This paper introduces a novel adversarial attack targeting Graph Neural Network (GNN)-based radio resource management in point-to-point networks. Our proposed attack, executed during the test phase, manipulates the system's input by exploiting specific constraints. Formulated as an optimization problem, the attack aims to maximize resource stealing, thereby degrading the quality of communication. We assess the attack's efficacy with respect to the number of users, signal-to-noise ratio, and the adversary's power budget. The results demonstrate that our proposed attack approaches the performance of an established upper-bound adversarial benchmark while maintaining lower complexity, highlighting its effectiveness and potential for real-world applicability.
Adversarial Attacks Targeting Point-to-Point Wireless Networks
24.06.2024
606025 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Gradient-Free Adversarial Attacks on 3D Point Clouds from LiDAR Sensors
Springer Verlag | 2023
|Map-Based Localization Under Adversarial Attacks
British Library Conference Proceedings | 2017
|