Examining Machine Learning for 5G and Beyond through an Adversarial Lens
It addresses security concerns for mobile network operators and researchers, but is incremental as it synthesizes existing adversarial ML concepts into the 5G domain.
The paper examines the adversarial risks of using AI/ML in 5G networks, highlighting vulnerabilities across supervised, unsupervised, and reinforcement learning methods through case studies, and proposes mitigation strategies and guidelines for robustness evaluation.
Spurred by the recent advances in deep learning to harness rich information hidden in large volumes of data and to tackle problems that are hard to model/solve (e.g., resource allocation problems), there is currently tremendous excitement in the mobile networks domain around the transformative potential of data-driven AI/ML based network automation, control and analytics for 5G and beyond. In this article, we present a cautionary perspective on the use of AI/ML in the 5G context by highlighting the adversarial dimension spanning multiple types of ML (supervised/unsupervised/RL) and support this through three case studies. We also discuss approaches to mitigate this adversarial ML risk, offer guidelines for evaluating the robustness of ML models, and call attention to issues surrounding ML oriented research in 5G more generally.