AbstractThe primary aim of this research is to develop and assess the innovative methods and techniques which are used to augment GPS using a variety of digital spatial data. It is well known that the use of GPS can be severely compromised by various error sources such as signal obstructions, multipath and poor satellite geometry etc., especially in highly built-up areas. In order to improve the accuracy and reliability of GPS, complementary data is often combined with GPS data for enhancing the performance of a standalone GPS receiver. Spatial data is one type of complementary data that can be used to augment GPS.
However, the potential of using various types of existing and newly acquired spatial data for enhancing GPS performance has not been fully realised. This is particularly true due to the fact that higher accuracy digital surface models (DSMs), which include buildings and vegetation, and digital maps, have only been made widely available in recent years.
This thesis will report on a number of experiments that used spatial data of various complexity and accuracy for enhancing GPS performance. These experiments include height aiding with different scale digital terrain models (DTMs); map-matching using odometer data, DTM and road centrelines; modelling and prediction of GPS satellite visibility using DSMs; and prediction of GPS multipath effect using DSMs and building footprints. These experiments are closely related to each other in the sense that GPS and spatial data are combined to provide value-added information for improved modelling and prediction of GPS positioning accuracy and reliability, for applications such as transport navigation and tracking...
Extensive fieldwork has been carried out to verify the developed techniques and methods. The results show that the accuracy of a standalone GPS receiver can be improved by height aiding using a higher resolution DTM and map-matching especially when the satellite geometry is poor. The mean error of single receiver GPS positioning for one particular dataset, on which the described map-matching algorithm was developed, is 8.8m compared with 53.7m for GPS alone. This work was carried out in collaboration with London Transport.
In terms of satellite visibility analysis, the results obtained from the fieldwork indicate that greater modelling accuracy has been achieved when using higher resolution DSMs. Furthermore, a ray tracing model was implemented in a 3D GIS environment in order to model reflected and diffracted GPS signals. The Double Differencing (DD) residuals were used to give an indication of the magnitude of the possible pseudorange multipath error caused by diffraction. A single-knife diffraction model was first implemented on 1m Light Detection And Ranging (LiDAR) DSMs, and verified by post-processing (i.e. large DD residuals occurred when the satellites are partially masked and unmasked by buildings), which indicate that GPS multipath prediction with LiDAR data and building footprints is feasible, and has the potential to offer greater modelling accuracy.
|Date of Award
|David Kidner (Supervisor)
- Global Positioning System
- spatial data