In todays world of technology use physical systems (CPS) play a crucial role, in various important sectors like healthcare, transportation and energy management. These interconnected systems require security measures to guard against cyber threats that could lead to major disruptions. Traditional security methods are not enough to handle the changing risks found in digital environments. This study presents an approach using a mix of self-adaptive systems (SAS) and Parametric Markov Decision Processes (pMDP) combined with attack trees to enhance CPS security. The goal is to improve how SAS can adapt dynamically to evolving security risks by calculating better policies. Thus by merging models, which offer a view of system states and changes, with attack trees that analyze possible attack pathways, the framework enables precise and context aware security policy creation. This allows the system to quickly assess and respond to threats in time, for resilience and operational stability. The framework was evaluated based on its ability to reduce system failures during attacks and the computational resources required for recalculating rewards and policies. This thorough evaluation ensures that the proposed solution not boosts security but also remains practical and efficient for real world use.
Nell'odierno mondo della tecnologia, l'uso dei sistemi cyber-fisici (CPS) svolge un ruolo cruciale in vari settori importanti come la sanità, i trasporti e la gestione dell'energia. Questi sistemi interconnessi richiedono misure di sicurezza avanzate per proteggersi dalle minacce informatiche che potrebbero causare gravi interruzioni. I metodi di sicurezza tradizionali non sono però sufficienti per gestire i rischi mutevoli degli ambienti digitali. Questo studio presenta un approccio che utilizza un mix di sistemi autoadattativi (SAS) e processi decisionali parametrici di Markov (pMDP) combinati con gli attack tree per migliorare la sicurezza dei CPS. L'obiettivo è perfezionare il modo in cui i SAS possono adattarsi dinamicamente all'evoluzione dei rischi per la sicurezza, calcolando politiche adatte. Ciò viene fatto unendo i modelli, che offrono una visione degli stati e dei cambiamenti del sistema, con gli attack tree, che analizzano i possibili percorsi di attacco, consentendo al framework di creare politiche di sicurezza precise e consapevoli del contesto. Ciò consente al sistema di valutare e rispondere rapidamente alle minacce in tempo utile, per garantire resilienza e stabilità operativa. Il framework è stato valutato in base alla sua capacità di ridurre i guasti del sistema durante gli attacchi e le risorse computazionali necessarie per ricalcolare le ricompense e le politiche. Questa valutazione approfondita non solo garantisce che la soluzione proposta aumenti la sicurezza, ma anche che questa rimanga pratica ed efficiente per l'uso nel mondo reale.
A hybrid self-adaptation framework for securing cyber-physical systems using Markov models and attack trees
POLETTI, LORENZO
2023/2024
Abstract
In todays world of technology use physical systems (CPS) play a crucial role, in various important sectors like healthcare, transportation and energy management. These interconnected systems require security measures to guard against cyber threats that could lead to major disruptions. Traditional security methods are not enough to handle the changing risks found in digital environments. This study presents an approach using a mix of self-adaptive systems (SAS) and Parametric Markov Decision Processes (pMDP) combined with attack trees to enhance CPS security. The goal is to improve how SAS can adapt dynamically to evolving security risks by calculating better policies. Thus by merging models, which offer a view of system states and changes, with attack trees that analyze possible attack pathways, the framework enables precise and context aware security policy creation. This allows the system to quickly assess and respond to threats in time, for resilience and operational stability. The framework was evaluated based on its ability to reduce system failures during attacks and the computational resources required for recalculating rewards and policies. This thorough evaluation ensures that the proposed solution not boosts security but also remains practical and efficient for real world use.File | Dimensione | Formato | |
---|---|---|---|
Lorenzo_Poletti_Summary_Final.pdf
accessibile in internet per tutti
Dimensione
951.28 kB
Formato
Adobe PDF
|
951.28 kB | Adobe PDF | Visualizza/Apri |
Lorenzo_Poletti_Thesis_Final.pdf
accessibile in internet per tutti
Dimensione
1.36 MB
Formato
Adobe PDF
|
1.36 MB | 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/10589/223066