Enhanced distributed state estimation with resilience to multiple disturbances and false data injection attacks

Disturbances and attacks co-exist widely in networked systems. Distributed state estimation of networked systems in the simultaneous presence of disturbances and attacks has been widely recognized as a challenging issue. In this article, an enhanced three-stage approach for the detection and rejection of attacks is established for networked systems subject to multiple disturbances and false data injection (FDI) attacks. First, a multi-layer distributed estimator with a disturbance observer layer is proposed. Then, an optimal observer-based attack detection scheme is designed to specify an attack detection logic. After disturbance rejection and attack detection, an attack-resilient estimator with a dynamic detection-triggered structure is proposed to actively reject FDI attacks online. This three-stage approach effectively separates the influence of disturbances and attacks in the distributed estimation process, enabling multiple heterogeneous disturbances and attacks across the sensor network to be rejected. A numeric example and comparisons are provided to illustrate the effectiveness of the proposed approach.