Singapore Academic Cybersecurity R&D
Harnessing R&D to Secure our Nation
Security by Design for Interconnected Critical Infrastructures
- Lead PIs : Aditya Mathur, Professor, SUTD ( firstname.lastname@example.org ) and Deeph Chana, Professor, Imperial College London
- Host Institution : Information Systems Technology Design Pillar, SUTD
- Partner Institution : Imperial College London
In this project, we study the cascading effects of cyber attacks on three interconnected critical infrastructures (CIs), namely water treatment plant, water distribution network, power generation and distribution network (these interconnected testbeds are available at iTrust, SUTD).
- Modelling based abstraction from system design for security analysis
- Impact & response analysis across interconnected CI using the model
- Upgrading the initial design to improve system resilience to cyber attacks
- State Condition graphs used to model individual & interconnected CI
- Axiomatic design theory concepts are adopted to design secured public CI
Axiomatic design theory principles from system design are used to model critical infrastructures. This modelling provides an abstract representation of critical infrastructures to understand the behaviour of infrastructures under potential attacks. Axiomatic Design principles start with functional requirements and defines the design parameters that meet those functional requirements. The design parameters represent the cyber-physical systems components, and the process of defining these parameters automatically sets their inter-relations. We consider WADI as a case study for modelling. The design matrices are derived for the second stage of the WADI which gives the relationship between set of functional requirements and design parameters. These design matrices can help for the detection of potential attacks and analyse the impact of real attacks in a cyber physical system .
To detect the attacks we followed two approaches, namely system design invariants and data invariants. Design invariant rules that define the physical conditions that must be maintained for the normal operation of industrial control system provide a means by which early detection of anomalous system states may be achieved, allowing for timely mitigating actions – such as fault checking, system shutdown – to be taken. However, many hidden invariant rules can be extremely challenging to identify, particularly in circumstances where insights are needed across numerous subsystems and where dependencies between a wide-range of physical metrics may be implicit rather than explicit. Therefore, we use a combination of machine learning and data mining techniques to systematically learn invariants from information contained within ICS operational data logs. The data logs were collected every second by running WADI non-stop for a total of 16-days. The system was operated under normal conditions (without any attacks) for a period of 14 days. During the remaining two days, 15 different types of attacks were launched on the testbed [2,3].
We have developed a generic modelling framework that can represent different Critical Infrastructures (CIs) like Water Distribution, Water Treatment Plant and Power Generation etc. The model created using the proposed framework provides an abstract representation of a CI. The framework is developed by using an ”agent-based” approach, to model CIs as an aid to understand their collective behaviour when subjected to cyber-attacks. In this work, the components of an infrastructure are categorized into 5 major classes namely commodity, carrier, resource, actuators and sensors. This categorization is not limited to a single type of infrastructure but works for any critical infrastructure. The framework was used to create a model of an operational Water Distribution (WADI) testbed as an illustration of the utility of the approach proposed. The model was validated experimentally on water distribution testbed. An invariant-based attack detection mechanism was implemented and several attacks launched on the system to understand how well the model can represent the system. Also, this model can interconnect multiple infrastructures even if they are of different types, for example, water and power making it a flexible approach .
- R. Palleti, Jude Victor, Arlindo Silva, “A Contribution of Axiomatic Design Principles to the Analysis and Impact of Attacks on Critical Infrastructures” Accepted in International Journal of Critical Infrastructure Protection.
- Mujeeb, Palleti V. R, Aditya Mathur, “WADI: A Water Distribution Testbed for Research in the Design of Secure Cyber Physical Systems” proceedings of the 3 rd international workshop on cyber-physical systems for smart water networks, 25-28, Pittsburgh, PA, USA, 21 April 2017.
- Cheng Feng, Venkata Reddy Palleti, Aditya P. Mathur, Deeph Chana, “Learning Invariant Rules from Data Logs for Anomaly Detection in Industrial Control systems” Submitted to NDSS conference.
- Vishruth Kumar, Venkata Reddy PalletiAditya P. Mathur, “An Agent-based Modeling Framework for Critical Infrastructures” submitted to IEEE transactions on Dependable and Secure Computing