Scholarlink

<p>Electric vehicle charging stations (EVCS) are becoming more and more common, so it is imperative to protect these systems from cyberattacks, especially Distributed Denial-of-Service (DDoS) assaults. The objective of this study is to enhance the model interpretability and detection accuracy of DDoS attacks in EVCS using the Personalized Federated Learning (PFL) technique. The research makes use of an IoT attack dataset with 33 attacks that were carried out over 105 devices in a topology that was divided into seven different categories. Using the Firefly Algorithm, the suggested PFL method selects a subset of features wisely to maximize the performance of the classification model. Promising outcomes are seen in the evaluation of several machine learning models, such as Random Forest, Gradient Boosting Machine (GBM), K-Nearest Neighbors, and Multilayer Perceptron. GBM and Random Forest demonstrate their promise for efficient DDoS detection in EVCS by achieving high accuracy rates of 99% and 98%, respectively, in detecting DDoS attacks. The overall detection performance is further improved by the feature selection model, which also increases the efficiency and interpretability of the classification model. These results imply that machine learning models can improve the security and resilience of EVCS against DDoS attacks when combined with the PFL technique. </p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fortifying EV charging stations: AI-powered detection and mitigation of DDoS attacks using personalized Federated learning

  • 期刊论文

  • 发表日期：

    2025年7月30日

查看全文
  • Fatma M. Talaat,
  • Mohamed Mohsen Elsaid Khoudier,
  • Ibrahim F. Moawad,
  • Amir El-Ghamry

摘要

Electric vehicle charging stations (EVCS) are becoming more and more common, so it is imperative to protect these systems from cyberattacks, especially Distributed Denial-of-Service (DDoS) assaults. The objective of this study is to enhance the model interpretability and detection accuracy of DDoS attacks in EVCS using the Personalized Federated Learning (PFL) technique. The research makes use of an IoT attack dataset with 33 attacks that were carried out over 105 devices in a topology that was divided into seven different categories. Using the Firefly Algorithm, the suggested PFL method selects a subset of features wisely to maximize the performance of the classification model. Promising outcomes are seen in the evaluation of several machine learning models, such as Random Forest, Gradient Boosting Machine (GBM), K-Nearest Neighbors, and Multilayer Perceptron. GBM and Random Forest demonstrate their promise for efficient DDoS detection in EVCS by achieving high accuracy rates of 99% and 98%, respectively, in detecting DDoS attacks. The overall detection performance is further improved by the feature selection model, which also increases the efficiency and interpretability of the classification model. These results imply that machine learning models can improve the security and resilience of EVCS against DDoS attacks when combined with the PFL technique.