Scholarlink

<p>IoT environments face significant security challenges due to their distributed, heterogeneous, and dynamic nature. Traditional security systems have been proven inadequate in addressing these complexities, prompting the exploration of advanced approaches like CTI and ML. However, their integration into IoT security is still underexplored. This paper provides a comprehensive study on applying CTI and ML to detect and mitigate cyber threats in IoT networks. The study reviews recent approaches, classifying them into technical subdomains and highlighting the most effective ML techniques, such as neural networks, federated learning, and hybrid algorithms. The findings demonstrate that while ML has proven to be an interesting tool for identifying anomalies and sophisticated attacks, its integration with CTI platforms remains limited, hindering collaboration and the ability to anticipate threats. Furthermore, the analysis identified significant gaps, including the need for improved model explainability and the implementation of scalable, energy-efficient solutions. This work also discusses the challenges of creating integrated frameworks that combine the adaptability of ML with CTI, ultimately aiming to improve cybersecurity for IoT through more robust and proactive solutions.</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

Toward using cyber threat intelligence with machine and deep learning for IoT security: a comprehensive study

  • 期刊论文

  • 发表日期：

    2025年10月3日

查看全文
  • Milton Lima,
  • Carlos Viana,
  • Wellison R. M. Santos,
  • Flávio Neves,
  • João R. Campos,
  • Fernando Aires

摘要

IoT environments face significant security challenges due to their distributed, heterogeneous, and dynamic nature. Traditional security systems have been proven inadequate in addressing these complexities, prompting the exploration of advanced approaches like CTI and ML. However, their integration into IoT security is still underexplored. This paper provides a comprehensive study on applying CTI and ML to detect and mitigate cyber threats in IoT networks. The study reviews recent approaches, classifying them into technical subdomains and highlighting the most effective ML techniques, such as neural networks, federated learning, and hybrid algorithms. The findings demonstrate that while ML has proven to be an interesting tool for identifying anomalies and sophisticated attacks, its integration with CTI platforms remains limited, hindering collaboration and the ability to anticipate threats. Furthermore, the analysis identified significant gaps, including the need for improved model explainability and the implementation of scalable, energy-efficient solutions. This work also discusses the challenges of creating integrated frameworks that combine the adaptability of ML with CTI, ultimately aiming to improve cybersecurity for IoT through more robust and proactive solutions.