1.School of Electrical and Information Engineering,Zhengzhou University,Zhengzhou 450001,Henan,China
2.Henan Key Laboratory of Network Cryptography Technology,Zhengzhou 450001,Henan,China
3.State Key Laboratory of Space-Ground Integrated Information Technology,Beijing 100086,China
Show less
文章历史+
Received
Published
2022-10-12
2023-10-24
Issue Date
2026-07-23
PDF (2368K)
摘要
目前关于射频指纹识别(radio frequency fingerprint identification,RFFI)的研究大多基于单个信号特征,存在识别准确率不高的问题。为此,提出了一种多特征融合多任务的射频指纹识别方法。该方法融合功率谱、基于STFT变换的时频谱、互功率谱三种信号特征,采用多任务学习(multi-task learning,MTL)策略,使用噪声信息作为先验知识来帮助网络训练,以设备分类为主任务,以信号噪声含量的分类作为网络第二个任务。仿真实验表明,本文提出的多特征融合多任务的方法较单特征单任务的方法有所提高,是一种有效的射频指纹识别方法。
Abstract
Most of the current research on Radio Frequency Fingerprint Identification (RFFI) is based on individual’s single signal feature, which has the problem of low identification accuracy. To this end, this paper proposes a multi-feature fusion multi-task RF fingerprint identification method, which fuses three signals: power spectrum, time-frequency spectrum based on STFT transform, and mutual power spectrum, and uses noise information as a priori knowledge to help network training, with device classification as the main task and classification of signal noise content as the second task of the network for multi-task learning (MTL). Simulation experiments show that the multi-feature fusion multi-task approach proposed in this paper is improved over the single-feature single-task approach and is an effective method for RF fingerprint identification.
射频指纹(radio frequency fingerprint,RFF)是设备在制造的过程中由于不可避免的电子器件细微误差所产生的,是一种不随信号传输场景变化而消失的信号特征信息。此类特征信息小且唯一,如同人类指纹一般,是唯一对应且不可复制的。射频指纹识别(radio frequency fingerprint identification,RFFI)是一种新兴的物联网安全技术,它通过无线设备内部硬件制造产生的唯一不可篡改的误差来实现设备的识别认证。射频指纹技术具有不会对设备造成多余开销、轻量级的特点,可以作为物联网设备接入认证的理想技术。
现有的射频指纹识别研究主要可分为基于特征工程的方法和基于深度学习的方法。基于特征工程的方法需要人工提取信号调制信息[4,5]、变换域信息、基本参数信息[6]、图像信息等统计信息或物理参数作为特征,再通过分类器进行设备识别。这种方法高度依赖特征提取算法,对信号处理方面的知识和经验要求较高,并且一些硬件特征相互关联,难以人工单独提取。近年来,深度学习迎来了快速发展,并逐渐应用到射频指纹识别领域。基于深度学习的物联网设备识别方法通常依靠分类神经网络来处理原始信号或其变换形式并直接推断设备身份,免除了繁杂的特征工程,引起了广泛关注。一些学者使用I/Q数据[7]、差分星座轨迹图[8]、短时傅里叶变换(short time Fourier transform,STFT)得到的时频谱[9~11]、功率谱[12,13]、双谱[14,15]、循环谱[16]作为神经网络的输入均取得了不错的效果,但使用单个信号特征在实验过程中仍存在着识别准确率不高的问题。
CUIB Q, SONGW K, WANGB L, et al. Xiaomi “smart phone × AIoT” security & privacy technology[J]. Journal of Wuhan University (Natural Science Edition), 2022, 68(1): 1-7. DOI: 10.14188/j.1671-8836.2021.2000(Ch ).
[4]
ZHANGJ Q, LIG Y, MARSHALLA, et al. A new frontier for IoT security emerging from three decades of key generation relying on wireless channels[J]. IEEE Access, 2020, 8: 138406-138446. DOI: 10.1109/ACCESS.2020.3012006 .
[5]
WANGS, PENGL N, FUH, et al. A convolutional neural network-based RF fingerprinting identification scheme for mobile phones[C]//IEEE INFOCOM 2020 ― IEEE Conference on Computer Communications Workshops(INFOCOM WKSHPS). New York: IEEE Press, 2020: 115-120. DOI: 10.1109/INFOCOMWKSHPS50562.2020.9163058 .
[6]
PENGL N, ZHANGJ Q, LIUM, et al. Deep learning based RF fingerprint identification using differential constellation trace figure[J]. IEEE Transactions on Vehicular Technology, 2020, 69(1): 1091-1095. DOI: 10.1109/TVT.2019.2950670 .
[7]
RUX H, LIUZ, HUANGZ T, et al. Evaluation of unintentional modulation for pulse compression signals based on spectrum asymmetry[J]. IET Radar, Sonar & Navigation, 2017, 11(4): 656-663. DOI: 10.1049/iet-rsn.2016.0248 .
[8]
WONGL J, HEADLEYW C, MICHAELSA J. Specific emitter identification using convolutional neural network-based IQ imbalance estimators[J]. IEEE Access, 2019, 7: 33544-33555. DOI: 10.1109/ACCESS.2019.2903444 .
NIUW Y, XUH, LIUY H, et al. Individual identification method based on PACGAN and differential constellation trace figure[J]. Journal of Signal Processing, 2021, 37(8): 1559-1567. DOI: 10.16798/j.issn.1003-0530.2021.08.024(Ch ).
[11]
SHENG X, ZHANGJ Q, MARSHALLA, et al. Radio frequency fingerprint identification for LoRa using spectrogram and CNN[C]//IEEE INFOCOM 2021 ― IEEE Conference on Computer Communications. New York: IEEE Press, 2021: 1-10. DOI: 10.1109/INFOCOM42981.2021.9488793 .
[12]
HUANGD, AL-HOURANIA, SITHAMPARANATHANK, et al. Deep learning methods for device authentication using RF fingerprinting[C]//2021 15th International Conference on Signal Processing and Communication Systems (ICSPCS). New York: IEEE Press, 2021: 1-7. DOI: 10.1109/ICSPCS53099.2021.9660226 .
[13]
CHENS L, XIEF Y, CHENY, et al. Identification of wireless transceiver devices using radio frequency (RF) fingerprinting based on STFT analysis to enhance authentication security[C]//2017 IEEE 5th International Symposium on Electromagnetic Compatibility. New York: IEEE Press, 2018: 1-5. DOI: 10.1109/EMC-B.2017.8260381 .
WANGJ, ZHANGB N, WEIG F, et al. Communication transmitter individual identification based on welch power spectrum and convolution neural network[J]. Telecommunication Engineering, 2021, 61(10): 1197-1204 (Ch). DOI: 10.3969/j.issn.1001-893x.2021.10.001 .
JIAJ C, QIL. RF fingerprint extraction method based on bispectrum[J]. Journal of Terahertz Science and Electronic Information Technology, 2021, 19(1): 107-111. DOI: 10.11805/TKYDA2019291(Ch ).
[20]
DINGL D, WANGS L, WANGF G, et al. Specific emitter identification via convolutional neural networks[J]. IEEE Communications Letters, 2018, 22(12): 2591-2594. DOI: 10.1109/LCOMM.2018.2871465 .
YANGX M, JINY J, LIY H. MATLAB simulation and analysis of the Welch method in the classical power spectrum estimation[J]. Electronic Test, 2011(7): 101-104. DOI: 10.3969/j.issn.1000-8519.2011.07.027(Ch ).
[25]
HEK M, ZHANGX Y, RENS Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). New York: IEEE Press, 2016: 770-778. DOI: 10.1109/CVPR.2016.90 .
BAXTERJ. A Bayesian/information theoretic model of learning to learn via multiple task sampling[J]. Machine Learning, 1997, 28(1): 7-39. DOI: 10.1023/A: 1007327622663 .
SHENG X, ZHANGJ Q, MARSHALLA, et al. Towards scalable and channel-robust radio frequency fingerprint identification for LoRa[J]. IEEE Transactions on Information Forensics and Security, 2022, 17: 774-787. DOI: 10.1109/TIFS.2022.3152404 .