Decentralized optimal H∞ fusion estimation for multi-sensor networked systems with two-channel hybrid attacks

• Published in: ISA transactions Vol. 156; pp. 168 - 178
• Main Authors: Zhang, Lei, Sun, Shuli
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.01.2025
• Subjects: Decentralized optimal H∝ fusion estimation; Exponentially mean-square stability; Hybrid attack; Linear matrix inequality; Multi-sensor networked system; Decentralized optimal H∝ fusion estimation; Multi-sensor networked system; Linear matrix inequality; Hybrid attack; Exponentially mean-square stability; Decentralized optimal H(∝) fusion estimation
• ISSN: 0019-0578; 1879-2022; 1879-2022
• DOI: 10.1016/j.isatra.2024.10.025

The article aims at the decentralized optimal H∞ fusion estimation issue for multi-sensor networked systems with insecure network communications, where hybrid attacks consisting of stochastic deception and denial-of-service attacks happen on both the sensor-to-local filter channel and the local filter-to-fusion center channel simultaneously. Some random variables obeying Bernoulli distributions are utilized to depict the hybrid attacks existing in two classes of communication channels in a unified framework. Relying on a novel augmentation method, the fusion estimation error system with globally internal dynamics is obtained. Two sufficient conditions to assure the corresponding H∞ performance and exponentially mean-square stability of the local and fusion estimation error systems are derived. To reduce the adverse effect of hybrid attacks, the decentralized optimal H∞ fusion filter with better H∞ performance index than each local H∞ filter is presented by linear matrix inequality technique. An actual civil aircraft system demonstrates the algorithms to be valid. •Hybrid attacks are launched on sensor-local filter channel and local filter-fusion center channels simultaneously.•A decentralized fusion H∞ filter is designed by linear matrix inequality.•The proposed fusion filter has smaller H∞ performance index than local filters.