Nowadays, positioning is important in every people’s activity and is considered necessary for all their actions. Moreover, positioning has an important role in research and control, for the environment, natural phenomena, human missions, and Earth Observation. Until recently, positioning services were only offered by the Global Navigation Satellite System (GNSS), which includes all satellite constellations, whose signals are used to estimate positions everywhere and at all times. So, GNSS has been the prevalent positioning, navigation and timing technology over the past few decades. Even though there has been consistent progress and improvement in GNSS techniques and accuracies, there are still some challenges that remain open. For example, densely built or inhabited urban environments represent obstructions to the signal, which can cause lack of observations or the multipath effect. That’s the reason why in the last years the idea of considering a hybrid configuration of GNSS and 5G services in positioning has developed. 5G could provide an important tool to improve urban positioning in scenarios where the main technique, GNSS, has problems and difficulties and can therefore compensate for its limitations and enhance its localization accuracy. My thesis objective is to explore the potential of 5G positioning together with Kalman Filter by exploiting the data collected during the GINTO5G experiment funded by the European Space Agency (ESA). In particular, the project consisted of collecting field measurements with 5G-like positioning signals to evaluate its potential. ESA collaborated with some other European companies for the experiment execution, and the test area able to accommodate all the equipment needed for the measurement campaign was selected in Germany. The field campaign consisted of estimating the coordinates of a moving trolley in different trajectories: some were indoor, some others outdoor and some moved from outdoor to indoor areas. The coordinates were estimated using two different positioning methods: Total Station, then used as a support ground truth, and 5G. In my thesis, 5G Time of Arrivals have been processed and analysed by comparing them with the ground truth. For the processing, two main methods have been implemented: the Least Squares and the Kalman Filter. In fact, the main objective of my work is to implement an algorithm capable of processing 5G observations, to compare the results given by these two different approaches, to assess the accuracy of 5G positioning, and to see if filtering the noise of the trajectories through Kalman Filter could improve the quality of the result significantly or not. A Python library has been implemented to perform 5G positioning using as input Time of Arrivals (ToA) measured on 5G signals. The algorithm implements Least Squares in single epoch and Kalman Filter on ToA, studying also the behaviour of the user unit clock offset. Also, some outlier rejection methods have been studied and implemented to treat the data in the best possible way, adapting to the environmental and geometrical features of the trajectories. All the results obtained have been analysed by empirical and statistical approaches, by specifically implemented libraries.
Oggigiorno, il posizionamento è fondamentale per ogni attività umana ed è considerato necessario per le azioni di chiunque. Inoltre, il posizionamento ha un ruolo importante nella ricerca e nel monitoraggio ambientale, dei fenomeni naturali, nelle missioni umane e nell'Osservazione Terrestre. Fino a poco tempo fa, i servizi di posizionamento venivano offerti soltanto dal Sistema di Navigazione Terrestre (GNSS), che comprende tutte le costellazioni satellitari, i cui segnali sono usati per stimare posizioni dovunque e in qualunque momento. Dunque, GNSS ha rappresentato la tecnologia di posizionamento, navigazione e temporizzazione prevalente nelle ultime decadi. Nonostante però ci sia stato un consistente progresso e miglioramento nelle tecniche GNSS e nella sua accuratezza, alcune sfide rimangono tuttora aperte. Ad esempio, ambienti urbani densamente costruiti o abitati rappresentano ostacoli per il segnale e possono causare una mancanza di osservazioni disponibili o l'effetto del multi-path. Questa è la ragione per cui negli ultimi anni si è sviluppata l'idea di considerare una configurazione ibrida di servizi GNSS e 5G nel posizionamento. Il 5G potrebbe fornire strumenti importanti nel miglioramento del posizionamento urbano in scenari in cui la principale tecnologia, il GNSS, presenta dei problemi e difficioltà, e potrebbe quindi compensare alle sue limitazioni ed aumentare l'accuratezza nella localizzazione. L'obiettivo della mia tesi è quello di esplorare il potenziale del posizionamento con il 5G insieme al filtre di Kalman, sfruttando i dati raccolti durante l'esperimento GINTO5G fondato dall'Agenzia Spaziale Eurepea (ESA). In particolare, il progetto consisteva nel raccogliere dati sul campo con segnali di posizionamento 5G per valutarne il potenziale. ESA ha collaborato con altre aziende Europee per l'esecuzione dell'esperimento, e l'area di test, capace di ospitare tutti gli strumenti necessari per la campagna di misurazione, è stata selezionata in Germania. La campagna su campo consisteva in stimare le coordinate di un trolley in movimento su diverse traiettorie: alcune indoor, altre outdoor, e altre ancora passanti da aree outdoor ad aree indoor. Le coordinate sono state estimate utilizzando due metodi di posizionamento diversi: la Total Station, usata poi come ground truth di supporto, e il 5G. Nella mia tesi, i 5G Time of Arrivals sono stati processati e analizzati comparandoli con la Total Station. Per il processamento, due metodi sono stati implementati: i Minimi Quadrati e il Filtro di Kalman. Infatti, l'obiettivo principale del mio lavoro è quello di implementare un algoritmo capace di processare osservazioni 5G, di comparare i risultati ottenuti da questi due metodi diversi, di valutare la precisione del posizionamento 5G e di vedere se filtrare il rumore delle traiettorie con il filtro di Kalman potrebbe migliorare significativamente o meno la qualità dei risultati. Una libreria Python è stata implementata per eseguire il posizionamento 5G utilizzando come input i Time of Arrivals (ToA) misurati con segnali 5G. L'algoritmo implementa i Minimi Quadrati in singola epoca e il filtro di Kalman sui ToA, e inoltre studia il comportamento del clock offset dello user. Inoltre, sono stati studiati ed implementati anche dei metodi di reiezione degli outliers, per trattare i dati nella maniera migliore possibile, adattandosi alle caratteristiche ambientali e geometriche delle traiettorie. Tutti i risultati ottenuti sono stati analizzati con approcci statistici ed empirici, tramite librerie implementate specificatamente a questo scopo.
Kalman filtering of undifferenced 5G time of arrivals: a first experimental analysis
ESPOSITO, MARTINA GIOVANNA
2022/2023
Abstract
Nowadays, positioning is important in every people’s activity and is considered necessary for all their actions. Moreover, positioning has an important role in research and control, for the environment, natural phenomena, human missions, and Earth Observation. Until recently, positioning services were only offered by the Global Navigation Satellite System (GNSS), which includes all satellite constellations, whose signals are used to estimate positions everywhere and at all times. So, GNSS has been the prevalent positioning, navigation and timing technology over the past few decades. Even though there has been consistent progress and improvement in GNSS techniques and accuracies, there are still some challenges that remain open. For example, densely built or inhabited urban environments represent obstructions to the signal, which can cause lack of observations or the multipath effect. That’s the reason why in the last years the idea of considering a hybrid configuration of GNSS and 5G services in positioning has developed. 5G could provide an important tool to improve urban positioning in scenarios where the main technique, GNSS, has problems and difficulties and can therefore compensate for its limitations and enhance its localization accuracy. My thesis objective is to explore the potential of 5G positioning together with Kalman Filter by exploiting the data collected during the GINTO5G experiment funded by the European Space Agency (ESA). In particular, the project consisted of collecting field measurements with 5G-like positioning signals to evaluate its potential. ESA collaborated with some other European companies for the experiment execution, and the test area able to accommodate all the equipment needed for the measurement campaign was selected in Germany. The field campaign consisted of estimating the coordinates of a moving trolley in different trajectories: some were indoor, some others outdoor and some moved from outdoor to indoor areas. The coordinates were estimated using two different positioning methods: Total Station, then used as a support ground truth, and 5G. In my thesis, 5G Time of Arrivals have been processed and analysed by comparing them with the ground truth. For the processing, two main methods have been implemented: the Least Squares and the Kalman Filter. In fact, the main objective of my work is to implement an algorithm capable of processing 5G observations, to compare the results given by these two different approaches, to assess the accuracy of 5G positioning, and to see if filtering the noise of the trajectories through Kalman Filter could improve the quality of the result significantly or not. A Python library has been implemented to perform 5G positioning using as input Time of Arrivals (ToA) measured on 5G signals. The algorithm implements Least Squares in single epoch and Kalman Filter on ToA, studying also the behaviour of the user unit clock offset. Also, some outlier rejection methods have been studied and implemented to treat the data in the best possible way, adapting to the environmental and geometrical features of the trajectories. All the results obtained have been analysed by empirical and statistical approaches, by specifically implemented libraries.| File | Dimensione | Formato | |
|---|---|---|---|
|
2023_12_Esposito.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: testo tesi
Dimensione
7.93 MB
Formato
Adobe PDF
|
7.93 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/214573