권예린, 허준범,

Vol. 34, No. 5, pp. 1059-1071, 10월. 2024
10.13089/JKIISC.2024.34.5.1059, Full Text:
Keywords: 5G/LTE, Signaling Attacks, Detection, DoS, clustering
Abstract

The 5G mobile network provides various services to numerous devices and applications, unlike LTE which focuses on smartphones. Features of 5G, such as low latency and massive connectivity, increase the overhead of the control plane(CP, signaling part) and make it difficult to detect abnormal devices due to random traffic patterns. In this paper, we propose a DBSCAN clustering-based detection method to counter signaling attacks, which are a type of 'Denial of Service(DoS)' attack targeting mobile networks. DBSCAN helps to create clusters of various shapes and can address dynamic traffic because the algorithm needs not to depend on past traffic statistics. We also use a real-time traced dataset for experiments to assess usability in real-world scenarios. According to the experiments, our method achieves 99.32% of accuracy and 0.03% of false-positive rates, demonstrating superior performance compared to previous works.

Statistics
Cite this article