Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cyber Attacks Prevention Towards Prosumer-based EV Charging Stations: An Edge-assisted Federated Prototype Knowledge Distillation Approach

Luyao Zou, Quang Hieu Vo, Kitae Kim, Huy Q. Le, Chu Myaet Thwal, Chaoning Zhang, Choong Seon Hong

TL;DR

This work targets cyber-attack prevention for prosumer-based EV charging stations by introducing an edge-assisted federated prototype knowledge distillation (E-FPKD) framework. It Combines feature selection via Pearson Correlation Coefficient, availability-aware federated learning, a teacher–student KD setup per client, and prototype aggregation to address non-IID NT data, followed by a rule-based intervention at edge servers. The approach demonstrates improved overall detection correctness (ODC) and accuracy on NSL-KDD, UNSW-NB15, and IoTID20, outperforming several baselines and maintaining data privacy through on-site training and prototype sharing. The proposed method offers scalable, privacy-preserving cyber-attack detection with practical intervention capabilities suitable for deployment in edge-enabled EV charging ecosystems.

Abstract

In this paper, cyber-attack prevention for the prosumer-based electric vehicle (EV) charging stations (EVCSs) is investigated, which covers two aspects: 1) cyber-attack detection on prosumers' network traffic (NT) data, and 2) cyber-attack intervention. To establish an effective prevention mechanism, several challenges need to be tackled, for instance, the NT data per prosumer may be non-independent and identically distributed (non-IID), and the boundary between benign and malicious traffic becomes blurred. To this end, we propose an edge-assisted federated prototype knowledge distillation (E-FPKD) approach, where each client is deployed on a dedicated local edge server (DLES) and can report its availability for joining the federated learning (FL) process. Prior to the E-FPKD approach, to enhance accuracy, the Pearson Correlation Coefficient is adopted for feature selection. Regarding the proposed E-FPKD approach, we integrate the knowledge distillation and prototype aggregation technique into FL to deal with the non-IID challenge. To address the boundary issue, instead of directly calculating the distance between benign and malicious traffic, we consider maximizing the overall detection correctness of all prosumers (ODC), which can mitigate the computational cost compared with the former way. After detection, a rule-based method will be triggered at each DLES for cyber-attack intervention. Experimental analysis demonstrates that the proposed E-FPKD can achieve the largest ODC on NSL-KDD, UNSW-NB15, and IoTID20 datasets in both binary and multi-class classification, compared with baselines. For instance, the ODC for IoTID20 obtained via the proposed method is separately 0.3782% and 4.4471% greater than FedProto and FedAU in multi-class classification.

Cyber Attacks Prevention Towards Prosumer-based EV Charging Stations: An Edge-assisted Federated Prototype Knowledge Distillation Approach

TL;DR

This work targets cyber-attack prevention for prosumer-based EV charging stations by introducing an edge-assisted federated prototype knowledge distillation (E-FPKD) framework. It Combines feature selection via Pearson Correlation Coefficient, availability-aware federated learning, a teacher–student KD setup per client, and prototype aggregation to address non-IID NT data, followed by a rule-based intervention at edge servers. The approach demonstrates improved overall detection correctness (ODC) and accuracy on NSL-KDD, UNSW-NB15, and IoTID20, outperforming several baselines and maintaining data privacy through on-site training and prototype sharing. The proposed method offers scalable, privacy-preserving cyber-attack detection with practical intervention capabilities suitable for deployment in edge-enabled EV charging ecosystems.

Abstract

In this paper, cyber-attack prevention for the prosumer-based electric vehicle (EV) charging stations (EVCSs) is investigated, which covers two aspects: 1) cyber-attack detection on prosumers' network traffic (NT) data, and 2) cyber-attack intervention. To establish an effective prevention mechanism, several challenges need to be tackled, for instance, the NT data per prosumer may be non-independent and identically distributed (non-IID), and the boundary between benign and malicious traffic becomes blurred. To this end, we propose an edge-assisted federated prototype knowledge distillation (E-FPKD) approach, where each client is deployed on a dedicated local edge server (DLES) and can report its availability for joining the federated learning (FL) process. Prior to the E-FPKD approach, to enhance accuracy, the Pearson Correlation Coefficient is adopted for feature selection. Regarding the proposed E-FPKD approach, we integrate the knowledge distillation and prototype aggregation technique into FL to deal with the non-IID challenge. To address the boundary issue, instead of directly calculating the distance between benign and malicious traffic, we consider maximizing the overall detection correctness of all prosumers (ODC), which can mitigate the computational cost compared with the former way. After detection, a rule-based method will be triggered at each DLES for cyber-attack intervention. Experimental analysis demonstrates that the proposed E-FPKD can achieve the largest ODC on NSL-KDD, UNSW-NB15, and IoTID20 datasets in both binary and multi-class classification, compared with baselines. For instance, the ODC for IoTID20 obtained via the proposed method is separately 0.3782% and 4.4471% greater than FedProto and FedAU in multi-class classification.

Paper Structure

This paper contains 39 sections, 2 theorems, 29 equations, 15 figures, 14 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Cyber-attack Detection for EV Charging Station
  4. Cyber-attack Intervention/Prevention
  5. Knowledge Distillation (KD) Technique in FL
  6. Prototype Aggregation (PA) Technique
  7. Nomenclature
  8. System Model and Problem Formulation
  9. Cyber-attack Detection Model via Deep Learning
  10. Malicious and Benign Traffic Boundary Model
  11. Problem Formulation
  12. Preliminaries Knowledge
  13. Background: Federated Learning (FL)
  14. Background: Knowledge Distillation (KD)
  15. Background: Prototype Aggregation (PA)
  16. ...and 24 more sections

Key Result

Lemma 1

The computational cost of eq. Opt_2 is high, if there exists many network traffic data per prosumer and/or high-dimensional feature space.

Figures (15)

  • Figure 1: System model of the prosumer-based EV charging stations.
  • Figure 2: Training process of federated learning approach.
  • Figure 3: Structure of the proposed solution for cyber-attack prevention towards the considered prosumer-based EVCSs.
  • Figure 4: Overall architecture of the proposed E-FPKD approach towards cyber-attacks prevention in both detection and intervention.
  • Figure 5: KD applied to per client in the proposed E-FPKD approach.
  • ...and 10 more figures

Theorems & Definitions (3)

  • Lemma 1
  • Lemma 2
  • Remark 1