Spanish Researchers Develop AI Agents to Protect EV Chargers
Researchers at the University of Malaga's NICS lab have created a multi-agent AI system that detects and prevents cyberattacks on EV charging infrastructure using OCPP, opinion dynamics consensus, and blockchain.
A Novel Approach to EV Charging Security
As the number of electric vehicles on roads worldwide continues to grow, so does the charging infrastructure—and with it, new cybersecurity risks. Researchers at the NICS lab of the University of Malaga in Spain have proposed an innovative solution: a multi-agent artificial intelligence system designed to protect EV charging stations from various cyber threats, ranging from fraud and energy theft to large-scale attacks that could destabilize power grids.
Cristina Alcaraz, the lead author of the study and an infrastructure security researcher, notes that the complex architecture of charging stations, which integrates both physical and digital components, creates a wide range of vulnerabilities. Current monitoring mechanisms based on the Open Charge Point Protocol (OCPP) often focus only on network traffic or local events, providing a limited view of the overall situation. This makes it difficult to identify where an anomaly occurs, which components are compromised, and how an attack might spread.
How the AI Agents Work
The proposed system employs multiple AI agents embedded in each charging station or relevant network component. These agents can analyze their environment, collect information, and collaborate with other agents to build a comprehensive view of the infrastructure's status. Each agent assesses the condition of chargers, communications, and connected devices to detect anomalies, operational failures, or security incidents. The agents then compare locally obtained data with that from nearby stations, providing a more complete and contextualized collaborative picture.
A key novelty is the consensus mechanism based on a mathematical framework called opinion dynamics. This approach mimics how humans exchange information within social networks to reach agreements. Applied to computer models, it allows AI agents to share observations and gradually adjust their assessments, building a collective understanding. This reduces false positives and enables the detection of anomalies that might be missed through local analysis alone.
The architecture also incorporates blockchain technology as a trust and validation mechanism. All transactions performed by the agents are recorded in an immutable distributed ledger, ensuring integrity and traceability.
Testing and Results
The multi-agent system was tested in a simulated OCPP-compliant charging environment. Agents were exposed to various anomaly scenarios, including component failures, communication link errors, and situations requiring coordinated responses. In all cases, the agents successfully identified local disturbances, shared observations, and collaborated to build a shared understanding of incidents.
Results showed that the combination of AI agents, the distributed consensus mechanism, and blockchain provided a global view of the network. The system detected both specific anomalies in individual devices and behavioral patterns affecting multiple charging stations. The consensus mechanism improved diagnostic accuracy by comparing observations from different agents.
The study was published in the International Journal of Critical Infrastructure Protection. The university lab expressed satisfaction, stating that the system offers a new way to guarantee the protection of electric-vehicle charging infrastructure.
