Scholarlink

<p>Maritime radar security issues include protecting radar systems on vessels against cyber threats. This is essential for preserving the integrity and reliability of navigation and collision avoidance systems, which are vital for ensuring safe maritime operations. It is essential to protect these systems against attacks since they are becoming more digitalized and interconnected. This research presents a novel strategy, Marine Radar Security with Personalized Federated Learning-based Intrusion Detection System (MRS-PFIDS), to increase maritime radar security. The proposed approach attempts to detect and classify various cyberattacks while protecting data privacy. The method includes preprocessing the dataset, standardizing numerical characteristics, and building a Convolutional Neural Network (CNN) within the Personalized federated learning framework utilizing TensorFlow Federated (TFF). Furthermore, the research included a non-IID data setting, where client datasets comprised diverse distributions of attack types, attempting real-world scenarios. The federated CNN model generates a classification accuracy of approximately 98.36% in IID settings and maintains continuously high accuracy in the non-IID setting, with client-specific accuracies ranging from 98.53% to 99.64%. These findings underscore the effectiveness of MRS-PFIDS in protecting marine radar systems from sophisticated cyber threats, highlighting its potential to advance maritime cybersecurity while balancing the need for data privacy, adaptability to heterogeneous environments, and advanced detection of attacks.</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

MRS-PFIDS: federated learning driven detection of network intrusions in maritime radar systems

  • 期刊论文

  • 发表日期：

    2025年3月10日

查看全文
  • Md. Alamgir Hossain,
  • Md Delwar Hossain,
  • Roya Choupani,
  • Erdoǧan Doǧdu

摘要

Maritime radar security issues include protecting radar systems on vessels against cyber threats. This is essential for preserving the integrity and reliability of navigation and collision avoidance systems, which are vital for ensuring safe maritime operations. It is essential to protect these systems against attacks since they are becoming more digitalized and interconnected. This research presents a novel strategy, Marine Radar Security with Personalized Federated Learning-based Intrusion Detection System (MRS-PFIDS), to increase maritime radar security. The proposed approach attempts to detect and classify various cyberattacks while protecting data privacy. The method includes preprocessing the dataset, standardizing numerical characteristics, and building a Convolutional Neural Network (CNN) within the Personalized federated learning framework utilizing TensorFlow Federated (TFF). Furthermore, the research included a non-IID data setting, where client datasets comprised diverse distributions of attack types, attempting real-world scenarios. The federated CNN model generates a classification accuracy of approximately 98.36% in IID settings and maintains continuously high accuracy in the non-IID setting, with client-specific accuracies ranging from 98.53% to 99.64%. These findings underscore the effectiveness of MRS-PFIDS in protecting marine radar systems from sophisticated cyber threats, highlighting its potential to advance maritime cybersecurity while balancing the need for data privacy, adaptability to heterogeneous environments, and advanced detection of attacks.