Cyber-Resilient Distributed MPC for Securing Microgrids Against ADMM-Based Manipulation Attacks

This work proposes a cyber-resilient Distributed Model Predictive Control framework to improve microgrid security and operational reliability against cyber threats. It specifically addresses manipulation attacks on the weight negotiation process within the Alternating Direction Method of Multipliers (ADMM) algorithm, where a malicious agent can alter weight updates, resulting in a cyber-attack. To counter these threats, the proposed approach integrates anomaly detection mechanisms that analyze negotiation patterns and identify malicious behavior through residual analysis. Upon detecting deviations in critical operational parameters, this system implements targeted mitigation actions to minimize the attack’s impact. Simulation results demonstrate the effectiveness of this approach in detecting and mitigating ADMM-based manipulation attacks, emphasizing the necessity of integrating cybersecurity mechanisms into distributed control frameworks for enhanced microgrid resilience.