Design of Load Forecast Systems Resilient Against Cyber-Attacks

Barreto, Carlos; Koutsoukos, Xenofon

doi:10.1007/978-3-030-32430-8_1

Cited by 15 publications

(14 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…У процесі модифікації алгоритму з метою визначення ризику, використовується наступне припущення. Абсолютна величина втрат у грошовому еквіваленті для конкретної ІКС залежить від ланцюжка елементів -загрози, уразливості, ЗЗІ [2,3]. Отже, кількість ризиків -це кількість комбінацій загроз та активів:…”

Section: теоретичні основи дослідженняunclassified

Optimization of Placement of Information Protection Means Based on the Application of a Genetic Algorithm

et al. 2022

View full text Add to dashboard Cite

the article considers the possibilities of modifying the genetic algorithm (GA) for solving the problem of selecting and optimizing the configurations of information protection means (IPR) for security circuits of information and communication systems (ICS). The scientific novelty of the work lies in the fact that in GA, as criteria for optimizing the composition of IPR, it is proposed to use the total value of risks from loss of information, as well as the integral indicator of IPR and cost indicators for each class of IPR. The genetic algorithm in the task of optimizing the selection of the composition of the IPR for ICS is considered as a variation of the problem associated with multiple selection. In such a statement, the optimization of the placement of IPR along the contours of ICS protection is considered as a modification of the combinatorial problem about the backpack. The GA used in the computing core of the decision support system (DSS) differs from the standard GA. As part of the GA modification, chromosomes are presented in the form of matrices, the elements of which are numbers that correspond to the numbers of the IPR in the ICS nodes. In the process of GA modification, k-point crossover was applied. The fitness function is represented as the sum of efficiency coefficients. At the same time, in addition to the traditional absolute indicators of the effectiveness of IPR, the total value of risks from loss of information, as well as cost indicators for each class of IPR are taken into account. The practical value of the research lies in the implementation of the DSS based on the proposed modification of the GA. Computational experiments on the selection of a rational software algorithm for the implementation of the model were performed. It is shown that the implementation of GA in DSS allows to speed up the search for optimal options for the placement of cyber security means (CS) for ICS by more than 25 times. This advantage allows not only to perform a quick review of various options of hardware and software IPR and their combinations for ICS, but also to further combine the proposed algorithm with existing models and algorithms for optimizing the composition of ICS cyber security circuits. Potentially, such a combination of models and algorithms will provide an opportunity to quickly rebuild ICS protection, adjusting its profiles in accordance with new threats and classes of cyberattacks.

show abstract

Section: теоретичні основи дослідженняunclassified

Optimization of Placement of Information Protection Means Based on the Application of a Genetic Algorithm

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Аналіз останніх досліджень і публікацій. Збільшення кількості і складності успішно реалізованих кібератак на різні ОБІ [1,2] породжує потребу в якісно нових процедурах формування складу комплексів СЗІ та кібербезпеки (КБ) для всіх контурів захисту інформаційних масивів ОБІ. Задача, що не втрачає актуальність формування ефективних контурів захисту інформації (ЗІ) та КБ ОБІ породила безліч теоретичних і прикладних досліджень, присвячених питанням оптимізації складу СЗІ та КБ [3,4].…”

Section: постановка проблемиunclassified

“…Рішення виконано на основі загального еволюційного алгоритму і його складових багатокритеріальних ГА. Основним при цьому став метод VEGA -Vector Evaluted Genetic Algorithm [1,2].…”

Section: методика дослідженняunclassified

Efficiency of the Indicators Investment Calculation Method in the Information Security System of Information Objects

et al. 2021

View full text Add to dashboard Cite

The article describes the methodology of multi-criteria optimization of costs for the information protection system of the object of informatization. The technique is based on the use of a modified VEGA genetic algorithm. A modified algorithm for solving the MCO problem of parameters of a multi-circuit information protection system of an informatization object is proposed, which makes it possible to substantiate the rational characteristics of the ISS components, taking into account the priority metrics of OBI cybersecurity selected by the expert. In contrast to the existing classical VEGA algorithm, the modified algorithm additionally applies the Pareto principle, as well as a new mechanism for the selection of population specimens. The Pareto principle applies to the best point. At this point, the solution, interpreted as the best, if there is an improvement in one of the cybersecurity metrics, and strictly no worse in another metric (or metrics). The new selection mechanism, in contrast to the traditional one, involves the creation of an intermediate population. The formation of an intermediate population occurs in several stages. At the first stage, the first half of the population is formed based on the metric - the proportion of vulnerabilities of the object of informatization that are eliminated in a timely manner. At the second stage, the second half of the intermediate population is formed based on the metric - the proportion of risks that are unacceptable for the information assets of the informatization object. Further, these parts of the intermediate population are mixed. After mixing, an array of numbers is formed and mixed. At the final stage of selection for crossing, specimens (individuals) will be taken by the number from this array. The numbers are chosen randomly. The effectiveness of this technique has been confirmed by practical results

show abstract

“…Most of the research on adversarial examples focus on classification problems; however, other applications also have vulnerabilities. [6,8] show how an adversary can manipulate sensor measurements to increase or decrease load forecasts.…”

Section: • Model Inversion Attacks Allow the Recovery Of The (Private)mentioning

confidence: 99%

@Pad

Ozdagli

Barreto

Koutsoukos

2020

Proceedings of the 7th Symposium on Hot Topics in the Science of Security

Self Cite

View full text Add to dashboard Cite

In this work, we study the vulnerabilities of protection systems that can detect cyber-attacks in power grid systems. We show that machine learning-based discriminators are not resilient against Denial-of-Service (DoS) attacks. In particular, we demonstrate that an adversarial actor can launch DoS attacks on specific sensors, render their measurements useless and cause the attack detector to classify a more sophisticated cyber-attack as a normal event. As a result of this, the system operator may fail to take action against attack-related faults leading to a decrease in the operation performance. To realize a DoS attack, we present an optimization problem to determine which sensors to attack within a given budget such that the existing classifier can be deceived. For linear classifiers, this optimization problem can be formulated as a mixed-integer linear programming problem. In this paper, we extend this optimization problem to find attacks for more complex classifiers such as neural networks. We demonstrate that a neural network, in particular, with RELU activation functions, can be represented as a set of logic formulas using Disjunctive Normal Form, and the optimization problem can be used to efficiently compute a DoS attack. In addition, we propose a defense model that improves the resilience of neural networks against DoS through adversarial training. Finally, we evaluate the efficiency of the approach using a dataset for classification in power systems.

show abstract

Design of Load Forecast Systems Resilient Against Cyber-Attacks

Cited by 15 publications

References 19 publications

Optimization of Placement of Information Protection Means Based on the Application of a Genetic Algorithm

Optimization of Placement of Information Protection Means Based on the Application of a Genetic Algorithm

Efficiency of the Indicators Investment Calculation Method in the Information Security System of Information Objects

@Pad

Contact Info

Product

Resources

About