Key points are not available for this paper at this time.
The resilience of Distributed State Estimation (DSE) in power systems depends not only on detecting cyber-attacks but also on the ability to robustly recover accurate system states during persistent malicious data injections. While existing trust-based frameworks can identify compromised nodes, they often lack a mechanism for continuous state restoration, a critical vulnerability under persistent attacks. This paper proposes a Generalized Distributed Information Kalman Filter (G-DIKF), a novel security scheme designed specifically for the post-attack restoration phase. When an agent is deemed untrustworthy by a cooperative trust management system, the G-DIKF activates a robust recovery protocol. This protocol employs an iterative, batch-mode regression that adaptively reweights all available information—including historical data, local observations, and data from trusted neighbors—based on dynamically assessed trust scores. This approach ensures that state variables are corrected without discarding potentially valuable information. A key feature of the G-DIKF is its scalability; the computationally intensive recovery process is localized, involving only the compromised agents and their immediate neighbors. The efficacy and robustness of this trust-aware restoration mechanism are validated through simulations on the IEEE 14-bus and 118-bus test systems, demonstrating significant improvements in estimation accuracy during multi-time-step attack scenarios.
Nasiri et al. (Mon,) studied this question.