Abstract
Radio frequency fingerprint identification (RFFI) provides a physical-layer credential for Internet of Things devices, but open-set decisions become fragile when a threshold calibrated on a source receiver is transferred to a target receiver. Receiver shift can lower the confidence of known transmitters and cause false rejection, while closed-set alignment can have the opposite effect by pulling unseen target transmitters into known regions, increasing false acceptance.
This letter presents CRODA-ST, a structure-first adaptation framework for single-source single-target cross-receiver open-set RFFI. Its two components target the bottlenecks behind unreliable source-calibrated rejection:
- Discriminative Structure Anchoring (DSA): Restores target-receiver known-class references from limited labeled target enrollment samples.
- Rejection-Oriented Alignment (ROA): Reduces receiver-sensitive confidence fluctuations around the anchored structure.
On the WiSig ManyTx dataset, CRODA-ST achieves 0.9092 known-class accuracy, 0.9692 AUROC, and 0.9580 OSCR. Score-sweep analysis further reduces FPR90 to 0.0469.
Blogger's Review: The CRODA-ST framework effectively addresses the challenges posed by receiver variability in open-set RFFI through its innovative structure-first strategy, showcasing significant results in enhancing identification accuracy and reducing false positives, highlighting its potential applications in IoT security.