The UTC Graduate School is pleased to announce that Mohamedelfateh Mohamedkhir will present Master's research titled, Analysis of Signal Resampling Effects on Attention-Driven SEI for IoT Systems on 10/10/2025 at 10 am - 12 pm in ECS 347. Everyone is invited to attend. 

Engineering

Chair: Donald R. Reising

Abstract:

The rapid expansion of the Internet of Things (IoT) has introduced billions of interconnected devices, many of which lack adequate security protections. Weak or absent safeguards make these devices attractive targets for malicious exploitation. To address this challenge, researchers have investigated Specific Emitter Identification (SEI) as a potential security mechanism. SEI is a passive technique that authenticates a device by its unique radio frequency signature, without requiring hardware modifications or alerting the device itself. Importantly, it is capable of distinguishing emitters down to individual serial numbers—one of the most demanding cases of identification—and can provide the “something you are” factor within zero-trust, multi-factor authentication frameworks. While SEI traditionally depends on high sampling rates to achieve strong identification accuracy, this requirement is difficult to satisfy in resource-constrained IoT environments. In this work, an attention-driven SEI process is assessed for its ability to distinguish emitters that differ only in serial numbers using signals collected at reduced sampling rates. Even with signals sampled at just 5 MHz and a dataset limited to 2,500 signals per emitter, the method consistently achieved identification accuracy above 97%, reducing storage demands by 87.5%. This performance highlights the practicality of SEI for real-world IoT applications where efficiency and scalability are critical.