Robust RF Signatures
- Our goal is to learn RF signatures that can distinguish between wireless devices sending exactly the same message, based on subtle imperfections unique to each device.
- Since the information in RF data resides in complex baseband, we employ CNNs with complex-valued weights to learn these signatures. We demonstrate its effectiveness for two wireless protocols - WiFi and ADS-B.
-
We show major pitfalls when data is collected over multiple days and locations, due to nuisance parameters such as the clock drift and variations in the wireless channel.
-
We develop augmentation strategies based on signal models for these effects, and show that they are essential for learning robust signatures.
Publications
-
M. Cekic, S. Gopalakrishnan, U. Madhow, “ Robust Wireless Fingerprinting: Generalizing Across Space and Time”, arXiv:2002.10791.
-
S. Gopalakrishnan, M. Cekic, U. Madhow, “ Robust Wireless Fingerprinting via Complex- Valued Neural Networks", in IEEE Global Communications Conference (Globecom), Waikoloa, 2019.