Scholarlink

Probe attacks are considered a first-stage threat because they typically involve an initial reconnaissance phase, during which attackers gather information about potential targets. Early detection of probe attacks is crucial, as it provides valuable insight into the intentions of malicious actors and the specific vulnerabilities they seek to exploit. This paper presents an Intrusion Detection System (IDS) based on supervised artificial intelligence to detect probe attacks in SDN environments. The results demonstrate the superior performance of our IDS, achieving 96.8% accuracy, a 3.2% false alarm rate (FAR), a training time of 158.3 s, and a testing time of 3.8 s using the KDD NSL dataset. Using the UNSW_NB15 dataset, the IDS achieves 96.1% accuracy, a 3.9% FAR, a training time of 142.6, and a testing time of 3.2 s.

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

ML-Driven Framework for Detecting Probing Attacks Against IoT Devices in SDN Environments

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Nader Karmous,
  • Wadii Jlassi,
  • Mohamed Ould-Elhassen Aoueileyine,
  • Ridha Bouallegue

摘要

Probe attacks are considered a first-stage threat because they typically involve an initial reconnaissance phase, during which attackers gather information about potential targets. Early detection of probe attacks is crucial, as it provides valuable insight into the intentions of malicious actors and the specific vulnerabilities they seek to exploit. This paper presents an Intrusion Detection System (IDS) based on supervised artificial intelligence to detect probe attacks in SDN environments. The results demonstrate the superior performance of our IDS, achieving 96.8% accuracy, a 3.2% false alarm rate (FAR), a training time of 158.3 s, and a testing time of 3.8 s using the KDD NSL dataset. Using the UNSW_NB15 dataset, the IDS achieves 96.1% accuracy, a 3.9% FAR, a training time of 142.6, and a testing time of 3.2 s.