Privacy-preserving resilience control for cyber-physical systems based on the dynamic heterogeneous redundancy architecture

This paper addresses the security control challenges of cyber-physical system (CPS) under coordinated false data injection and eavesdropping attacks. Existing research often relies on stacking multiple defense techniques to counter complex and diverse coordinated attacks, resulting in increased system complexity, interference among defense mechanisms, and reduced resource efficiency. To overcome these limitations, we propose a cyber-physical fusion proactive-reactive defense architecture. This architecture is built upon dynamic heterogeneous redundancy (DHR) and integrates confidential interaction protocols to defend against coordinated false data injection and eavesdropping attacks in the cyber domain. We design a lightweight confidential interaction protocol based on elliptic curve cryptography that ensures data transmission security while meeting CPS real-time requirements. Additionally, we propose a DHR scheduling algorithm based on historical trustworthiness and heterogeneity that balances detection reliability, heterogeneity maximization, and system stability. To address challenges faced by DHR architecture in practical applications—including limited heterogeneous resources, underlying homogeneity risks, and switching frequency constraints—we design an attack detection filter in the physical domain as a complementary measure. The developed proactive-reactive defense control algorithm enhances system unpredictability through dynamic controller switching (proactive defense) while timely identifying compromised controllers using attack detection filters (reactive defense). Theoretical analysis demonstrates that the closed-loop system maintains stability even under unknown attacks. Simulation experiments across various CPS scenarios confirm that our proposed resilient control method enhances the system's ability to resist coordinated false data injection and eavesdropping attacks, validating the effectiveness of our approach.