Caputo, Francesco (2025) Smart and Secure Low-power transducer networks. [Tesi di dottorato]

Documento PDF Caputo_PhDThesis_ITEE37.pdf Visibile a [TBR] Amministratori dell'archivio Download (11MB) | Richiedi una copia

Abstract

This thesis focuses on the energy efficiency estimation and security analysis of Internet of Things (IoT) devices. In terms of energy efficiency, a public framework for evaluating the performance of embedded devices while running machine learning models has been explored. An improvement of the framework has been proposed to achieve more accurate and reproducible energy performance measurements. As security is concerned, the focus is on side-channel attacks (SCAs) based on power measurements. In particular, two main aspects are addressed: (i) the improvement of the efficiency of power attacks and (ii) the optimization of Vulnerability Assessment tests. The efficiency improvement of power attacks aims to reduce the effort and time needed to reveal the secret key of a cryptographic algorithm, while the optimization of Vulnerability Assessment tests aims to better define the robustness of a crypto device. Two different methods are explored to improve the efficiency of attacks. The first method is based on a new optimization technique using a fractional experimental design, able to identify the optimal parameters to maximize the number of discovered key bytes while minimizing the attack cost in terms of time and resources. The second method uses machine learning models for "profiled" attacks, and also introduces an approach to estimate uncertainty in order to evaluate the reliability of the model in executing the attack.

Downloads

Actions (login required)

Modifica documento