Authors:
Panagiotis Dedousis
1
;
George Stergiopoulos
1
;
2
;
George Arampatzis
3
and
Dimitris Gritzalis
1
Affiliations:
1
Dept. of Informatics, Athens University of Economics & Business, Athens, Greece
;
2
Dept. of Information & Communication Systems Engineering, University of the Aegean, Samos, Greece
;
3
School of Production Engineering & Management, Technical University of Crete, Chania, Greece
Keyword(s):
Critical Infrastructure Protection, Component Cascading Failures, Dependency Risk Graphs, Resilience.
Abstract:
During the past decades, and especially since the Stuxnet event, there has being a growing concern around the protection of critical infrastructures. Even though the protection of such systems and services has been an international security priority, still, even after all those years, relevant research either focuses on individual ICS systems security (PLC, RTU and SCADA network protection and attacks), or uses high-level models to perform risk assessments, mostly from a system-of-systems scope that studies interdependencies. From an engineering perspective, current approaches address system resilience from an efficiency perspective (i.e. focusing on the availability of physical processes) while neglecting the security dimension of their components. Still, the availability and reliability requirements of such systems are directly affected by security incidents. To our knowledge, there is currently no process to integrate security-by-design in industrial critical infrastructure engine
ering. To this end, we present a method to integrate security risk assessment analysis into engineering design practices. We do this by modeling internal dependencies between physical components in critical industrial production processes to identify possible hotspots of system failures that are challenging to handle later in the development lifecycle, especially during operation. To validate our approach, we model and assess the present situation in a portion of an actual oil refining plant, thereby establishing a baseline model. Then we introduce risk mitigation measures by altering the design of the baseline model, resulting in a reduction of the overall cascade risk.
(More)