Scholarlink

<p>The rapid adoption of Internet of Things (IoT)-enabled Electric Vehicles (EVs) has significantly enhanced intelligent transportation and energy management systems. It has also exposed EV infrastructures to sophisticated cyberattacks, such as Denial of Service (DoS) and False Data Injection (FDI). These attacks can severely affect vehicle stability, state estimation accuracy, battery management, and smart grid reliability. Baseline deep learning-based cyber-defense techniques often suffer from high computational complexity, less adaptability to nonlinear attack patterns, and increased latency during real-time edge deployment. Therefore, it aims to develop an efficient and resilient framework for accurate state estimation and robust cyber-attack mitigation in IoT-enabled EV systems operating under dynamic adversarial conditions. Here, a novel hybrid framework integrating Convolutional Kolmogorov-Arnold Networks (CKAN) with Lotus Effect Optimization (LEA) (LEA-CKAN) is proposed. CKAN model employs learnable B-spline-based nonlinear mappings to identify complex attack signatures, while LEA optimizes battery charge–discharge behavior and improves system stability by repelling malicious data perturbations via swarm-intelligence-based optimization. Experimental evaluation was simulated via the CICEVSE2024 dataset under multiple cyber-attack scenarios, including FDI, DoS, and host-based attacks. The proposed LEA-CKAN model attained an increased F1-score of 98.5%, maintained Mean Squared Error (MSE) below 10, and minimized processing latency to 4.2&#xa0;ms per sample, outperforming conventional CNN, EVNH, and CPDN-based approaches. Outcomes confirm that the proposed framework gives a scalable, minimized-latency, and robust cyber-defense solution for secure real-time IoT-enabled EV operation and future intelligent transportation systems.</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

A Novel Approach to Enhanced State Estimation and Robust Control in IoT-Enabled Electric Vehicle under Cyber Attacks

  • 期刊论文

  • 发表日期：

    2026年7月7日

查看全文
  • G. Kannayeram,
  • S. Udaiyakumar,
  • G. Venkatesan,
  • R. Saravanan

摘要

The rapid adoption of Internet of Things (IoT)-enabled Electric Vehicles (EVs) has significantly enhanced intelligent transportation and energy management systems. It has also exposed EV infrastructures to sophisticated cyberattacks, such as Denial of Service (DoS) and False Data Injection (FDI). These attacks can severely affect vehicle stability, state estimation accuracy, battery management, and smart grid reliability. Baseline deep learning-based cyber-defense techniques often suffer from high computational complexity, less adaptability to nonlinear attack patterns, and increased latency during real-time edge deployment. Therefore, it aims to develop an efficient and resilient framework for accurate state estimation and robust cyber-attack mitigation in IoT-enabled EV systems operating under dynamic adversarial conditions. Here, a novel hybrid framework integrating Convolutional Kolmogorov-Arnold Networks (CKAN) with Lotus Effect Optimization (LEA) (LEA-CKAN) is proposed. CKAN model employs learnable B-spline-based nonlinear mappings to identify complex attack signatures, while LEA optimizes battery charge–discharge behavior and improves system stability by repelling malicious data perturbations via swarm-intelligence-based optimization. Experimental evaluation was simulated via the CICEVSE2024 dataset under multiple cyber-attack scenarios, including FDI, DoS, and host-based attacks. The proposed LEA-CKAN model attained an increased F1-score of 98.5%, maintained Mean Squared Error (MSE) below 10, and minimized processing latency to 4.2 ms per sample, outperforming conventional CNN, EVNH, and CPDN-based approaches. Outcomes confirm that the proposed framework gives a scalable, minimized-latency, and robust cyber-defense solution for secure real-time IoT-enabled EV operation and future intelligent transportation systems.