Πλοήγηση ανά Συγγραφέας "Daoutis, Georgios"

Τώρα δείχνει 1 - 2 of 2
  • Μικρογραφία εικόνας
    Τεκμήριο
    Lightweight security mechanisms for μικρής κλίμακας ενσωματωμένες συσκευές.
    (ΕΛ.ΜΕ.ΠΑ., ΣΧΟΛΗ ΜΗΧΑΝΙΚΩΝ (ΣΜΗΧ), Τμήμα Ηλεκτρολόγων Μηχανικών και Μηχανικών Υπολογιστών, 2020-07-16) Δαούτης, Γεώργιος; Daoutis, Georgios
    Στόχος αυτής της πτυχιακής εργασίας είναι η ασφαλής ανάθεση και εκτέλεση επεξεργαστικών εργασιών σε έναν μικροεπεξεργαστή ενσωματωμένου συστήματος, μέσω τεχνικών αυθεντικοποίησης, κρυπτογράφησης και ταυτόχρονα δέσμευσης πόρων, ξεχωριστών, απομονωμένων και δεσμευμένων για αυτόν τον στόχο. Επιπλέον, η διαδικασία χρονοδρομολόγησης εργασιών έχει και αυτή με τη σειρά της απομονωθεί και έχει αφιερωθεί/δεσμευτεί ένας ανεξάρτητος μικροελεγκτής για την εκτέλεση αυτής και μόνον της εργασίας.
  • Μικρογραφία εικόνας
    Τεκμήριο
    Privacy preserving cloud computation for data collected from microcontrollers.
    (ΕΛ.ΜΕ.ΠΑ., Σχολή Μηχανικών (ΣΜΗΧ), ΠΜΣ Πληροφορική και Πολυμέσα, 2023-04-12) Daoutis, Georgios; Δαούτης, Γεώργιος
    The growth of the Internet of Things has given us the ability to monitor and control various situations from distance, mostly by collecting data from the edge, then storing and distributing them with the use of servers. This data can be the location of a person, medical data, industrial data, or other sensitive information, so we must protect them from unauthorized users. If the server where the data will be stored is semi-trusted, then we must protect take action to protect those data. As semi-honest, we referred to a party that adheres to the protocol correctly while also keeping a log of all its intermediate calculations. The proposed solution to this problem is the use of homomorphic encryption which give us the ability to perform operations on encrypted data such as additions and multiplications without the need of decrypting the data first. This makes homomorphic cryptography ideal for this use because the server can perform operations and at the same time, he cannot have access to the data. In this project, we will build the above scenario for data collected from microcontrollers. The microcontroller will have various data stored on its local memory and when the user wants to offload them and send them to the cloud for storage, the microcontroller will first encrypt those data with homomorphic encryption and then send them to the cloud. The microcontroller will connect with the outside world via NFC. When the user wants to extract some information from those data, he will ask the server to execute an algorithm on those data and then send back the result encrypted that only the user can decrypt with the Android application using his secret key. The query implemented is an encrypted inference based on a dataset that determines the presents of humans in a room. For that reason, we trained a perceptron and designed a that can run on the server without the server can access the private values that the board collected.