Calibrated RF-Fingerprinting Under Co-Channel Interference: Multi-Label 1D-CNN with Guarantees on False Negatives

Forum topic · 小凯 · 2026-09-20

Summary

RF-fingerprinting identifies wireless transmitters from hardware impairments embedded in their emitted signals, but most prior work assumes only one transmitter is active at a time, limiting real-world use. This paper extends RF-fingerprinting to co-channel interference, where multiple signals overlap in both time and frequency. The authors formulate the task as multi-label classification using a 1D convolutional neural network (CNN) and calibrate the model to derive confidence thresholds on label probabilities, providing guarantees on an upper bound for the average number of false negatives—ensuring reliable detection of real spectrum violations. The method is validated on real data collected from the POWDER 5G testbed involving Wi-Fi (802.11a), 4G LTE, and 5G NR waveforms. Calibrated accuracy ranges from 73% to 97% depending on channel conditions; calibration for different average false-negative upper bounds drives micro recall toward (1 - calibrated false-negative rate), and the approach shows robustness to out-of-distribution interference. (arXiv: 2609.20765, authors: Tariq Abdul-Quddoos, Xiangfang Li, Lijun Qian)

Overview

Field: Machine Learning
Authors: Tariq Abdul-Quddoos, Xiangfang Li, Lijun Qian
Published: 2026-09-17
arXiv: 2609.20765

Summary (translated)

Radio Frequency (RF)-fingerprinting is a spectrum monitoring technique that identifies specific transmitters based on hardware impairments imprinted within the emitted signal. Although widely researched, prior studies almost exclusively consider scenarios where only one transmitter is emitting at a time, limiting real-world applicability. This work extends the study of RF-fingerprinting to co-channel interference, with multiple emitted signals interfering with each other and overlapping in time and frequency. Specifically, the problem is formulated as multi-label classification and solved with a 1D convolutional neural network (CNN). The models are further calibrated so that confidence thresholds for the label probabilities are derived, with guarantees on an upper bound for the average number of false negatives, ensuring trustworthy detection when real spectrum violations occur.

Original Abstract (excerpt)

> Radio Frequency (RF)-Fingerprinting is a spectrum monitoring technique that identifies specific transmitters based on hardware impairments imprinted within the emitted signal. Although widely researched, studies almost exclusively consider scenarios where only one transmitter is emitting at a time, limiting real world applicability. In this work, we further the study of RF-Fingerprinting by considering co-channel interference, with multiple emitted signals interfering with each other, overlapping in time and frequency. Specifically, we formulate this problem as a multi-label classification problem and employ a 1D convolutional neural network (CNN). Furthermore, the models are calibrated such that the confidence thresholds for the label probabilities are derived, with guarantees on the upper...

Key Results

Validated on real data from the POWDER 5G testbed with Wi-Fi (802.11a), 4G LTE, and 5G NR waveforms.
Calibrated accuracy ranges from 73% to 97% depending on channel conditions.
Calibration tuned to different average false-negative upper bounds drives micro recall toward (1 - calibrated false-negative rate).
The calibration is robust to out-of-distribution interference, demonstrating potential for highly contested wireless environments.

---

*Auto-collected on 2026-09-20.*