Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Kolmogorov-Arnold Network for Explainable Detection of Cyberattacks on EV Chargers

Ahmad Mohammad Saber, Max Mauro Dias Santos, Mohammad Al Janaideh, Amr Youssef, Deepa Kundur

TL;DR

The paper addresses cyberattack detection for EV charging infrastructure using only locally measured power signals. It introduces a Kolmogorov-Arnold Network (KAN) that decomposes multivariate inputs into sums of univariate functions, enabling deep, interpretable models suitable for edge deployment. Trained offline on a large, labeled dataset, the KAN achieves high precision ($0.99$) and strong F1-score ($0.92$) in distinguishing normal from malicious charging, and can extract symbolic formulas to explain its decisions with competitive, though slightly reduced, accuracy ($87.28\%$) via symbolic regression. The approach demonstrates edge-ready, scalable performance with fast inference ($\approx12.5\text{ ms}$ per sample) and offers a practical, explainable solution to securing EV charging infrastructure against cyber threats.

Abstract

The increasing adoption of Electric Vehicles (EVs) and the expansion of charging infrastructure and their reliance on communication expose Electric Vehicle Supply Equipment (EVSE) to cyberattacks. This paper presents a novel Kolmogorov-Arnold Network (KAN)-based framework for detecting cyberattacks on EV chargers using only power consumption measurements. Leveraging the KAN's capability to model nonlinear, high-dimensional functions and its inherently interpretable architecture, the framework effectively differentiates between normal and malicious charging scenarios. The model is trained offline on a comprehensive dataset containing over 100,000 cyberattack cases generated through an experimental setup. Once trained, the KAN model can be deployed within individual chargers for real-time detection of abnormal charging behaviors indicative of cyberattacks. Our results demonstrate that the proposed KAN-based approach can accurately detect cyberattacks on EV chargers with Precision and F1-score of 99% and 92%, respectively, outperforming existing detection methods. Additionally, the proposed KANs's enable the extraction of mathematical formulas representing KAN's detection decisions, addressing interpretability, a key challenge in deep learning-based cybersecurity frameworks. This work marks a significant step toward building secure and explainable EV charging infrastructure.

A Kolmogorov-Arnold Network for Explainable Detection of Cyberattacks on EV Chargers

TL;DR

The paper addresses cyberattack detection for EV charging infrastructure using only locally measured power signals. It introduces a Kolmogorov-Arnold Network (KAN) that decomposes multivariate inputs into sums of univariate functions, enabling deep, interpretable models suitable for edge deployment. Trained offline on a large, labeled dataset, the KAN achieves high precision () and strong F1-score () in distinguishing normal from malicious charging, and can extract symbolic formulas to explain its decisions with competitive, though slightly reduced, accuracy () via symbolic regression. The approach demonstrates edge-ready, scalable performance with fast inference ( per sample) and offers a practical, explainable solution to securing EV charging infrastructure against cyber threats.

Abstract

The increasing adoption of Electric Vehicles (EVs) and the expansion of charging infrastructure and their reliance on communication expose Electric Vehicle Supply Equipment (EVSE) to cyberattacks. This paper presents a novel Kolmogorov-Arnold Network (KAN)-based framework for detecting cyberattacks on EV chargers using only power consumption measurements. Leveraging the KAN's capability to model nonlinear, high-dimensional functions and its inherently interpretable architecture, the framework effectively differentiates between normal and malicious charging scenarios. The model is trained offline on a comprehensive dataset containing over 100,000 cyberattack cases generated through an experimental setup. Once trained, the KAN model can be deployed within individual chargers for real-time detection of abnormal charging behaviors indicative of cyberattacks. Our results demonstrate that the proposed KAN-based approach can accurately detect cyberattacks on EV chargers with Precision and F1-score of 99% and 92%, respectively, outperforming existing detection methods. Additionally, the proposed KANs's enable the extraction of mathematical formulas representing KAN's detection decisions, addressing interpretability, a key challenge in deep learning-based cybersecurity frameworks. This work marks a significant step toward building secure and explainable EV charging infrastructure.

Paper Structure

This paper contains 13 sections, 7 equations, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Preliminaries and Threat Model
  3. Electric Vehicle Charging Systems
  4. Threat Model
  5. A KAN-based Framework for Detection of Cyberattacks on EV Chargers
  6. Performance Evaluation Results
  7. Perfromance Evaluation Scenarios and Dataset
  8. Training of the KAN-based Cyberattack Detection Model
  9. Testing Results
  10. Accuracy of Trained KAN Model
  11. Accuracy of KAN Model's Extracted Symbolic Formulas
  12. Scalability and Computational Efficiency
  13. Conclusion and Future Work

Figures (5)

  • Figure 1: The proposed KAN-based framework for detecting cyberattacks against EVSE.
  • Figure 2: High-level idea of KANs with activation functions on edges instead of nodes.
  • Figure 3: Scatter plot of the dataset after applying PCA transformation, blue and red dots denote benign and malicious cases, respectively
  • Figure 4: An illustration of utilized KAN model's initial architecture with 5 hidden neurons, a cubic spline, and 3 grid intervals
  • Figure 5: Proposed KAN's confusion matrix in percentages.