The spreading use of Bluetooth Low Energy (BLE) devices and the limitations of techniques based on logical identifiers, which are easily spoofed or randomized, highlight the need for alternative approaches to reliable device identification. This thesis proposes a fingerprinting technique based on the analysis of Carrier Frequency Offset (CFO), an intrinsic parameter of radio hardware that can be detected in a completely passive manner. The main objective is to assess whether the variation in CFO as a function of ambient temperature can be modeled in a stable and distinctive way for each device, so as to constitute a unique thermal signature useful for identification purposes. To this end, an acquisition system based on Software Defined Radio (SDR) was developed, and an experimental study was conducted in a controlled environment on eight BLE devices. For each device, the CFO-temperature curve was estimated and modeled using linear regression. The curves obtained were used in various classification strategies, based on both single observations and multi-point sequences. The experimental results show that the proposed model allows nominally identical devices to be distinguished with accuracies of up to 70.9%, even under limited observation conditions. These results confirm the validity of the CFO–temperature relationship as a robust and persistent physical fingerprint, opening the door to applications in areas such as passive authentication and, in particular, forensic analysis. In this context, the ability to associate a physical device with a presence in space and time, even in the absence of reliable logical identifiers, can offer significant technical support for investigative activities and security scenarios.
La crescente diffusione dei dispositivi Bluetooth Low Energy (BLE) e le limitazioni delle tecniche basate su identificatori logici, facilmente soggetti a spoofing o randomizzazione, evidenziano la necessità di approcci alternativi per l'identificazione affidabile dei dispositivi. Questa tesi propone una tecnica di fingerprinting basata sull’analisi del Carrier Frequency Offset (CFO), un parametro intrinseco dell’hardware radio, rilevabile in modo completamente passivo. L’obiettivo principale è valutare se la variazione del CFO in funzione della temperatura ambientale possa essere modellata in modo stabile e distintivo per ciascun dispositivo, così da costituire una firma termica univoca utile ai fini identificativi. A tal fine, è stato realizzato un sistema di acquisizione basato su Software Defined Radio (SDR), ed è stata condotta una campagna sperimentale in ambiente controllato su otto dispositivi BLE. Per ciascun dispositivo è stata stimata la curva CFO–temperatura, modellata mediante regressione lineare. Le curve così ottenute sono state impiegate in diverse strategie di classificazione, sia basate su osservazioni singole che su sequenze multi-punto. I risultati sperimentali mostrano che il modello proposto consente di distinguere dispositivi nominalmente identici con accuratezze fino al 70,9%, anche in condizioni di osservazione limitata. Questi risultati confermano la validità della relazione CFO–temperatura come impronta fisica robusta e persistente, aprendo la strada ad applicazioni in ambiti come l’autenticazione passiva e, in particolare, l’analisi forense. In questo contesto, la possibilità di associare un dispositivo fisico a una presenza nello spazio e nel tempo, anche in assenza di identificatori logici attendibili, può offrire un supporto tecnico rilevante per attività investigative e scenari di sicurezza.
Leveraging environmental context for accurate BLE devices fingerprinting
Magnelli, Camilla
2024/2025
Abstract
The spreading use of Bluetooth Low Energy (BLE) devices and the limitations of techniques based on logical identifiers, which are easily spoofed or randomized, highlight the need for alternative approaches to reliable device identification. This thesis proposes a fingerprinting technique based on the analysis of Carrier Frequency Offset (CFO), an intrinsic parameter of radio hardware that can be detected in a completely passive manner. The main objective is to assess whether the variation in CFO as a function of ambient temperature can be modeled in a stable and distinctive way for each device, so as to constitute a unique thermal signature useful for identification purposes. To this end, an acquisition system based on Software Defined Radio (SDR) was developed, and an experimental study was conducted in a controlled environment on eight BLE devices. For each device, the CFO-temperature curve was estimated and modeled using linear regression. The curves obtained were used in various classification strategies, based on both single observations and multi-point sequences. The experimental results show that the proposed model allows nominally identical devices to be distinguished with accuracies of up to 70.9%, even under limited observation conditions. These results confirm the validity of the CFO–temperature relationship as a robust and persistent physical fingerprint, opening the door to applications in areas such as passive authentication and, in particular, forensic analysis. In this context, the ability to associate a physical device with a presence in space and time, even in the absence of reliable logical identifiers, can offer significant technical support for investigative activities and security scenarios.File | Dimensione | Formato | |
---|---|---|---|
2025_07_Magnelli_Tesi_01.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Tesi
Dimensione
1.81 MB
Formato
Adobe PDF
|
1.81 MB | Adobe PDF | Visualizza/Apri |
2025_07_Magnelli_Executive_Summary_02.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Executive Summary
Dimensione
494.08 kB
Formato
Adobe PDF
|
494.08 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/240593