Gotta Detect 'Em All: Fake Base Station and Multi-Step Attack Detection in Cellular Networks

Kazi Samin Mubasshir, Imtiaz Karim, Elisa Bertino

34th USENIX Security Symposium · Day 1

Cellular networks, with billions of connected devices globally, are critical infrastructure and, consequently, prime targets for malicious actors. A significant and persistent threat in this landscape is the proliferation of **Fake Base Stations (FBSes)**, also known as false base stations or rogue base stations. These unauthorized devices impersonate legitimate cellular towers, luring unsuspecting user equipment (UEs) to connect to them, thereby enabling a range of sophisticated **Multi-Step Attacks (MSAs)**. This paper introduces **FBSDetector**, a novel, effective, and efficient solution designed to detect FBSes and MSAs directly from Layer-3 network traces at the UE side, leveraging advanced machine learning (ML) techniques.

AI review

Solid systems security work that actually ships something usable. The POWDER-based dataset generation is the real contribution here—first large-scale FBS/MSA dataset that isn't synthetic garbage. Detection numbers are good but not earth-shattering; the graph-learning MSA approach is clever but the 86% accuracy with 3.28% FPR means you're still going to annoy users in the real world.