Secure Control for Cyber-Physical Systems under Stealthy Attacks via a Successive Convex Optimization Algorithm

This article investigates the secure control scheme for discrete-time cyber-physical systems (CPSs) under stealthy attacks. The goal is to design a controller such that the system runs safely under attack. Firstly, the stealthiness of attacks is characterized by the Kullback-Leibler divergence (KLD)...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on industrial cyber-physical systems s. 1 - 11
Hlavní autoři: Chen, Yufan, Ren, Yingying, Ding, Da-Wei
Médium: Journal Article
Jazyk:angličtina
Vydáno: IEEE 12.12.2025
Témata:
Actuators
Convex functions
Covariance matrices
Cyber-physical system
Cyber-physical systems
Linear matrix inequalities
Observers
reachable set
Robustness
Safety
Security
stealthy attack
successive convex optimization algorithm
Symmetric matrices
ISSN:2832-7004
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This article investigates the secure control scheme for discrete-time cyber-physical systems (CPSs) under stealthy attacks. The goal is to design a controller such that the system runs safely under attack. Firstly, the stealthiness of attacks is characterized by the Kullback-Leibler divergence (KLD), while the reachable set is introduced to quantify system safety. On this basis, an <inline-formula><tex-math notation="LaTeX">H_{\infty }</tex-math></inline-formula> performance is introduced to attenuate the effects of noise and attacks on the controlled output. Based on the reachable set and the <inline-formula><tex-math notation="LaTeX">H_{\infty }</tex-math></inline-formula> performance, the controller design can be depicted as a nonconvex optimization problem with bilinear matrix inequalities (BMIs) constraints. A successive convex optimization (SCO) algorithm is proposed to iteratively solve the problem, which ensures that the reachable set resides within a safe set while satisfying the required performance level. Eventually, numerical simulations and experiments validate the efficacy of the proposed method.
ISSN:2832-7004
DOI:10.1109/TICPS.2025.3643827