Bio-inspired system identification attacks in noisy networked control systems

Abstract

The possibility of cyberattacks in Networked Control Systems (NCS), along with the growing use of networked controllers in industry and critical infrastructures, is motivating studies about the cybersecurity of these systems. The literature on cybersecurity of NCSs indicates that accurate and covert model-based attacks require high level of knowledge about the models of the attacked system. In this sense, recent works recognize that Bio-inspired System Identification (BiSI) attacks can be considered an effective tool to provide the attacker with the required system models. However, while BiSI attacks have obtained sufficiently accurate models to support the design of model-based attacks, they have demonstrated loss of accuracy in the presence of noisy signals. In this work, a noise processing technique is proposed to improve the accuracy of BiSI attacks in noisy NCSs. The technique is implemented along with a bio-inspired metaheuristic that was previously used in other BiSI attacks: the Backtracking Search Optimization Algorithm (BSA). The results indicate that, with the proposed approach, the accuracy of the estimated models improves. With the proposed noise processing technique, the attacker is able to obtain the model of an NCS by exploiting the noise as a useful information, instead of having it as a negative factor for the performance of the identification process.