Fusion-Based FDI Attack Detection in Cyber-Physical Systems

Abstract

This brief studies the alarm response problem of attack detection for cyber-physical systems under false data injection attacks. Notice that the faster the alarm response of the attack detection is, the less the system performance loss is. In this case, the distributed fusion strategy, which can potentially provide higher accuracy and reliability, is introduced, for the first time, to improve the speed of alarm response. Meanwhile, to ensure real-time anomaly detection online, a group of sensors deployed by different positions send their messages to a monitoring center through communication networks. Due to the limited bandwidth, multiple finite-level logarithmic quantizers are used to reduce the size of data packages containing residual messages (generated by local state estimators). Then, an optimal weighting fusion criterion is designed by establishing convex optimization problems such that the residual evaluation threshold is more accurate in the presence of quantization errors. Finally, an F-404 aircraft example is used to illustrate that the response speed of attack detection becomes faster as compared with the case of single sensor.