Due to the increasing demands of computational power, energy efficiency, and thermal constraints, multi-core microprocessors are being increasingly adopted in embedded real- time systems. Compared with their single-core counterparts, these solutions are typically more cost-effective and provide better performance at lower clock speeds by handling multiple tasks in parallel. However, the increased transistor density that made it possible to integrate multiple cores on a single die also increased the system susceptibility to transient faults. This issue is further accentuated by the harsh physical environments in which embedded safety-critical systems often operate, where they can be affected by radiation, temperature fluctuations, and electromagnetic interference. In the literature, software solutions known as Software-Implemented Hardware Fault Tolerance (SIHFT) have been proposed to detect hardware transient faults. However, most of these techniques do not exploit the additional resources of multi-core architectures, leading to sub-optimal performance. It is therefore necessary to explore new techniques that are more suitable for these types of systems. This work shows how an existing state-of-the-art SIHFT technique can be adapted to perform in parallel with the protected program. It also introduces a novel overhead reduction technique. Furthermore, a Real-Time Operating System (RTOS) is compiled with this scheme, hardening both real-time tasks and kernel features. Finally, the technique was validated through testing on a real board powered by an RP2040 microcontroller, demonstrating a remarkably high Control-Flow Error (CFE) detection rate, with improved performance when coupled with the overhead reduction technique.
A causa delle crescenti richieste di potenza computazionale, efficienza energetica e vin- coli termici, i microprocessori multi-core vengono adottati sempre più frequentemente nei sistemi embedded real-time. Rispetto alle loro controparti single-core, queste soluzioni sono tipicamente più convenienti dal punto di vista economico e forniscono prestazioni migliori a frequenze di clock inferiori, gestendo più task in parallelo. Tuttavia, l’aumento di densità dei transistor che ha reso possibile integrare più core su un singolo die ha anche incrementato la suscettibilità del sistema ai guasti transitori. Questo problema è ulteriormente accentuato dagli ambienti fisici in cui operano spesso i sistemi embedded safety-critical, dove possono essere influenzati da radiazioni, fluttuazioni di temperatura e interferenze elettromagnetiche. In letteratura, per rilevare i guasti hardware transitori sono state proposte soluzioni software note come SIHFT. Tuttavia, la maggior parte di queste tecniche non sfrutta le risorse aggiuntive delle architetture multi-core, portando a prestazioni non ottimali. È quindi necessario esplorare nuove soluzioni più adatte a questi tipi di sistema. Questo lavoro mostra come una tecnica SIHFT possa essere adattata per funzionare in parallelo con il programma protetto, proponendo inoltre una nuova tecnica di riduzione dell’overhead. La tesi tratta anche di come un sistema operativo real-time possa essere compilato con questo schema, proteggendo sia i task real-time che le fun- zionalità del kernel. Infine, la tecnica sviluppata è stata validata attraverso test su una vera board basata su un microcontrollore RP2040, dimostrando una notevole capacità di rilevamento di errori nel control flow e prestazioni migliorate quando accoppiata con la tecnica di riduzione dell’overhead.
Enabling inter-core control flow checking in embedded real-time systems
Rigoli, Fabio
2024/2025
Abstract
Due to the increasing demands of computational power, energy efficiency, and thermal constraints, multi-core microprocessors are being increasingly adopted in embedded real- time systems. Compared with their single-core counterparts, these solutions are typically more cost-effective and provide better performance at lower clock speeds by handling multiple tasks in parallel. However, the increased transistor density that made it possible to integrate multiple cores on a single die also increased the system susceptibility to transient faults. This issue is further accentuated by the harsh physical environments in which embedded safety-critical systems often operate, where they can be affected by radiation, temperature fluctuations, and electromagnetic interference. In the literature, software solutions known as Software-Implemented Hardware Fault Tolerance (SIHFT) have been proposed to detect hardware transient faults. However, most of these techniques do not exploit the additional resources of multi-core architectures, leading to sub-optimal performance. It is therefore necessary to explore new techniques that are more suitable for these types of systems. This work shows how an existing state-of-the-art SIHFT technique can be adapted to perform in parallel with the protected program. It also introduces a novel overhead reduction technique. Furthermore, a Real-Time Operating System (RTOS) is compiled with this scheme, hardening both real-time tasks and kernel features. Finally, the technique was validated through testing on a real board powered by an RP2040 microcontroller, demonstrating a remarkably high Control-Flow Error (CFE) detection rate, with improved performance when coupled with the overhead reduction technique.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_07_Rigoli_Tesi.pdf
solo utenti autorizzati a partire dal 01/07/2026
Descrizione: Tesi
Dimensione
1 MB
Formato
Adobe PDF
|
1 MB | Adobe PDF | Visualizza/Apri |
|
2025_07_Rigoli_Executive_Summary.pdf
solo utenti autorizzati a partire dal 01/07/2026
Descrizione: Executive Summary
Dimensione
394.22 kB
Formato
Adobe PDF
|
394.22 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/10589/240640