Scholarlink

This study addresses the security vulnerabilities inherent in the SS7 protocol, with a specific focus on mitigating Man-in-the-Middle (MITM) attacks. Despite technological advancements, persistent security issues in Signaling System No. 7 (SS7) underscore the need to enhance user protection and minimize risks. The primary objective is to leverage Explainable Artificial Intelligence (XAI) to make AI decisions in telecommunications more transparent and justifiable. This research employs advanced machine learning algorithms, including Random Forest, Autoencoders, and K-Means Clustering, integrated with SHapley Additive exPlanations (SHAP) to enhance the interpretability of AI models. Given the limited availability of specific SS7 datasets, the study extrapolates data from existing scholarly articles. The research methodology involves systematic data collection and preprocessing, followed by the implementation and optimization of algorithms to effectively detect and analyze vulnerabilities. The integration of XAI with robust analytical tools aims to make the machine learning detection process transparent, thus improving the security of the SS7 protocol. This approach is crucial to identify potential attack vectors and reduce associated risks. The results demonstrate high precision, with the Random Forest algorithm achieving 94% accuracy, the autoencoders showing a low loss around ±0.10, and K-Means Clustering achieving a high Silhouette Score of 0.999. Furthermore, SHAP values provide information on the distinctions and similarities between the algorithms.

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

Transparency in Detecting Man-in-the-Middle Attacks on SS7 Networks Using SHAP Explainable AI

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Fauzan Rizqi Muhammad,
  • Parman Sukarno,
  • Aulia Arif Wardana

摘要

This study addresses the security vulnerabilities inherent in the SS7 protocol, with a specific focus on mitigating Man-in-the-Middle (MITM) attacks. Despite technological advancements, persistent security issues in Signaling System No. 7 (SS7) underscore the need to enhance user protection and minimize risks. The primary objective is to leverage Explainable Artificial Intelligence (XAI) to make AI decisions in telecommunications more transparent and justifiable. This research employs advanced machine learning algorithms, including Random Forest, Autoencoders, and K-Means Clustering, integrated with SHapley Additive exPlanations (SHAP) to enhance the interpretability of AI models. Given the limited availability of specific SS7 datasets, the study extrapolates data from existing scholarly articles. The research methodology involves systematic data collection and preprocessing, followed by the implementation and optimization of algorithms to effectively detect and analyze vulnerabilities. The integration of XAI with robust analytical tools aims to make the machine learning detection process transparent, thus improving the security of the SS7 protocol. This approach is crucial to identify potential attack vectors and reduce associated risks. The results demonstrate high precision, with the Random Forest algorithm achieving 94% accuracy, the autoencoders showing a low loss around ±0.10, and K-Means Clustering achieving a high Silhouette Score of 0.999. Furthermore, SHAP values provide information on the distinctions and similarities between the algorithms.