Scholarlink

This article presents a comprehensive and scalable system for the detection and mitigation of Rogue Access Points (APs) in Wi-Fi networks, addressing a significant security risk in distributed infrastructures. The proposed solution integrates multiple analytical layers, including passive and active traffic analysis, anomaly detection (Isolation Forest, One-Class SVM), supervised machine learning (Random Forest, XGBoost), and explainable artificial intelligence (XAI) mechanisms. A local whitelist is used for initial verification, with unknown devices triggering advanced analysis. Detected threats generate alerts via Snort and are logged and visualized in real time using OpenSearch and a lightweight browser plugin. Unlike previous approaches, this system combines detection with automated response, real-time stream analytics, and interpretable decision-making within a unified architecture. Experimental results in real-world conditions demonstrate over 95% detection accuracy and strong resilience to false positives. The solution shows high potential for deployment in high-risk, dynamic wireless network environments.

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

Anomaly-Based Detection of Rogue Access Points in High-Risk Network Infrastructures

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Łukasz Lemieszewski,
  • Grzegorz Remiszewski,
  • Piotr Borkowski,
  • Aleksander Wolak,
  • Janusz Szymczyk,
  • Jarosław Becker

摘要

This article presents a comprehensive and scalable system for the detection and mitigation of Rogue Access Points (APs) in Wi-Fi networks, addressing a significant security risk in distributed infrastructures. The proposed solution integrates multiple analytical layers, including passive and active traffic analysis, anomaly detection (Isolation Forest, One-Class SVM), supervised machine learning (Random Forest, XGBoost), and explainable artificial intelligence (XAI) mechanisms. A local whitelist is used for initial verification, with unknown devices triggering advanced analysis. Detected threats generate alerts via Snort and are logged and visualized in real time using OpenSearch and a lightweight browser plugin. Unlike previous approaches, this system combines detection with automated response, real-time stream analytics, and interpretable decision-making within a unified architecture. Experimental results in real-world conditions demonstrate over 95% detection accuracy and strong resilience to false positives. The solution shows high potential for deployment in high-risk, dynamic wireless network environments.