A cybersecurity researcher introduces their work on black-box adversarial attacks and anomaly detection at the 6G physical layer using Sionna PHY

Hi everyone,

I’m excited to join the NVIDIA 6G Developer Program community. My name is Kabiru M. Salawu, and I’m based in Nigeria. My background is in cybersecurity, and I recently joined this program to explore how that expertise intersects with 6G research — an area I believe is critically underserved.

My current project focuses on AI/ML security at the physical layer. Specifically, I’m investigating black-box evasion attacks on neural channel estimators in OFDM-based 6G systems, using adversarial perturbations injected via pilot contamination. Alongside the attack, I’m building an anomaly detection layer as a countermeasure — to produce a paired attack-defense contribution that is both practically relevant and publishable.

I’m working with Sionna PHY (v2.0) as my primary simulation environment, and I plan to use the Aerial Omniverse Digital Twin for sandboxed attack testing in later stages.

As someone newer to the ML/communications side of things (my strength is on the security and threat modeling end), I’m looking forward to learning from this community and contributing a cybersecurity perspective that I hope adds value to the broader 6G research conversation here.

Happy to connect with anyone working on related topics — particularly around AI/ML robustness, O-RAN security, or federated learning vulnerabilities in 6G.

Looking forward to being part of this community!

Best,
Kabiru M. Salwu