The frequency and severity of cyberattacks on critical infrastructure is concerning not only for private sector companies, but for governments as well.

Along with those concerns are the costs associated with cybersecurity, making the efficient allocation of resources paramount.

To that end, a framework featuring a more holistic picture of the cybersecurity landscape, along with a model that explicitly represents multiple dimensions of the potential impacts of successful cyberattacks, is currently in development.

As critical infrastructure such as electric power grids become more sophisticated, they are also becoming increasingly more reliant on digital networks and smart sensors to optimize their operations, which can increase attack surface and make them more vulnerable.

Over the past couple of years, cyberattacks on critical infrastructure have become ever more complex and disruptive, causing systems to shut down, disrupting operations, or enabling attackers to remotely control affected systems. Importantly, the impacts of successful attacks on critical cyber-physical systems are multidimensional in nature, which means impacts are not only limited to losses incurred by the operators of the compromised system, but also economic losses to other parties relying on their services as well as public safety or environmental hazards.

Schneider Bold

According to the study just published in the journal Risk Analysis, this makes it important to have a tool that distinguishes between different dimensions of cyber risks and also allows for the design of security measures able to make the most efficient use of limited resources.

The authors set out to answer two main questions:

  1. Whether it is possible to find vulnerabilities, the exploitation of which opens ways for several attack scenarios to proceed
  2. If it is possible to take advantage of this knowledge and deploy countermeasures to simultaneously protect the system from several threats

One of the ways in which cyber threats are commonly managed, is to conduct an analysis of individual attack scenarios through risk matrices, prioritizing the scenarios according to their perceived urgency (depending on their likelihoods of occurrence and severity of potential impacts), and then addressing them in order until all the resources available for cybersecurity are spent. According to the authors, this approach may however lead to suboptimal resource allocations, given that potential synergies between different attack scenarios and among available security measures are not taken into consideration.

Cost vs. Losses
“Existing assessment frameworks and cyber security models assume the perspective of the operator of the system and support her cost-benefit analysis, in other words, the cost of security measures versus potential losses in the case of a successful cyberattack. Yet, this approach is not satisfactory in the context of security of critical infrastructure, where the potential impacts are multidimensional and may affect multiple stakeholders. We endeavored to address this problem by explicitly modeling multiple relevant impact dimensions of successful cyberattacks,” said lead author Piotr Żebrowski, a researcher in the Exploratory Modeling of Human-natural Systems Research Group of the International Institute for Applied Systems Analysis (IIASA) Advancing Systems Analysis Program.

IIASA is an international research institute that advances systems analysis and applies its research methods to identify policy solutions to reduce human footprints, enhance the resilience of natural and socioeconomic systems, and help achieve the Sustainable Development Goals.

To overcome this shortcoming, researchers propose a quantitative framework that features a more holistic picture of the cybersecurity landscape that encompasses multiple attack scenarios, thus allowing for a better appreciation of vulnerabilities.

To do this, the team developed a Bayesian network model representing a cybersecurity landscape of a system. This method has gained popularity in the last few years due to its ability to describe risks in probabilistic terms and to explicitly incorporate prior knowledge about them into a model that can be used to monitor the exposure to cyber threats and allow for real-time updates if some vulnerabilities have been exploited.

In addition to this, the researchers built a multi-objective optimization model on top of the Bayesian network that explicitly represents multiple dimensions of the potential impacts of successful cyberattacks.

Broader Perspective
The framework adopts a broader perspective than the standard cost-benefit analysis and allows for the formulation of more nuanced security objectives. The study also proposes an algorithm that is able to identify a set of optimal portfolios of security measures that simultaneously minimize various types of expected cyberattack impacts, while also satisfying budgetary and other constraints.

The researchers note while the use of models like this in cybersecurity is not entirely unheard of, the practical implementation of such models usually requires extensive study of systems vulnerabilities. In their study, the team however suggests how such a model can be built based on a set of attack trees, which is a standard representation of attack scenarios commonly used by the industry in security assessments. The researchers demonstrated their method with the help of readily available attack trees presented in security assessments of electric power grids in the U.S.

“Our method offers the possibility to explicitly represent and mitigate the exposure of different stakeholders other than system operators to the consequences of successful cyberattacks. This allows relevant stakeholders to meaningfully participate in shaping the cybersecurity of critical infrastructure,” Żebrowski said.

The researchers said it is important to have a systemic perspective on the issue of cybersecurity. This is crucial in terms of establishing a more accurate landscape of cyber threats to critical infrastructure and in the efficient and inclusive management of important systems in the interest of multiple stakeholders.

Pin It on Pinterest

Share This