.png){kind=icon}
.png){kind=icon}
Return
Toward Channel-Robust RF Fingerprint Identification Using Spectrum Averaging and High-Order Difference
.png){kind=icon}
PRE
AI
L
J
DOI：10.1109/tifs.2026.3714140.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper proposes a novel radio frequency fingerprint identification (RFFI) system that eliminates the channel effects in the channel state information (CSI) measurements and extracts hardware-related features for wireless device identification, using Wi-Fi as a case study. Specifically, a spectrum averaging algorithm was designed to mitigate channel effects by exploiting the spatial diversity of a multi-antenna receiver. Furthermore, a high-order difference algorithm was developed to further suppress channel effects and extract RFF features by operating in the logarithmic domain, thereby preserving hardware-specific fingerprint characteristics. To validate the proposed system, we built a testbed comprising 12 commercial off-the-shelf Wi-Fi devices and a four-antenna USRP X310 software-defined radio (SDR) receiver. Extensive experiments were conducted under three practical scenarios with varying transmitter mobility and environmental dynamics. The experimental results demonstrated that the proposed RFFI system achieved an average classification accuracy of 90.28% across all training scenarios and outperformed state-of-the-art methods by 2.97%–9.73% in classification accuracy.
Keywords:
Radio frequency fingerprint identification
channel state information
channel-robust
Journal
IF:
8
Papers:
5.2K
Citations:
2.3W
Organization
Cited Papers