The thesis analyzes the transition from cloud to edge in the context of the Internet of Things (IoT), linking architectural choices to quality-of-service requirements and security implications. After defining IoT, its growth, and the main application classes with their constraints on latency, reliability, and privacy, it outlines the technological and systemic background underpinning the scenarios considered. The work presents and compares the cloud and edge paradigms with respect to latency, bandwidth usage, privacy, energy efficiency, and context awareness, proposing a three-layer model (devices–edge–cloud) as a reference for task partitioning. On the security front, it defines an attacker model and proposes an end-to-end classification of threats across the stack: in IoT/Low-power and Lossy Networks (LLNs) there are passive attacks (eavesdropping, traffic analysis) and active ones (jamming, collisions, HELLO flooding, selective forwarding/grayhole/blackhole, sinkhole, wormhole, Sybil, node replication, routing loops, and misdirection). In the edge infrastructure, prominent threats include Distributed Denial of Service (DDoS), malware injection, side-channel attacks, and authentication/authorization weaknesses. Building on this evidence, the thesis sets out mitigation guidelines compatible with low-power platforms: encryption and key management, authentication and access control, network filtering and anomaly detection, and hardware/operating-system isolation. The goal is to preserve performance and quality of service while limiting exposure and attack surface. The main contribution is a perspective that links cloud/edge partitioning decisions to the most likely attack surfaces and to practical controls in real-world scenarios, offering criteria for designing more secure and resilient decentralized IoT architectures.
La tesi analizza la transizione dal cloud all’edge nel contesto dell’Internet of Things (IoT), met- tendo in relazione le scelte architetturali con i requisiti di qualità del servizio e le implicazioni di sicurezza. Dopo aver definito l’IoT, la sua crescita e le principali classi applicative con i relativi vincoli di latenza, affidabilità e privacy, viene delineato il quadro tecnologico e sistemico alla base degli scenari considerati. Il lavoro presenta e confronta i paradigmi cloud ed edge rispetto a latenza, uso di banda, privacy, efficienza energetica e consapevolezza di contesto, proponendo un modello a tre strati (dispositivi–edge–cloud) come riferimento per la ripartizione dei compiti. Sul piano della sicurezza viene definito un modello di attaccante e viene proposta una classifi- cazione end-to-end delle minacce lungo lo stack: nelle reti IoT/Low-power and Lossy Network (LLN) compaiono attacchi passivi (eavesdropping, analisi del traffico) e attivi (jamming, col- lision, HELLO-flooding, selective forwarding/grayhole/blackhole, sinkhole, wormhole, Sybil, node replication, routing loop e misdirection). Nell’infrastruttura edge spiccano Distributed Denial of Service (DDoS), malware injection, side-channel e debolezze di autenticazione/au- torizzazione. A partire da tali evidenze, la tesi espone linee guida di mitigazione compatibili con piattaforme a bassa potenza: cifratura e gestione delle chiavi, autenticazione e controllo degli accessi, filtraggio e rilevamento di anomalie in rete, isolamento hardware e del sistema operativo. L’obiettivo è preservare prestazioni e qualità del servizio limitando l’esposizione e la superficie d’attacco. Il contributo principale consiste in una lettura che collega le decisioni di ripartizione cloud/edge alle superfici d’attacco più probabili e ai controlli praticabili in scenari reali, offrendo criteri per progettare architetture IoT decentralizzate più sicure e resilienti.
Dal Cloud all’Edge: Vulnerabilità, Minacce e Soluzioni di Sicurezza per un’Architettura IoT Decentralizzata
PESCE, MATTIA
2024/2025
Abstract
The thesis analyzes the transition from cloud to edge in the context of the Internet of Things (IoT), linking architectural choices to quality-of-service requirements and security implications. After defining IoT, its growth, and the main application classes with their constraints on latency, reliability, and privacy, it outlines the technological and systemic background underpinning the scenarios considered. The work presents and compares the cloud and edge paradigms with respect to latency, bandwidth usage, privacy, energy efficiency, and context awareness, proposing a three-layer model (devices–edge–cloud) as a reference for task partitioning. On the security front, it defines an attacker model and proposes an end-to-end classification of threats across the stack: in IoT/Low-power and Lossy Networks (LLNs) there are passive attacks (eavesdropping, traffic analysis) and active ones (jamming, collisions, HELLO flooding, selective forwarding/grayhole/blackhole, sinkhole, wormhole, Sybil, node replication, routing loops, and misdirection). In the edge infrastructure, prominent threats include Distributed Denial of Service (DDoS), malware injection, side-channel attacks, and authentication/authorization weaknesses. Building on this evidence, the thesis sets out mitigation guidelines compatible with low-power platforms: encryption and key management, authentication and access control, network filtering and anomaly detection, and hardware/operating-system isolation. The goal is to preserve performance and quality of service while limiting exposure and attack surface. The main contribution is a perspective that links cloud/edge partitioning decisions to the most likely attack surfaces and to practical controls in real-world scenarios, offering criteria for designing more secure and resilient decentralized IoT architectures.| File | Dimensione | Formato | |
|---|---|---|---|
|
Pesce_Mattia.pdf
Accesso riservato
Dimensione
1.15 MB
Formato
Adobe PDF
|
1.15 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/91740