Mitigation of Multipath-induced Errors in Satellite Navigation
Author
Gulie, Ioana
Term
4. term
Education
Publication year
2014
Submitted on
2014-06-11
Pages
81
Abstract
Globale satellitnavigationssystemer (GNSS, f.eks. GPS) beregner en brugers position ved at måle, hvornår radiosignaler fra satellitter når modtageren. I byområder kan signalerne reflekteres fra bygninger og jordoverflader, hvilket skaber flervejsudbredelse, der kan forringe positionsnøjagtigheden. Denne afhandling undersøger, om nøjagtigheden kan forbedres ved fælles estimering af ankomsttiden for både det direkte og det reflekterede signal, frem for kun at bruge de sporingssløjfer, som findes i konventionelle modtagere. En maksimum-likelihood-estimator (en statistisk metode, der søger den mest sandsynlige løsning) undersøges og implementeres. Da denne metode er beregningstung, anvendes signalkomprimering for at reducere datamængde og beregningskrav, og der foreslås og udvikles forskellige begrænsninger (constraints) for at styre estimeringen. Til evaluering udvikles en simuleringsplatform, der genererer direkte og reflekterede signaler og muliggør implementering af maksimum-likelihood-algoritmen. Estimatorens ydeevne i multipath-miljøer sammenlignes med klassiske sporingssløjfer.
Global Navigation Satellite Systems (GNSS, e.g., GPS) determine a user’s position by timing when radio signals from satellites reach the receiver. In built-up areas, signals can bounce off buildings or the ground, creating multipath reflections that delay and distort the signal and reduce accuracy. This thesis examines whether position accuracy can be improved by jointly estimating the arrival time of both the direct and the reflected signal, rather than relying only on the tracking loops used in conventional receivers. A maximum likelihood estimator (a statistical method that seeks the most probable solution) is investigated and implemented. Because this approach is computationally intensive, signal compression is used to reduce data size and processing load, and different constraints are proposed and developed to guide the estimation. For evaluation, a simulation platform is built to generate direct and reflected signals and to implement the maximum likelihood algorithm. The estimator’s performance in multipath environments is compared with classical tracking loops.
[This abstract was generated with the help of AI]
Documents