Attack-resilient adaptive distributed neurodynamic approach for solving noncooperative games.

Area of Science:

• Control Theory
• Game Theory
• Network Security

Background:

• Noncooperative games face risks from exploratory and causative attacks.
• Existing methods with fixed communication switching sequences are vulnerable to information leakage.
• Private inequality constraints and external disturbances complicate game dynamics.

Purpose of the Study:

• To develop robust adaptive distributed neurodynamic approaches for noncooperative games under adversarial conditions.
• To enhance resilience against exploratory and causative attacks.
• To ensure reliable convergence and disturbance rejection.

Main Methods:

• Proposed an adaptive distributed neurodynamic approach with continuous-time randomly switching communication topologies.
• Developed an attack-resilient adaptive distributed approach with discrete-time switching communication.
• Utilized singular perturbation techniques for convergence proofs and disturbance-observer-based control for disturbance rejection.

Main Results:

• Randomized communication switching enhances defense against eavesdropping.
• The discrete-time approach guarantees exponential convergence to the Nash Equilibrium (NE) even under attack.
• Proposed methods demonstrate faster convergence, improved resilience, and greater adaptability compared to existing approaches.

Conclusions:

• The novel neurodynamic approaches offer superior performance in adversarial noncooperative game scenarios.
• Validated through application to autonomous vehicle networks, confirming real-world applicability.
• These strategies provide a strong foundation for secure and efficient decentralized decision-making.