| Summary: | Global Navigation Satellite Systems (GNSS) are the main systems that provide positioning, navigation and timing at a global level. They are being used in numerous applications in different sectors including transport, military, oil & gas, agriculture as well as location based services. A significant number of these applications require centimetre-level positioning accuracy, a challenging feat due to the many error sources that affect GNSS measurements. These include errors at the satellite, propagation medium, and receiver levels. Most of these errors can be mitigated by modeling, or by exploiting their spatial and temporal correlation characteristics. However, multipath errors, which result from the combination of the direct signal with reflected signals in the vicinity of the receiver antenna, are difficult to model and therefore, difficult to mitigate. Furthermore, high accuracy positioning applications typically rely on linear combinations of measurements at different frequencies (e.g. L1 and L2 in the case of the Global Positioning System) to mitigate frequency-dependent errors such as ionospheric errors (i.e. ionosphere free combination) or otherwise facilitate position calculation (e.g. Wide Lane observable). The multipath errors associated with such combinations are significantly larger than those of individual signals. The dependency of the multipath error on the environment and its low level in single frequency measurements (i.e. up to quarter of wavelength) makes modelling and mitigating it very difficult. Current techniques attempt to mitigate multipath errors for measurements at each individual frequency, independently of the error at other frequencies, even when linear combinations of measurements are used. The literature review carried out in this thesis has drawn three main conclusions regarding carrier multipath mitigation. Firstly, existing carrier multipath mitigation techniques are inaccurate, impractical or not effective. Secondly most of the practical techniques attempt to mitigate the error by de-weighting the measurements which are most prone to the multipath error (i.e measurement at low elevation). Thirdly, existing weighting techniques are oversimplified and do not reflect the error level accurately. In this research and for the first time, carrier multipath errors have been studied directly at the linear combination level. This is by exploiting the dispersive nature of multipath errors in order to model and correct them. New carrier multipath mitigation techniques applicable to linear combinations of measurements have been developed in this thesis on the basis of a new error model and a new observable referred to as the IFM (Inter-Frequency carrier Multipath). The IFM is computed from carrier phase measurements at two different frequencies, and corresponds to the combined multipath errors of those signals. In addition to multipath mitigation, this observable has various other applications. The well-defined relationship between the IFM and carrier multipath errors is used in this thesis to develop multipath mitigation techniques based on two approaches: multipath correction and measurement weighting. The new mitigation techniques are applicable to linear combinations of observations such as Wide Lane (WL) and Ionosphere Free (IF) carrier phase measurements in double differenced mode. The new multipath mitigation techniques have been validated using real data and the results compared with those obtained using the elevation weighting technique. The results show that the new methods developed in this thesis improve the mean error of horizontal position by up to 33% when using the IF combination. The results also show improvements of up to 78% in the time it takes to resolve ambiguities when using the WL combination.
|
|---|
