity parameter, we use single-difference ambiguity as the parameter to be estimated, which is not affected by reference satellite switching and wavelength limitation. The paper first analyzes the stability of differential inter-systems code biases and inter-frequency code biases with several zero and ultra-short baselines. Then, the paper tests the positioning performance under occlusion environment. The experiment shows that the tightly-combined model is more robust than the loosely-combined model and the average time to first fix is shortened by approximately 22.4%. At last, the paper compares the positioning performance of tightly-combined model, loosely-combined model, and tightly-combined model after calibration under the medium and long baselines. The experiment shows that the tightly-combined model after calibration is better than the tightly-combined model and loosely-combined model in the ambiguity resolution rate, initialization speed and positioning accuracy, respectively. Compared with intra-system loosely-combined model, the positioning accuracy of tightly-combined model after calibration is improved by 42.1%, 25.5%, and 32.3% in the three-dimensional directions, respectively.
Read full abstract