Fixed-Time Optimal Consensus of Multiagent Systems Under Cyber-Attacks: A Hierarchical Control Approach

Area of Science:

• Control Systems Engineering
• Artificial Intelligence
• Cybersecurity

Background:

• Multiagent systems (MASs) face challenges in achieving consensus under cyber-attacks like Denial of Service (DoS) and False Data Injection (FDI).
• Existing methods often lack fixed-time convergence or struggle with unknown system dynamics and complex attack scenarios.

Purpose of the Study:

• To develop a fixed-time optimal leader-following consensus strategy for unknown MASs under DoS and FDI attacks.
• To enhance system resilience and performance despite cyber threats.
• To reduce computational complexity for settling time determination.

Main Methods:

• A novel fixed-time stability theorem for DoS attacks.
• Deep Neural Networks (DNNs) with projection operators to approximate unknown system dynamics.
• A hierarchical control approach with distributed and Luenberger-based observers for state estimation.
• An event-triggered mechanism (ETM) for optimal control strategy design.
• A critic-only reinforcement learning (RL) algorithm for online optimal control learning.

Main Results:

• Achieved fixed-time state reconstruction using developed observers under multiple malicious attacks.
• Demonstrated an optimal control policy capable of tracking observation states within a fixed time.
• Validated the effectiveness of the proposed technique through simulation of automated vehicle platooning control.

Conclusions:

• The proposed hierarchical control framework effectively addresses the fixed-time optimal leader-following consensus problem in unknown MASs under sophisticated cyber-attacks.
• The integration of DNNs, RL, and ETM provides a robust and computationally efficient solution.
• The method ensures stability and optimal performance, applicable to real-world scenarios like autonomous vehicle coordination.