Research units A-Z

FDI Modelling, Detection, and Mitigation for Modern EPES

The digitalization of electric power and energy systems (EPES) and Smart Grids (SGs) has significantly increased cyberattacks, such as, false data injection (FDI). False Data Injection Attacks (FDIA) are one of the most challenging threats to grid operations. Different from other attacks such as jamming and flooding, FDIA can be stealthily launched without being detected by bad data detection (BDD) algorithms. By tampering sensor’s and meter’s measurements, FDI attacks interfere with power systems’ regular operation. The direct, immediate, and critical consequences of FDI in SGs, such as, generators tripping, transmission lines overloading, and cascading blackouts may lead to equipment damage and endangering lives, and necessitate resilient protection strategies for these critical infrastructures. Consequently, it is imperative to investigate FDI modelling strategies adopted by adversaries, including data-driven approaches, attacks based on partial system knowledge, and attacks with targeted constraints.

Recently, in-network threat detection techniques have gained traction by leveraging the programmability of extended Berkeley Packet Filtering (eBPF). This approach enhances performance by enabling timely, layered, and systematic protection for smart cities, transportation systems, and smart grids. Through the integration of DL-based approximate solutions within the kernel network stack, accelerated real-time detection, reporting, and mitigation of FDI attacks is enabled. This is achieved by developing specifically designed packet-level primitives on a reduced instruction set architecture. Variant of active and passive mitigation strategies support flexible deployment with resilient protection against ever increasing cyberattacks on OT networks.

 

Publications