Observer-based ADP for secure resource allocation in high-order nonlinear multi-agent systems under FDI attacks

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Artificial Intelligence

Background:

• Multi-agent systems (MASs) are vulnerable to false data injection (FDI) attacks.
• Attacks can target both actuators and sensors, disrupting control and resource allocation.
• Existing resource allocation methods may fail under such sophisticated attacks.

Purpose of the Study:

• To develop a secure resource allocation framework for high-order nonlinear MASs under simultaneous actuator and sensor FDI attacks.
• To ensure closed-loop stability and accurate resource allocation despite adversarial conditions.
• To propose a novel observer-based adaptive dynamic programming (ADP) approach.

Main Methods:

• An adaptive attack-estimation mechanism is designed to counteract FDI.
• A deviation dynamic system incorporates gradient and multiplier dynamics to meet Karush-Kuhn-Tucker conditions.
• A single-critic neural network approximates the Hamilton-Jacobi-Bellman equation for online control.
• The framework jointly optimizes allocation error, control energy, and attack compensation.

Main Results:

• The proposed method ensures practical fixed-time stability for closed-loop error signals.
• The resource allocation satisfies approximate Karush-Kuhn-Tucker conditions within a fixed time.
• Numerical simulations validate the effectiveness of the secure allocation controller.

Conclusions:

• The observer-based ADP framework effectively secures resource allocation in MASs against FDI attacks.
• The method provides robust and stable control performance under cyber threats.
• This research contributes to secure and intelligent control of complex systems.