Asynchronous inference for security applications
Asynchronous inference for security applications
Ασύγχρονος συμπερασμός για εφαρμογές ασφάλειας
Διπλωματική Εργασία
Diploma Work
2022-10-182022enThis work studies distributed inference for privacy-non intrusive security applications. Specifically, minimum mean square estimation (MMSE) of channel parameters is proposed, with asynchronous and distributed implementation of Gaussian Belief Propagation (GBP). Wireless channel is measured by distributed wireless terminals/sensors, which exchange messages to infer the wireless channel through asynchronous GBP. Variations of the channel between consecutive measurements indicate the presence (or absence)
of a moving person, without other means (e.g., cameras, microphones or infrared sensors). The method is tested in simulations and compared to synchronous GBP in terms of convergence; in addition, the method is contrasted against centralized least-squares (LS) and centralized MMSE, at the WiFi carrier frequency of 2.4 GHz, at various signal-to-noise ratios. It is found that mobility changes channel estimate metrics by at least one order of magnitude, compared to the static (immobile) case. The method could be exploited in indoor environments, where distributed presence of embedded wireless radios is widespread, without reverting to privacy intrusive microphones or cameras. In addition, the method is truly asynchronous and distributed, and hence more robust compared to synchronous counterparts.
http://creativecommons.org/licenses/by/4.0/Πολυτεχνείο Κρήτης::Σχολή Ηλεκτρολόγων Μηχανικών και Μηχανικών ΥπολογιστώνSifakis_Zacharias-Iosif_Dip_2022.pdfChania [Greece]Library of TUC2022-10-17application/pdf818.8 kBembargo
Sifakis Zacharias-Iosif
Σηφακης Ζαχαριας-Ιωσηφ
Bletsas Aggelos
Μπλετσας Αγγελος
Lagoudakis Michail
Λαγουδακης Μιχαηλ
Karystinos Georgios
Καρυστινος Γεωργιος
Πολυτεχνείο Κρήτης
Technical University of Crete
Asynchrony
MMSE
LS
Channel estimation
Distributed
GBP
Inference
Asynchronous inference