Συντάχθηκε 27-07-2021 13:19
Τόπος: Η παρουσίαση θα γίνει με τηλεδιάσκεψη
Σύνδεσμος τηλεδιάσκεψης
Έναρξη: 29/07/2021 16:00
Λήξη: 29/07/2021 17:00
ΠΟΛΥΤΕΧΝΕΙΟ ΚΡΗΤΗΣ
Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών
Πρόγραμμα Μεταπτυχιακών Σπουδών
ΠΑΡΟΥΣΙΑΣΗ ΜΕΤΑΠΤΥΧΙΑΚΗΣ ΕΡΓΑΣΙΑΣ
ΚΩΝΣΤΑΝΤΙΝΟΣ ΣΚΥΒΑΛΑΚΗΣ
θέμα
Χαμηλού Κόστους Ραδιόφωνα Ελεγχόμενα από Λογισμικό & Τεχνικές για Ρομποτικά RFID Συστήματα
Low-Cost Software-Defined Radios (SDR) & Techniques for Robotic RFID Systems
Εξεταστική Επιτροπή
Καθηγητής Άγγελος Μπλέτσας (επιβλέπων)
Καθηγητής Γεώργιος Καρυστινός
Καθηγητής Μιχαήλ Πατεράκης
Abstract
The continuously increasing deployment of RFID tags has sparked strong academic and industrial interest on improved RFID tag throughput and localization. This work proposes a collision resolution method to boost the inventorying capacity of RFID readers, as well as a novel phase-based method for multistatic 2D/3D localization. The first part of this thesis proposes a Viterbi joint sequence detector that can resolve a collision between two tags in the physical layer. In sharp contrast to prior art, the proposed closed-form signal model takes into account the asynchrony level between the two collided tag responses, which is not uncommon with commercial, ultra-low-cost RFID tags that follow industry's Gen2 protocol. Asynchrony is considered as the time offset τ between the beginnings of the two tags' responses and is modeled through a derived shaping matrix that depends on the delayed tag information. Performance evaluation of the proposed detectors with simulated data under Ricean fading, as well as experimental data with software-defined radio (SDR), reveals improved performance compared to prior art, under various operating regimes. It is also shown that for different values of the parameter τ, BER does not present a monotonic behavior. As a collateral dividend, it is found that clustering techniques on the filtered received signal should explicitly take into account the time offset τ, since the latter modifies the number of observed clusters. The second part of this thesis offers elliptical direction-of-arrival (DoA) estimation and 2D/3D localization techniques, using phase-based, narrowband measurements. The method exploits a multistatic architecture, where illuminating and receiving antennas are placed in the same line. Ambiguities inherent in phase measurements are analytically addressed. Experimental results with Gen2 UHF RFID tags show similar performance in terms of DoA estimation compared to the MUSIC algorithm. In terms of 2D localization accuracy, the proposed method outperformed state-of-the-art algorithms in all experimental cases at the lab, offering mean absolute localization error as small as 9 cm, at the expense of additional effort for calibration. As a collateral dividend, proof-of-concept is offered via simulation that the multistatic 2D localization method, could possibly be applied in real-time motion tracking of a mobile robotic platform equipped with a Gen2 UHF RFID tag. Extension of the localization method for the 3D case is also offered, with promising simulation results.
Meeting ID: 969 7887 3417
Password: 564407
