TY - BOOK

T1 - On integrity prediction for network-RTK positioning in urban environments

AU - Karimidoona, Ali

PY - 2024

Y1 - 2024

N2 - Autonomous transportation systems require navigation performance with a high level of integrity. GNSS real time kinematic (RTK) solution provides absolute cm level accuracy, which is needed to ensure lane level accuracy. These solutions should be trustworthy, which necessitates integrity monitoring. Integrity is usually introduced by the parameters position error (PE), protection level (PL) and alert limit (AL). PE is calculated from the positioning solution, for defining the PL, the integrity risk or the required probability that the integrity can be violated. The AL depends on the application, e.g., for local streets, it can be defined around 30 cm for longitudinal and lateral components. The main purpose of this research is evaluating the possibility of predicting the integrity of Network-RTK positioning in urban environments. For predicting the integrity, every compo- nent of the positioning algorithm should be simulated. The observations, the design matrix and the C/N0 for weighting, should be predicted, to simulate the whole positioning algorithm. Using ray tracing algorithms, along with digital 3D city models, it is possible to predict the satellite visibility status for a specific position on the ground. Regarding the situation of satel- lite, receiver and blocking object, different conditions may exist. If the receiver is in direct line of sight of the satellite, it is LOS; if the signal arrives at the receiver antenna with reflection in addition to LOS, it is multipath (MP); if the signal comes to the receiving antenna only from a reflection, then it is called non-line of sight (NLOS). There is also another case where the signal can reach the receiver through diffraction. Furthermore, the signal can be attenuated passing through vegetation. For this research, a kinematic field experiment was conducted. Four different RTK-enabled GNSS receivers were installed on top of a van. Three of them are geodetic grade receivers from Septentrio, Leica and Trimble, and one of them is a low-cost receiver from u-blox, integrated with an antenna into one housing. The van was driven in a 1-km trajectory near the Leibniz University campus in Hannover. The trajectory consists of both open-sky parts and limited sky visibility. The main findings of this research can be outlined as follows. The satellite visibility can be predicted with an accuracy of more than 90 % in terms of confusion matrix. For predicting the C/N0, the attenuation of the signal from satellite to receiver should be considered. The power losses include free space loss, and satellite and receiver antenna gain. Moreover, losses due to reflection or diffraction or passing through vegetation, should be considered. When the C/N0 is predicted, it can be used as a very beneficial information added to the prediction process, while many functions depend on this parameter. The precision of the observation is determined using C/N0, it is the crucial parameter for the acquisition and tracking parts. The C/N0 can be predicted by a mean value of 2.21 dB-Hz difference to the real value in LOS situations. For this purpose, a positioning simulation is developed to emulate RTK positioning. As there is already a very precise reference trajectory available, a linearized Kalman filter is used. The observations consist of double difference (DD) residuals that can be assumed to be the station dependent errors in urban environments, i.e., those errors from MP, NLOS, and diffraction. The predicted C/N0 is used as a criterion to see if the code and phase observations can be present or not, and also for defining a weighting scheme. The positioning simulation provides PE and PL, which are compared to the real values. The true positive rate (TPR) or sensitivity for prediction of nominal operation cases as well as detection of fixed solutions can reach near 90 %.

AB - Autonomous transportation systems require navigation performance with a high level of integrity. GNSS real time kinematic (RTK) solution provides absolute cm level accuracy, which is needed to ensure lane level accuracy. These solutions should be trustworthy, which necessitates integrity monitoring. Integrity is usually introduced by the parameters position error (PE), protection level (PL) and alert limit (AL). PE is calculated from the positioning solution, for defining the PL, the integrity risk or the required probability that the integrity can be violated. The AL depends on the application, e.g., for local streets, it can be defined around 30 cm for longitudinal and lateral components. The main purpose of this research is evaluating the possibility of predicting the integrity of Network-RTK positioning in urban environments. For predicting the integrity, every compo- nent of the positioning algorithm should be simulated. The observations, the design matrix and the C/N0 for weighting, should be predicted, to simulate the whole positioning algorithm. Using ray tracing algorithms, along with digital 3D city models, it is possible to predict the satellite visibility status for a specific position on the ground. Regarding the situation of satel- lite, receiver and blocking object, different conditions may exist. If the receiver is in direct line of sight of the satellite, it is LOS; if the signal arrives at the receiver antenna with reflection in addition to LOS, it is multipath (MP); if the signal comes to the receiving antenna only from a reflection, then it is called non-line of sight (NLOS). There is also another case where the signal can reach the receiver through diffraction. Furthermore, the signal can be attenuated passing through vegetation. For this research, a kinematic field experiment was conducted. Four different RTK-enabled GNSS receivers were installed on top of a van. Three of them are geodetic grade receivers from Septentrio, Leica and Trimble, and one of them is a low-cost receiver from u-blox, integrated with an antenna into one housing. The van was driven in a 1-km trajectory near the Leibniz University campus in Hannover. The trajectory consists of both open-sky parts and limited sky visibility. The main findings of this research can be outlined as follows. The satellite visibility can be predicted with an accuracy of more than 90 % in terms of confusion matrix. For predicting the C/N0, the attenuation of the signal from satellite to receiver should be considered. The power losses include free space loss, and satellite and receiver antenna gain. Moreover, losses due to reflection or diffraction or passing through vegetation, should be considered. When the C/N0 is predicted, it can be used as a very beneficial information added to the prediction process, while many functions depend on this parameter. The precision of the observation is determined using C/N0, it is the crucial parameter for the acquisition and tracking parts. The C/N0 can be predicted by a mean value of 2.21 dB-Hz difference to the real value in LOS situations. For this purpose, a positioning simulation is developed to emulate RTK positioning. As there is already a very precise reference trajectory available, a linearized Kalman filter is used. The observations consist of double difference (DD) residuals that can be assumed to be the station dependent errors in urban environments, i.e., those errors from MP, NLOS, and diffraction. The predicted C/N0 is used as a criterion to see if the code and phase observations can be present or not, and also for defining a weighting scheme. The positioning simulation provides PE and PL, which are compared to the real values. The true positive rate (TPR) or sensitivity for prediction of nominal operation cases as well as detection of fixed solutions can reach near 90 %.

U2 - 10.15488/18268

DO - 10.15488/18268

M3 - Dissertation

T3 - Wissenschaftliche Arbeiten der Fachrichtung Geodäsie und Geoinformatik der Leibniz-Universität Hannover

ER -