Fuse and Federate: Enhancing EV Charging Station Security with Multimodal Fusion and Federated Learning
Rabah Rahal, Abdelaziz Amara korba, Yacine Ghamri-Doudane · 2025
Read Fuse and Federate: Enhancing EV Charging Station Security with Multimodal Fusion and Federated Learning by Rabah Rahal, Abdelaziz Amara korba, Yacine Ghamri-Doudane (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-06730.txt.
Introduction
Rabah Rahal and co-authors guard electric car chargers with a security system that learns from two senses at once. Charging stations sit exposed on the smart grid and draw reconnaissance, backdoor intrusions, and denial-of-service attacks that single-view detectors miss. Their framework fuses network traffic with kernel events from the station's own machine.
Learning is federated: stations train together without shipping raw data anywhere, preserving privacy across the fleet. In decentralized tests the system caught over ninety-eight percent of attacks with precision above ninety-seven percent, beating existing solutions. The authors pitch the design as a scalable shield for infrastructure that must stay both open to drivers and closed to attackers.
Catalog record
| CREATOR | Rabah Rahal, Abdelaziz Amara korba, Yacine Ghamri-Doudane |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-06730
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fuse and Federate: Enhancing EV Charging Station Security with Multimodal Fusion and Federated Learning Rabah Rahal, Abdelaziz Amara korba, Yacine Ghamri-Doudane Abstract The rapid global adoption of electric vehicles (EVs) has established electric vehicle supply equipment (EVSE) as a critical component of smart grid infrastructure. While essential for ensuring reliable energy delivery and accessibility, EVSE systems face significant cybersecurity challenges, including network reconnaissance, backdoor intrusions, and distributed denial-of-service (DDoS) attacks. These emerging threats, driven by the interconnected and autonomous nature of EVSE, require innovative and adaptive security mechanisms that go beyond traditional intrusion detection systems (IDS). Existing approaches, whether network-based or host-based, often fail to detect sophisticated and targeted attacks specifically crafted to exploit new vulnerabilities in EVSE infrastructure. This paper proposes a novel intrusion detection framework that leverages multimodal data sources, including network traffic and kernel events, to identify complex attack patterns. The framework employs a distributed learning approach, enabling collaborative intelligence across EVSE stations while preserving data privacy through federated learning. Experimental results demonstrate that the proposed framework outperforms existing solutions, achieving a detection rate above 98% and a precision rate exceeding 97% in decentralized environments. This solution addresses the evolving challenges of EVSE security, offering a scalable and privacy-preserving response to advanced cyber threats. I Introduction The global electric vehicle (EV) market has surged from 26 million EVs on the road in 2022 to over 40 million in 2024, reflecting a 53.85% growth rate in sales [ 1 ] . This rapid expansion makes it imperative to adapt current smart grid infrastructure, including Electric Vehicle Supply Equipment (EVSE), to meet the demands of this fast-growing sector. Each EVSE deployed must provide flexible payment solutions for access to services, such as credit card payments, mobile applications, or prepaid monthly cards. However, this accessibility to sensitive payment and user information has made EVSEs an attractive target for cyber threats, emphasizing the need for robust cybersecurity measures to protect user confidentiality and privacy. Despite advances in securing traditional infrastructure, EVSEs continue to exhibit significant vulnerabilities. Recent incidents underscore the associated risks: in 2024, ransomware attacks on charging stations surged by 90%, with cybercriminals encrypting station systems and demanding ransom payments for restoration [ 2 ] . Furthermore, research has identified critical flaws in the widely-used Open Charge Point Protocol (OCPP), making charging sessions vulnerable to disruptions and unauthorized data access [ 3 , 4 ] . These incidents emphasize the urgent need for advanced cybersecurity solutions to protect EVSEs from a wide array of threats. Current security mechanisms, such as network-based [ 4 , 5 ] and host-based IDS, often rely on modeling the behavior of EV charging stations using a single type of log within their local scope. This approach may lack the depth and adaptability required to detect sophisticated attack patterns and multi-layered cyber threats targeting EVSEs. Current intrusion detection systems (IDS) for EV charging infrastructure have made progress in addressing cyber threats but face significant limitations. Many rely on complex architectures or multiple sub-models, which can be resource-intensive and unsuitable for deployment on constrained EVSEs [ 6 , 7 ] . Others focus on specific attack scenarios like injection attacks in vehicle-to-grid communication but lack flexibility for diverse threats [ 8 , 9 ] . Emerging AI-driven solutions incorporating blockchain and reinforcement learning (RL) show promise - blockchain ensures transaction security [ 10 , 11 ] despite interoperability and storage challenges, while RL enables dynamic adaptation [ 12 , 13 ] though requiring extensive data and raising ethical concerns. However, these advanced approaches still struggle with high computational costs, poor cross-network generalization, and the fundamental trade-off between scalability, adaptability, and resource efficiency. This highlights the critical need for more robust IDS frameworks capable of leveraging diverse data sources while overcoming these persistent limitations. Developing a robust IDS for EVSEs is crucial to address cyberattacks like vulnerability scanning, network intrusions, and host-targeted attacks. Existing IDS solutions often focus on a single data source, such as network traffic [ 14 , 15 , 16 , 17 ] or kernel logs [ 18 ] , offering limited threat insight. Network traffic lacks host-level visibility, while kernel logs miss external attack patterns. Centralizing EVSE logs for global IDS training raises significant privacy concerns. To address these challenges, our framework employs a comprehensive approach by intelligently processing and fusing diverse log types generated by EVSEs during operation. This enables the IDS to correlate information across multiple data sources, uncovering complex attack patterns that cannot be detected by analyzing a single log type alone. • Multimodal Data Fusion : Our approach utilizes diverse data types generated by EVSEs, including network traffic and kernel events, to construct a comprehensive view of station activity. While this paper focuses on these two data sources, the framework’s design, leveraging latent representation extraction and compression, is inherently extensible to incorporate other types of data logs, such as power consumption and voltage metrics. • Privacy-Preserving Collaboration : Through Federated Learning (FL), the framework capitalizes on the diversity of logs collected by a large number of geographically distributed EVSEs, enabling collaborative model training without compromising data privacy. • Decentralized Security : This decentralized approach ensures high detection accuracy while preserving the confidentiality of user and operational data. By combining multimodal data fusion and federated learning, the framework significantly improves the detection of sophisticated attack patterns, providing a robust, scalable, and privacy-preserving cybersecurity solution for EVSE infrastructure. The remainder of this paper is organized as follows: Section 2 reviews background and literature review; Section 3 defines the problem and model framework; Section 4 introduces the proposed methodology; Section 5 presents experiments and results; and Section 6 concludes with future research directions. II Background and Related Work This section reviews EVSE foundations, focusing on key features, protocols, attack scenarios, and recent intrusion detection research. II-A EVSE Characteristics and Communication Protocols The Electric Vehicle Supply Equipment (EVSE) ecosystem depends on key communication protocols for interoperability, security, and efficiency. Two main protocols are central to this: the Open Charge Point Protocol (OCPP) and ISO 15118. OCPP, developed by the Open Charge Alliance, is a widely adopted protocol for managing communication between EV chargers and central systems. It facilitates features such as smart charging, remote diagnostics, and enhanced security, ensuring compatibility and centralized control across diverse EVSE networks [ 19 ] . On the other hand, ISO 15118, an international standard for EVs, governs communication between vehicles and charging stations. This protocol introduces Plug & Charge functionality, allowing automatic vehicle authentication, which improves user convenience and security. Additionally, ISO 15118 supports bidirectional energy transfer, enabling advanced vehicle-to-grid (V2G) services that enhance grid stability and energy efficiency [ 20 ] . Fig. 1 : Federated Training of the Intrusion Detection Model II-B Related Work In [ 6 ] , the authors proposed an IDS using ensemble learning for EVSE networks, targeting attack detection in both centralized and decentralized infrastructures. Their framework combines multiple models trained on diverse EVSE data to improve detection accuracy. However, relying on three classification sub-models may not suit all EVSEs, particularly those with limited computational resources. Researchers in [ 7 ] proposed an IDS for IoT-based EV charging stations to detect cyber threats, using a real IoT dataset with both binary and