False Data Injection Attack Detection Under Uncertain Parameters in Cyber-Physical Systems

Abstract

The problem of detecting false data injection (FDI) attacks in a class of cyber-physical systems with discrete-time information under uncertain parameters is investigated in this article. In practical scenarios, it is noteworthy that accurately modeling the system matrix is often challenging, and traditional detection methods frequently result in false positives and missed detections. Therefore, this article introduces a novel approach for accurately detecting FDI attacks in the presence of parameter uncertainties. Assuming that system parameters reside within a convex polytope, three distinct FDI attack detection estimators with varying performance levels are proposed. Subsequently, a convex optimization problem is established to obtain the necessary gains for the detectors. The results demonstrate that the proposed detection method achieves higher accuracy under uncertain parameters. Furthermore, the method allows for flexible selection based on application-specific requirements, whether prioritizing accuracy or efficiency. Finally, the effectiveness and applicability of the proposed techniques are demonstrated through simulation examples using the F-404 aerospace engine system model.