Back to results list
Please use this identifier to cite or link to this item:
|Title:||Modeling systematic errors and improving computation efficiency in GNSS positioning||Authors:||Yu, Wenkun||Advisors:||Ding, Xiaoli (LSGI)||Keywords:||Global Positioning System
Artificial satellites in navigation
|Issue Date:||2019||Publisher:||The Hong Kong Polytechnic University||Abstract:||The emergence of multiple Global Navigation Satellite Systems (GNSSs) improves the observation redundancy and positioning accuracy, especially in areas where sky views are restricted. However, joint use of multi-GNSS observations also brings challenges to data processing due to the extra biases, systematic errors, heteroscedastic error structures, more frequent multiple-outliers and higher computational resources required. This study focuses on four aspects of multi-GNSS positioning. First, the advantages of using multi-GNSS observations in positioning are investigated. It is shown that compared with GPS-only solutions the total number of satellites in the case of the current multiple GNSS constellations increases by 280% and the position dilution of precision (PDOP) reduces by 52.4%. The improvements of the satellite number and PDOP will further rise to 340% and 57.1% respectively when all the satellite systems reach their full constellations. It is especially advantageous to use multiple GNSSs for positioning in difficult observational environments with significant blockage of satellite signals. The real-data static relative positioning test shows that the positioning error decreases by up to 52.4% when using the current quad-constellations (i.e. GPS, GLONASS, BeiDou, and Galileo). Simulation experiments show that when all four systems reach their full constellations the positioning accuracy can improve by 48.5-69.0%. Second, a new positioning model is developed to account for systematic errors in multi-GNSS positioning. Unmodeled systematic errors, such as multipath effects and residual atmospheric delays, can impact on the estimated positions and the variance components. A semiparametric estimation model is developed for better mitigation of the systematic errors. Test results with simulated systematic errors show that systematic errors can be accurately estimated with the proposed approach. A test with three-day real GNSS observations from a short baseline demonstrates that, compared with standard least-squares estimation, when combined with variance component estimation the proposed method can improve the accuracy of the estimated static baseline by 35.6%. A simplified procedure based on least-squares residuals is proposed to enhance the determination of smoothing parameters, which has been proved practical and effective. Experimental results indicate that the proposed approach is about 100 times faster than the traditional generalized cross-validation-based method.
Third, a new model based on the mixed use of time-differenced and undifferenced carrier phase observations is developed for kinematic multi-GNSS precise point positioning (PPP). The approach can reduce the number of constant parameters to be estimated and effectively mitigates systematic errors. Test results show that when an accurate initial position is available the proposed approach can attain up to 71% improvement in positioning accuracy compared to the traditional PPP. Four, a two-step positioning approach is proposed whereby a subset of observations considered to be of high quality is first selected to estimate an initial position. The derived initial position is then used to remove outliers and ambiguities in the rest of the observations. All the available observations can be processed together after the removal of outliers and ambiguities to strengthen and refine the positioning. Experimental results show that the new approach outperforms the traditional multiple GNSSs approach by 4.8% and 21.4% in standard point positioning and static relative positioning respectively. The two-step method has a higher efficiency compared with the traditional method, the computation time can be reduced by 40.4% and 27.7% respectively in standard point positioning and static relative positioning. The research results contribute to an improved understanding of the impacts of systematic errors in multi-GNSS positioning, and enhancing the accuracy, reliability, and efficiency in such positioning applications. The study can be extended to long-term and high-rate positioning applications, such as deformation monitoring and un-manned aerial vehicle navigation.
|Description:||xii, 108 pages : color illustrations
PolyU Library Call No.: [THS] LG51 .H577P LSGI 2019 Yu
|URI:||http://hdl.handle.net/10397/81460||Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|991022270059803411_link.htm||For PolyU Users||168 B||HTML||View/Open|
|991022270059803411_pira.pdf||For All Users (Non-printable)||5.23 MB||Adobe PDF||View/Open|
Citations as of Nov 20, 2019
Citations as of Nov 20, 2019
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.