Positioning with supported GPS, A simulation study
Abstract: The satellite based on positioning system GPS has a very high potential in accuracy and general performance. It has also low cost, low weight and low power consumption.However, among all good proprieties there exist some bad ones. The most serious disadvantages of GPS are that the satellite signal is often disturbed by other electromagnetic signals and that it is shaded by physical obstacles. This diminishes accuracy and reliability, especially when GPS data is used in real time.To decrease this uncertainly, the GPS system should be supported by a complementary sensor. For this supported GPS system the following questions are raised:-how should this positioning system look like and what mathematical methods should be used to get most accurate estimate of the navigation parameters?-how should the achieved accuracy be estimated?-how could erroneous observations be sorted out?This report suggests several methods to solve these problems. Some of these are tested with for this purpose developed software called GPSlab. The main idea is that this is an optimization problem that should be common for all types of sensors to be integrated. A realization of the solution of these problems as software, is here called “A Navigation core”Suggestions presented in this report are:-an optimal estimate of parameters should be carried out on the row data level. This has the advantage to preserve the position information content,-with help of a supporting device it is possible to use several epochs to calculate a more accurate point solution,-if the ambiguities are solved to a number of satellites, also if the number is less than four, a new adjustment of the coordinates should BE Carried out. In this adjustment the observations with solved ambiguities are given a high weight and the pseud oranges are given a low.-a kalman filter could be used on the raw data level. It could detect deviations in the observations such as malfunctions, multipath, cycle slips and other significant errors,- a calibration of the supporting device should be performed when the ambiguities to observed satellites are solved.-it is proposed that Förstner’s method could be used to estimate the proper weights-a method is proposed to detect large errors-the accuracy of the solved parameters is estimated with the help of an optimal weighted point solution, with or without the solved ambiguity-a method to decrease the search volume of possible solutions for ambiguity is proposed,- a method to integrate sensors with different properties is suggested. The simulations show that with the used positioning method, it seems likely to solve the ambiguities, with the support of an Medium Accuracy Navigation System. It also seems possible to solve the ambiguities with the support of an odometer as the only supporting sensor. In that case, however, there have to be at least six satellites, good satellite geometry, small height variations and small errors in the odometer.Furthermore, the azimuth information of the GPS should be used and the observations should be weighted. This system must be initialized with at least five satellites when the receiver is not moving and with at least, probably 7-8 satellites that the receiver moves.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.