Resilient distributed optimization for cyber-physical systems under adversarial environments: An event-based method.

Area of Science:

• Cyber-Physical Systems (CPSs)
• Distributed Optimization
• Control Theory

Background:

• Cyber-physical systems require robust optimization algorithms to function reliably under adversarial conditions.
• Existing distributed optimization methods often lack resilience against sophisticated network attacks.
• Event-triggering mechanisms offer potential for efficient and secure distributed control.

Purpose of the Study:

• To develop a resilient distributed optimization algorithm for cyber-physical systems (CPSs) that operates effectively under adversarial environments.
• To address two specific attack scenarios: f-total and f-local attacks.
• To ensure the convergence and optimality of benign agents despite malicious agent behavior.

Main Methods:

• An event-triggering mechanism is integrated into a distributed optimization framework.
• The algorithm is designed to optimize average convex cost functions for multiple agents.
• Theoretical analysis is used to derive convergence and optimality conditions under attack scenarios.

Main Results:

• The proposed resilient distributed optimization with event-triggering (RDO-E) algorithm guarantees convergence of benign agents to a bounded error range under both f-total and f-local attacks.
• Optimality conditions are established, ensuring convergence to a safety interval even with malicious agents.
• Numerical examples demonstrate superior accuracy and reduced communication overhead compared to existing resilient algorithms.

Conclusions:

• The RDO-E algorithm provides a robust and efficient solution for distributed optimization in adversarial CPS environments.
• The algorithm's practical viability is validated through successful application to a resilient economic dispatch problem in a multi-microgrid system.