Attack Detection and State Estimation Based on Gaussian Mixture Model

Control Engineering of China ›› 2022, Vol. 29 ›› Issue (6) : 1027-1032.

Author information +

History +

Abstract

We consider the attack detection and state estimation of multi-sensor linear time-invariant systems against static sensor attacks. An attack detection algorithm based on Gaussian mixture model (GMM) is designed. Based on this, a new state estimation fusion method is proposed. Firstly, the local state estimates are modeled by Gaussian mixture model, and the expectation-maximization (EM) algorithm is used to cluster the local estimates. Then, the detection algorithm is designed according to the clustering results, which realizes the attack detection. Finally, based on the detection results, the local data are fused to achieve remote state estimation. Experimental results show that the proposed method can accurately locate the damaged sensors while guaranteeing the estimation accuracy, which effectively improves the detection accuracy and efficiency.

Key words

Cyber physical system / Gaussian mixture model / attack detection / secure state estimation

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

WANG Zi-le, ZHU Cui, ZANG Ze-yuan, Li Yu-xuan, ZHENG Huan-ming. Attack Detection and State Estimation Based on Gaussian Mixture Model[J]. Control Engineering of China, 2022, 29(6): 1027-1032