Secure State Estimation Against Sparse Sensor Attacks: A Distributed Adaptive Saturation Method

Qingjie Wang

doi:10.1109/TASE.2025.3649048

Secure State Estimation Against Sparse Sensor Attacks: A Distributed Adaptive Saturation Method

2026-01-01

0

PRE

AI

DOI：10.1109/TASE.2025.3649048

原文链接

原文求助

摘要

En 中文

This paper investigates the problem of distributed secure state estimation for cyber-physical systems under sparse sensor attacks. Traditional centralized methods struggle with high computational complexity in large-scale systems, while existing distributed approaches either require strong robustness constraints for graphs or face challenges due to local unobservability. To address these challenges, we propose a distributed adaptive saturation method. This method designs an adaptive saturation function for each node to limit the impact of attacks on the estimation residual. Additionally, it integrates a consensus protocol to enable cooperative estimation of nodes, effectively mitigating the influence of malicious attacks. Theoretical analysis shows that the algorithm ensures consensus estimation under fixed attack channels, with estimation errors ultimately bounded. Furthermore, by refining the saturation mechanism for each node, the algorithm is extended to scenarios with time-varying attacks. Simulation experiments validate that the proposed method effectively handles sparse sensor attacks, reduces computational complexity, and significantly enhances system robustness. Note to Practitioners—This paper addresses the security of state estimation in cyber-physical systems (e.g., power grids, automated factories) when sensors are compromised by malicious attacks. Existing solutions typically demand excessive computational resources or rely on impractical network constraints. We propose a distributed adaptive saturation method that does not require detection of attacks and achieves accurate state estimation. Each remote estimation node dynamically adjusts its sensitivity to anomalous data while collaborating with neighboring nodes via a consensus protocol to jointly mitigate attack impacts. Compared to centralized approaches, this method significantly reduces computational burdens and operates reliably under standard network connectivity without requiring highly robust topologies. It handles both fixed and dynamically shifting attack patterns, making it suitable for evolving threats. Simulation tests on autonomous vehicles confirm robust performance under intense sensor attacks. Current applicability is limited to linear systems; extensions to nonlinear dynamics are planned. For engineers, this approach provides a practical solution to enhance security in critical infrastructure like smart grids (countering false data injection) and automated plants (resisting sensor tampering).

Keyword:

Secure state estimation

adaptive saturation

cyber-physical systems

distributed algorithms

Secure State Estimation Against Sparse Sensor Attacks: A Distributed Adaptive Saturation Method

摘要

期刊

机构

引用论文