Boundary-Focused False Data Injection Attacks in Decentralized AC Power Systems

The shift toward decentralized control in AC power systems has introduced new vulnerabilities, especially at boundary buses, which are key nodes where inter-area data is exchanged to ensure coordinated state estimation across regions. While prior research on false data injection attacks (FDIAs) has largely focused on centralized or DC-based systems, limited attention has been given to boundary vulnerabilities in decentralized AC systems. This paper introduces an optimized FDIA model that targets boundary buses to disrupt decentralized state estimation processes. To overcome the inherent challenge of nonlinearity in AC system, we construct a linear basis for AC state estimation leveraging the grid's topology information, enabling a more tractable formulation. We then frame the distributed FDIA problem as a mixed-integer second-order cone programming (MISOCP) problem, which strategically allows for a resource-efficient attack design that maximizes disruption across interconnected power system areas while remaining computationally feasible. Extensive simulations on IEEE 14-bus and 118-bus systems validate the model's effectiveness in exploiting boundary bus vulnerabilities, underscoring the need for reinforced security in AC decentralized frameworks. These insights provide essential guidance for enhancing the resilience and security of smart grids, supporting proactive measures against future cyber-threats.