WO2021109672A1 - 一种噪声增强射频指纹识别方法及装置 - Google Patents

一种噪声增强射频指纹识别方法及装置 Download PDF

Info

Publication number
WO2021109672A1
WO2021109672A1 PCT/CN2020/115323 CN2020115323W WO2021109672A1 WO 2021109672 A1 WO2021109672 A1 WO 2021109672A1 CN 2020115323 W CN2020115323 W CN 2020115323W WO 2021109672 A1 WO2021109672 A1 WO 2021109672A1
Authority
WO
WIPO (PCT)
Prior art keywords
wireless device
signal
mahalanobis distance
feature vector
noise
Prior art date
Application number
PCT/CN2020/115323
Other languages
English (en)
French (fr)
Inventor
方昊
周新宇
Original Assignee
南京东科优信网络安全技术研究院有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 南京东科优信网络安全技术研究院有限公司 filed Critical 南京东科优信网络安全技术研究院有限公司
Publication of WO2021109672A1 publication Critical patent/WO2021109672A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/309—Measuring or estimating channel quality parameters
    • H04B17/336—Signal-to-interference ratio [SIR] or carrier-to-interference ratio [CIR]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Definitions

  • the invention relates to the technical field of information security, in particular to a noise-enhanced radio frequency fingerprint identification method and device.
  • radio frequency fingerprint uses the characteristics of the hardware device itself, which is a comprehensive performance of the tolerances generated by the hardware on the circuit during production. It is unique and difficult to clone, which has caused a lot of research and practice.
  • the current radio frequency fingerprint technology consists of three parts: signal acquisition and processing, signal feature extraction and feature recognition. Most of the existing schemes only complete the experiment under the ideal signal-to-noise ratio. When the signal-to-noise ratio changes, its effectiveness is questionable. In order to achieve performance under different signal-to-noise ratios, traditional solutions need to build models under multiple signal-to-noise ratios. The repeated training process consumes time and computing resources. When the signal-to-noise ratio is low, simple Monte Carlo simulation is difficult. The influence of exhausting noise causes a significant drop in performance under low signal-to-noise ratio. Therefore, there is an urgent need for a lightweight radio frequency fingerprint identification method that can operate in a variety of channel environments.
  • the present invention provides a noise-enhanced radio frequency fingerprint identification method and device. After acquiring the signal to be identified, the present invention first performs signal processing on it, and then adjusts the system parameters according to the signal-to-noise ratio. Recognizing signals can effectively adapt to various signal-to-noise ratio channel environments and improve the recognition performance of low-power devices in actual use.
  • the noise-enhanced radio frequency fingerprint identification method of the present invention includes:
  • step (2) Collect the output signal of the wireless device to be identified, estimate the signal-to-noise ratio of the output signal, and process according to step (2) and step (3) to obtain the feature vector of the wireless device to be identified;
  • the preprocessing in step (2) includes: sequential down-conversion, over-sampling, signal detection and interception, energy normalization, estimation and compensation of signal frequency deviation and phase deviation, and I/Q signal extraction .
  • step (3) specifically includes:
  • y k () represents the target symbol selected from the preprocessed signal of the wireless device k
  • y′ k (t) represents the superimposed signal
  • T represents the symbol period
  • N is the number of superimpositions
  • F k [real(y′ k (T S )), imag(y′ k (T S )), real(y′ k (2T S )), imag(y′ k (2T S )), .. ., real(y′ k (T)), imag(y′ k (T))]
  • F k represents the feature vector of the wireless device k
  • real() represents the real part
  • imag() represents the imaginary part
  • T S is the sampling time
  • k 1,...,K.
  • step (6) specifically includes:
  • SNR a represents the estimated signal-to-noise ratio, Represents noise variance
  • ⁇ * represents the Mahalanobis distance parameter
  • ⁇ represents the basic Mahalanobis distance parameter
  • I represents the unit matrix
  • N represents the number of target symbols.
  • step (7) specifically includes:
  • Mah(F test , ⁇ k ) represents the Mahalanobis distance between the feature vector F test of the wireless device to be identified and the template ⁇ k of the wireless device model k,
  • the noise-enhanced radio frequency fingerprint identification device of the present invention includes:
  • the acquisition module is used to collect the output signals of a number of wireless devices of known device models in a high signal-to-noise ratio environment during the training phase, and to collect the output signals of the wireless devices to be identified in the recognition phase;
  • the preprocessing module is used to preprocess the signals collected by the acquisition module;
  • the feature vector acquisition module is used to select several target symbols from the preprocessed signal to perform periodic superposition and expansion to obtain the feature vector of the corresponding wireless device; wherein the target symbol is the symbol 0 repeatedly sent in the preprocessed signal;
  • the first parameter acquisition module is used to calculate the mean value of the characteristic vector of each wireless device as a template of the corresponding wireless device model, and calculate the mean value of the covariance of the characteristic vectors of all wireless devices as the basic Mahalanobis distance parameter;
  • the signal-to-noise ratio estimation module is used to estimate the signal-to-noise ratio of the signal collected by the second acquisition module;
  • the second preprocessing module is used to preprocess the signals collected by the acquisition module in the recognition phase;
  • the second parameter acquisition module is configured to calculate the Mahalanobis distance parameter according to the signal-to-noise ratio estimated by the signal-to-noise ratio estimation module and the Mahalanobis distance basic parameters obtained by the first parameter acquisition module;
  • the device identification module is used to calculate the Mahalanobis distance between the feature vector of the wireless device to be identified and the template of all wireless device models obtained by the first parameter acquisition module according to the Mahalanobis distance parameter obtained by the second parameter acquisition module, and select the minimum Mahalanobis distance
  • the corresponding wireless device model is used as the device model of the wireless device to be identified.
  • the pre-processing performed by the pre-processing module includes: sequential down-conversion, over-sampling, signal detection and interception, energy normalization, estimation and compensation of signal frequency deviation and phase deviation, and I/Q signal extraction .
  • the feature vector acquiring module specifically includes:
  • the target symbol acquisition unit is configured to select several target symbols from the preprocessed signal of each wireless device, where the target symbol is the symbol 0 repeatedly sent in the preprocessed signal;
  • the symbol superimposition unit is used to superimpose the selected target symbols according to the symbol period.
  • the specific formula is:
  • y k () represents the target symbol selected from the preprocessed signal of the wireless device k
  • y′ k (t) represents the superimposed signal
  • T represents the symbol period
  • N is the number of superimpositions
  • the symbol expansion unit is used to expand the superimposed signal according to I/Q and stack it into a feature vector:
  • F k [real(y′ k (T S )), imag(y′ k (T S )), real(y′ k ((2T S )), imag(y′ k (2T S )),. .., real(y′ k (T)), imag(y′ k (T))]
  • F k represents the feature vector of the wireless device k
  • real() represents the real part
  • imag() represents the imaginary part
  • T S is the sampling time
  • k 1,...,K.
  • the first parameter acquisition module specifically includes:
  • the noise variance obtaining unit is used to calculate the noise variance according to the estimated signal-to-noise ratio according to the following formula:
  • SNR a represents the estimated signal-to-noise ratio, Represents noise variance
  • the parameter acquisition unit is configured to update the basic Mahalanobis distance parameter according to the noise variance to obtain the Mahalanobis distance parameter:
  • ⁇ * represents the Mahalanobis distance parameter
  • ⁇ represents the basic Mahalanobis distance parameter
  • I represents the unit matrix
  • N represents the number of superpositions.
  • the device identification module specifically includes:
  • the Mahalanobis distance calculation unit is configured to calculate the Mahalanobis distance between the feature vector of the wireless device to be identified and the templates of all wireless device models according to the Mahalanobis distance parameter according to the following formula:
  • Mah(F test , ⁇ k ) represents the Mahalanobis distance between the feature vector F test of the wireless device to be identified and the template ⁇ k of the wireless device model k,
  • the device model confirmation unit is used to select the wireless device model corresponding to the minimum Mahalanobis distance as the device model of the wireless device to be identified.
  • the present invention has a significant advantage: through the method of the present invention, the received signal signal-to-noise ratio estimated based on measurement can be used to adjust the identification parameters of the system, thereby achieving robustness under different signal-to-noise ratios. Radio frequency fingerprint identification. It can be obtained through simulations and experiments that the use of the present invention can greatly improve the recognition performance of low-power devices, especially in environments with different signal-to-noise ratios.
  • FIG. 1 is a schematic flowchart of an embodiment of a noise-enhanced radio frequency fingerprint identification method provided by the present invention.
  • This embodiment provides a noise-enhanced radio frequency fingerprint identification method, as shown in FIG. 1, including the following steps:
  • the signal When collecting the output signal of the wireless device, the signal is collected through a direct coaxial line plus an attenuator connection, or the signal is collected in a wireless receiving environment where the signal-to-noise ratio is higher than the preset value at close range, visible distance.
  • 50 ZigBee wireless transmitting modules are selected as target wireless devices, and numbered as 1-50.
  • USRP equipment is used to collect line-of-sight transmission signals at close range, and the signal-to-noise ratio of the collected signals is 26dB.
  • preprocessing includes: down-conversion, over-sampling, signal detection and interception, energy normalization, signal frequency offset and phase offset estimation and compensation, and I/Q signal extraction.
  • the symbol rate of the original ZigBee device is 1Mbps
  • the signal is directly down-converted to the baseband signal during acquisition, and the sampling rate is 10Mbps.
  • the received signal is framed according to the sampling point change, the energy of each frame is normalized, and the frequency deviation of the signal is finally performed And phase deviation processing (refer to patent 201510797097.8 for specific methods).
  • This step specifically includes:
  • y k () represents the target symbol selected from the preprocessed signal of the wireless device k
  • y′ k (t) represents the superimposed signal
  • T represents the symbol period
  • N is the number of superimpositions
  • k 1, ..., K
  • F k [real(y′ k (T S )), imag(y′ k (T S )), real(y′ k (2T S )), imag(y′ k (2T S )), .. ., real(y′ k (T)), imag(y′ k (T))]
  • F k represents the feature vector of the wireless device k
  • real() represents the real part
  • imag() represents the imaginary part
  • T S is the sampling time
  • k 1,...,K.
  • the feature vector comes from K multivariate Gaussian distributions with different means and the same covariance. It can be seen that when identifying the device under test, only the posterior probability of the mean value of the signal under test and the existing multivariate Gaussian distribution is calculated. The signal under test should belong to the device number with the largest posterior probability. Under the condition of equivalence of the appearance of the equipment, calculating the posterior probability is equivalent to calculating the Mahalanobis distance from the mean.
  • step (2) Collect the output signal of the wireless device to be identified, estimate the signal-to-noise ratio of the output signal, and process according to step (2) and step (3) to obtain the feature vector of the wireless device to be identified.
  • This step specifically includes:
  • SNR a represents the estimated signal-to-noise ratio, Represents the noise variance; in the additive white Gaussian noise channel, the system noise presents a Gaussian distribution and has nothing to do with the signal sent by the device. Therefore, the eigenvector of the signal received in the channel still obeys the Gaussian distribution, and can be based on the signal to be identified.
  • Signal-to-noise ratio SNR a to calculate variance;
  • ⁇ * represents the Mahalanobis distance parameter
  • ⁇ represents the basic Mahalanobis distance parameter
  • I represents the unit matrix
  • N represents the number of superpositions
  • This step specifically includes:
  • Mah(F test , ⁇ k ) represents the Mahalanobis distance between the feature vector F test of the wireless device to be identified and the template ⁇ k of the wireless device model k. Since the energy is normalized in the process of receiving the signal, Gaussian white noise Does not affect the mean, so the new system mean is
  • the accuracy of identification and classification of low-power consumption devices can be effectively improved.
  • the use of high signal-to-noise ratio as training and Mahalanobis distance recognition results in a severe drop below 15dB.
  • the recognition accuracy has been significantly improved.
  • the present invention also provides a noise-enhanced radio frequency fingerprint identification device, which includes:
  • the acquisition module is used to collect the output signals of a number of wireless devices of known device models in a high signal-to-noise ratio environment during the training phase, and to collect the output signals of the wireless devices to be identified in the recognition phase;
  • the preprocessing module is used to preprocess the signals collected by the acquisition module;
  • the feature vector acquisition module is used to select several target symbols from the preprocessed signal to perform periodic superposition and expansion to obtain the feature vector of the corresponding wireless device; wherein the target symbol is the symbol 0 repeatedly sent in the preprocessed signal;
  • the first parameter acquisition module is used to calculate the mean value of the characteristic vector of each wireless device as a template of the corresponding wireless device model, and calculate the mean value of the covariance of the characteristic vectors of all wireless devices as the basic Mahalanobis distance parameter;
  • the signal-to-noise ratio estimation module is used to estimate the signal-to-noise ratio of the signal collected by the second acquisition module;
  • the second preprocessing module is used to preprocess the signals collected by the acquisition module in the recognition phase;
  • the second parameter acquisition module is configured to calculate the Mahalanobis distance parameter based on the signal-to-noise ratio estimated by the signal-to-noise ratio estimation module and the Mahalanobis distance basic parameter obtained by the first parameter acquisition module;
  • the device identification module is used to calculate the Mahalanobis distance between the feature vector of the wireless device to be identified and the template of all wireless device models obtained by the first parameter acquisition module according to the Mahalanobis distance parameter obtained by the second parameter acquisition module, and select the minimum Mahalanobis distance
  • the corresponding wireless device model is used as the device model of the wireless device to be identified.
  • the pre-processing performed by the pre-processing module includes: sequential down-conversion, over-sampling, signal detection and interception, energy normalization, estimation and compensation of signal frequency deviation and phase deviation, and I/Q signal extraction .
  • the feature vector acquiring module specifically includes:
  • the target symbol acquisition unit is configured to select several target symbols from the preprocessed signal of each wireless device, where the target symbol is the symbol 0 repeatedly sent in the preprocessed signal;
  • the symbol superimposition unit is used to superimpose the selected target symbols according to the symbol period.
  • the specific formula is:
  • y k () represents the target symbol selected from the preprocessed signal of the wireless device k
  • y′ k (t) represents the superimposed signal
  • T represents the symbol period
  • N is the number of superimpositions
  • the symbol expansion unit is used to expand the superimposed signal according to I/Q and stack it into a feature vector:
  • F k [real(y′ k (T S )), imag(y′ k (T S )), real(y′ k (2T S )), imag(y′ k (2T S )), .. ., real(y′ k (T)), imag(y′ k (T))]
  • F k represents the feature vector of the wireless device k
  • T S is the sampling time
  • k 1,...,K.
  • the first parameter acquisition module specifically includes:
  • the noise variance obtaining unit is used to calculate the noise variance according to the estimated signal-to-noise ratio according to the following formula:
  • SNR a represents the estimated signal-to-noise ratio, Represents noise variance
  • the parameter acquisition unit is configured to update the basic Mahalanobis distance parameter according to the noise variance to obtain the Mahalanobis distance parameter:
  • ⁇ * represents the Mahalanobis distance parameter
  • ⁇ represents the basic Mahalanobis distance parameter
  • I represents the unit matrix
  • N represents the number of superpositions.
  • the device identification module specifically includes:
  • the Mahalanobis distance calculation unit is configured to calculate the Mahalanobis distance between the feature vector of the wireless device to be identified and the templates of all wireless device models according to the Mahalanobis distance parameter according to the following formula:
  • Mah(F test , ⁇ k ) represents the Mahalanobis distance between the feature vector F test of the wireless device to be identified and the template ⁇ k of the wireless device model k,
  • the device model confirmation unit is used to select the wireless device model corresponding to the minimum Mahalanobis distance as the device model of the wireless device to be identified.
  • the device corresponds to the above-mentioned method one-to-one, and the above-mentioned method is referred to for details that are not exhaustive, and will not be repeated.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Electromagnetism (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Monitoring And Testing Of Transmission In General (AREA)

Abstract

本发明公开了一种噪声增强射频指纹识别方法及装置，其中方法包括：采集高信噪比环境中若干已知设备型号的无线设备的输出信号；(2)对输出信号进行预处理；(3)选取若干目标符号进行周期叠加和展开，得到特征向量；(4)计算特征向量均值，作为对应的设备型号模板，并计算马氏距离基础参数；(5)采集待识别无线设备的输出信号并进行信噪比估计，再经过处理后得到待识别无线设备的特征向量；(6)根据估计的信噪比和所述马氏距离基础参数计算得到马氏距离参数；(7)根据马氏距离参数计算待识别无线设备的特征向量与所有型号模板的马氏距离，选择最小值对应的型号作为识别出的设备型号。本发明安全有效，可应用于多种信噪比信道环境中。

Description

一种噪声增强射频指纹识别方法及装置 技术领域
本发明涉及信息安全技术领域，尤其涉及一种噪声增强射频指纹识别方法及装置。
背景技术
随着无线通信技术的不断发展，尤其是物联网设备在各个领域的普及，其安全问题也越来越严峻。无线信号传输的开放性引起了一系列的安全挑战，如IP/MAC仿冒，重放攻击和拒绝服务攻击等。并且，由于大部分物联网设备成本低廉，难以支持复杂运算，所以轻量的物理层射频指纹安全技术得到了广泛的关注。射频指纹利用硬件设备本身的特征，是电路上硬件在生产制造时产生的容差的综合表现，具有唯一性，且难以克隆，引起了大量的研究与实践。
现在的射频指纹技术由信号采集与处理、信号特征提取与特征识别三个部分组成。大部分现有方案都只是在理想信噪比下完成实验，当信噪比发生变化时，其有效性存疑。传统方案为了实现不同信噪比下的性能，需要构建多种信噪比条件下的模型，重复的训练过程耗费时间与计算资源，并且当信噪比较低时，简单的蒙特卡洛仿真难以穷尽噪声的影响，造成低信噪比下性能大幅下降。因此，一种轻量的能够在多种信道环境下运行的射频指纹识别方法是迫切需求的。
发明内容
发明目的：本发明针对现有技术存在的问题，提供一种噪声增强射频指纹识别方法及装置，本发明在获取待识别信号后，先对其进行信号处理，随后根据信噪比调节系统参数，对信号进行识别，可以有效的适应各种信噪比信道环境，提高低功耗设备在实际使用中的识别性能。
技术方案：本发明所述的噪声增强射频指纹识别方法包括：
(1)采集高信噪比环境中若干已知设备型号的无线设备的输出信号；
(2)对每一无线设备的输出信号进行预处理；
(3)从每一无线设备的预处理后信号中选取若干目标符号进行周期叠加和展开，得到每一无线设备的特征向量；其中，所述目标符号为预处理后信号中重复发送的符号0；
(4)计算每一无线设备的特征向量的均值，作为对应的无线设备型号 的模板，并计算所有无线设备特征向量的协方差均值，作为马氏距离基础参数；
(5)采集待识别无线设备的输出信号，对该输出信号进行信噪比估计，并按照步骤(2)和步骤(3)处理得到待识别无线设备的特征向量；
(6)根据估计的信噪比和所述马氏距离基础参数计算得到马氏距离参数；
(7)根据所述马氏距离参数计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离，选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
进一步的，步骤(2)中所述预处理包括：依次进行的下变频、过采样、信号检测和截取、能量归一化、信号频偏和相偏的估计与补偿和I/Q路信号提取。
进一步的，步骤(3)具体包括：
(3.1)从每一无线设备的预处理后信号中选取若干目标符号，所述目标符号为预处理后信号中重复发送的符号0；
(3.2)将选取的目标符号按符号周期进行叠加，其具体公式为：
Figure PCTCN2020115323-appb-000001
式中，y k()表示从无线设备k的预处理后信号中选取的目标符号，y′ k(t)表示叠加后的信号，T表示符号周期，N是叠加的数量，k＝1，...，K，K表示无线设备的个数；
(3.3)对叠加后的信号按I/Q进行展开，堆叠成特征向量：
F k＝[real(y′ k(T S))，imag(y′ k(T S))，real(y′ k(2T S))，imag(y′ k(2T S))，...，real(y′ k(T))，imag(y′ k(T))]
式中，F k表示无线设备k的特征向量，real()表示取实部，imag()表示取虚部，T S为采样时间，k＝1，...，K。
进一步的，步骤(6)具体包括：
(6.1)根据估计的信噪比按照下式计算噪声方差：
Figure PCTCN2020115323-appb-000002
式中，SNR a表示估计的信噪比，
Figure PCTCN2020115323-appb-000003
表示噪声方差；
(6.2)根据噪声方差更新所述马氏距离基础参数，得到马氏距离参数：
Figure PCTCN2020115323-appb-000004
式中，∑ *表示马氏距离参数，∑表示马氏距离基础参数，I表示单位矩阵，N表示目标符号个数。
进一步的，步骤(7)具体包括：
(7.1)根据所述马氏距离参数按照下式计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离：
Figure PCTCN2020115323-appb-000005
式中，Mah(F test，μ k)表示待识别无线设备的特征向量F test与无线设备型号k的模板μ k的马氏距离，
Figure PCTCN2020115323-appb-000006
(7.2)选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
本发明所述的噪声增强射频指纹识别装置包括：
采集模块，用于在训练阶段采集高信噪比环境中若干已知设备型号的无线设备的输出信号，以及在识别阶段采集待识别无线设备的输出信号；
预处理模块，用于对采集模块采集的信号进行预处理；
特征向量获取模块，用于从预处理后信号中选取若干目标符号进行周期叠加和展开，得到对应无线设备的特征向量；其中，所述目标符号为预处理后信号中重复发送的符号0；
第一参数获取模块，用于计算每一无线设备的特征向量的均值，作为对应的无线设备型号的模板，并计算所有无线设备特征向量的协方差均值，作为马氏距离基础参数；
信噪比估计模块，用于对第二采集模块采集的信号进行信噪比估计；
第二预处理模块，用于对识别阶段采集模块采集的信号进行预处理；
第二参数获取模块，用于根据信噪比估计模块估计的信噪比和第一参数获取模块得到的马氏距离基础参数计算得到马氏距离参数；
设备识别模块，用于根据第二参数获取模块得到的马氏距离参数计算待识别无线设备的特征向量与第一参数获取模块得到的所有无线设备型号的模板的马氏距离，选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
进一步的，所述预处理模块进行的预处理包括：依次进行的下变频、过采样、信号检测和截取、能量归一化、信号频偏和相偏的估计与补偿和I/Q路信号提取。
进一步的，所述特征向量获取模块具体包括：
目标符号获取单元，用于从每一无线设备的预处理后信号中选取若干目标符号，所述目标符号为预处理后信号中重复发送的符号0；
符号叠加单元，用于将选取的目标符号按符号周期进行叠加，其具体公式为：
Figure PCTCN2020115323-appb-000007
式中，y k()表示从无线设备k的预处理后信号中选取的目标符号，y′ k(t)表示叠加后的信号，T表示符号周期，N是叠加的数量，k＝1，...，K，K表示无线设备的个数；
符号展开单元，用于对叠加后的信号按I/Q进行展开，堆叠成特征向量：
F k＝[real(y′ k(T S))，imag(y′ k(T S))，real(y′ k((2T S))，imag(y′ k(2T S))，...，real(y′ k(T))，imag(y′ k(T))]
式中，F k表示无线设备k的特征向量，real()表示取实部，imag()表示取虚部，T S为采样时间，k＝1，...，K。
进一步的，所述第一参数获取模块具体包括：
噪声方差获取单元，用于根据估计的信噪比按照下式计算噪声方差：
Figure PCTCN2020115323-appb-000008
式中，SNR a表示估计的信噪比，
Figure PCTCN2020115323-appb-000009
表示噪声方差；
参数获取单元，用于根据噪声方差更新所述马氏距离基础参数，得到马氏距离参数：
Figure PCTCN2020115323-appb-000010
式中，∑ *表示马氏距离参数，∑表示马氏距离基础参数，I表示单位矩阵，N表示叠加的数量。
进一步的，所述设备识别模块具体包括：
马氏距离计算单元，用于根据所述马氏距离参数按照下式计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离：
Figure PCTCN2020115323-appb-000011
式中，Mah(F test，μ k)表示待识别无线设备的特征向量F test与无线设 备型号k的模板μ k的马氏距离，
Figure PCTCN2020115323-appb-000012
设备型号确认单元，用于选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
有益效果：本发明与现有技术相比，其显著优点是：通过本发明方法，可以基于测量估计出的接收信号信噪比，调节系统的识别参数，从而在不同信噪比下实现鲁棒的射频指纹识别。通过仿真和实验可以得到，使用本发明将可以极大提高低功耗设备的识别性能，尤其在不同信噪比环境中有较大的改进效果。
附图说明
图1是本发明提供的噪声增强射频指纹识别方法的一个实施例的流程示意图。
具体实施方式
本实施例提供了一种噪声增强射频指纹识别方法，如图1所示，包括以下步骤：
(1)采集高信噪比环境中若干已知设备型号的无线设备的输出信号。
采集无线设备的输出信号时，通过直接同轴线加衰减器连接采集信号，或者是在近距离、可视距、信噪比高于预设值的无线接收环境下采集信号。本实施例中，选用50个ZigBee无线发射模块作为目标无线设备，并按1-50编号。采用USRP设备在近距离采集视距传输信号，采集信号的信噪比为26dB。
(2)对每一无线设备的输出信号进行预处理。
其中，预处理包括：下变频、过采样、信号检测和截取、能量归一化、信号频偏和相偏的估计与补偿、I/Q路信号提取。原始ZigBee设备的符号速率为1Mbps，采集时直接将信号下变频至基带信号，采样速率为10Mbps，对接收信号按采样点变化分帧后，每帧进行能量归一化，最后进行信号的频偏和相偏处理(具体方法参照专利201510797097.8)。
(3)从每一无线设备的预处理后信号中选取若干目标符号进行周期叠加和展开，得到每一无线设备的特征向量；其中，所述目标符号为预处理后信号中重复发送的符号0。
该步骤具体包括：
(3.1)从每一无线设备的预处理后信号中选取若干目标符号，所述目标符号为预处理后信号中重复发送的符号0；
(3.2)将选取的目标符号按符号周期进行叠加，其具体公式为：
Figure PCTCN2020115323-appb-000013
式中，y k()表示从无线设备k的预处理后信号中选取的目标符号，y′ k(t)表示叠加后的信号，T表示符号周期，N是叠加的数量，k＝1，...，K，K表示无线设备的个数；本实施例中，N＝60，在具体的应用中，叠加的目标符号个数由系统实际所处环境和应用要求的安全等级决定；
(3.3)对叠加后的信号按I/Q进行展开，堆叠成特征向量：
F k＝[real(y′ k(T S))，imag(y′ k(T S))，real(y′ k(2T S))，imag(y′ k(2T S))，...，real(y′ k(T))，imag(y′ k(T))]
式中，F k表示无线设备k的特征向量，real()表示取实部，imag()表示取虚部，T S为采样时间，k＝1，...，K。
(4)计算每一无线设备的特征向量的均值，作为对应的无线设备型号的模板，并计算所有无线设备特征向量的协方差均值，作为马氏距离基础参数。
在训练系统模型时，根据实验采集数据，认为特征向量来源于K个均值不同，协方差相同的多元高斯分布。由此可知，在识别待测设备时，只需计算待测信号与已有多元高斯分布的均值的后验概率，待测信号应属于后验概率最大的设备号。在设备出现的等概的条件下，计算后验概率等价于计算与均值的马氏距离。因此，对K个设备中每个设备的特征向量分别计算均值μ k，k＝1，...，K，作为其模板，同时计算每个设备的协方差，取所有设备协方差的平均∑作为马氏距离基础参数。
(5)采集待识别无线设备的输出信号，对该输出信号进行信噪比估计，并按照步骤(2)和步骤(3)处理得到待识别无线设备的特征向量。
(6)根据估计的信噪比和所述马氏距离基础参数计算得到马氏距离参 数。
该步骤具体包括：
(6.1)根据估计的信噪比按照下式计算噪声方差：
Figure PCTCN2020115323-appb-000014
式中，SNR a表示估计的信噪比，
Figure PCTCN2020115323-appb-000015
表示噪声方差；在加性高斯白噪声信道中，系统噪声呈现高斯分布，且与设备发送信号无关，因此，在该信道中接收的信号的特征向量仍然服从高斯分布，并且可以根据待识别信号的信噪比SNR a来进行计算方差；
(6.2)信号方差除受能量归一化影响外，也受到高斯噪声的影响，因此根据噪声方差更新所述马氏距离基础参数，得到马氏距离参数：
Figure PCTCN2020115323-appb-000016
式中，∑ *表示马氏距离参数，∑表示马氏距离基础参数，I表示单位矩阵，N表示叠加的数量
(7)根据所述马氏距离参数计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离，选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
该步骤具体包括：
(7.1)根据所述马氏距离参数按照下式计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离：
Figure PCTCN2020115323-appb-000017
式中，Mah(F test，μ k)表示待识别无线设备的特征向量F test与无线设 备型号k的模板μ k的马氏距离，由于接收信号过程中进行了能量归一化，高斯白噪声不影响均值，因此新的系统均值为
Figure PCTCN2020115323-appb-000018
(7.2)选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
通过本发明方法，可以有效的提升对低功耗设备的识别和分类的准确率。如下表所示，单纯以高信噪比作为训练，使用马氏距离识别，其结果在15dB以下有严重下降，使用本发明方法调整参数后，识别正确率有了明显的提升。
Figure PCTCN2020115323-appb-000019
本发明还提供了一种噪声增强射频指纹识别装置，包括：
采集模块，用于在训练阶段采集高信噪比环境中若干已知设备型号的无线设备的输出信号，以及在识别阶段采集待识别无线设备的输出信号；
预处理模块，用于对采集模块采集的信号进行预处理；
特征向量获取模块，用于从预处理后信号中选取若干目标符号进行周期叠加和展开，得到对应无线设备的特征向量；其中，所述目标符号为预处理后信号中重复发送的符号0；
第一参数获取模块，用于计算每一无线设备的特征向量的均值，作为对应的无线设备型号的模板，并计算所有无线设备特征向量的协方差均值，作为马氏距离基础参数；
信噪比估计模块，用于对第二采集模块采集的信号进行信噪比估计；
第二预处理模块，用于对识别阶段采集模块采集的信号进行预处理；
第二参数获取模块，用于根据信噪比估计模块估计的信噪比和第一参 数获取模块得到的马氏距离基础参数计算得到马氏距离参数；
设备识别模块，用于根据第二参数获取模块得到的马氏距离参数计算待识别无线设备的特征向量与第一参数获取模块得到的所有无线设备型号的模板的马氏距离，选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
进一步的，所述预处理模块进行的预处理包括：依次进行的下变频、过采样、信号检测和截取、能量归一化、信号频偏和相偏的估计与补偿和I/Q路信号提取。
进一步的，所述特征向量获取模块具体包括：
目标符号获取单元，用于从每一无线设备的预处理后信号中选取若干目标符号，所述目标符号为预处理后信号中重复发送的符号0；
符号叠加单元，用于将选取的目标符号按符号周期进行叠加，其具体公式为：
Figure PCTCN2020115323-appb-000020
式中，y k()表示从无线设备k的预处理后信号中选取的目标符号，y′ k(t)表示叠加后的信号，T表示符号周期，N是叠加的数量，k＝1，...，K，K表示无线设备的个数；
符号展开单元，用于对叠加后的信号按I/Q进行展开，堆叠成特征向量：
F k＝[real(y′ k(T S))，imag(y′ k(T S))，real(y′ k(2T S))，imag(y′ k(2T S))，...，real(y′ k(T))，imag(y′ k(T))]
式中，F k表示无线设备k的特征向量，T S为采样时间，k＝1，...，K。
进一步的，所述第一参数获取模块具体包括：
噪声方差获取单元，用于根据估计的信噪比按照下式计算噪声方差：
Figure PCTCN2020115323-appb-000021
式中，SNR a表示估计的信噪比，
Figure PCTCN2020115323-appb-000022
表示噪声方差；
参数获取单元，用于根据噪声方差更新所述马氏距离基础参数，得到马氏距离参数：
Figure PCTCN2020115323-appb-000023
式中，∑ *表示马氏距离参数，∑表示马氏距离基础参数，I表示单位矩阵，N表示叠加的数量。
进一步的，所述设备识别模块具体包括：
马氏距离计算单元，用于根据所述马氏距离参数按照下式计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离：
Figure PCTCN2020115323-appb-000024
式中，Mah(F test，μ k)表示待识别无线设备的特征向量F test与无线设备型号k的模板μ k的马氏距离，
Figure PCTCN2020115323-appb-000025
设备型号确认单元，用于选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
该装置与上述方法一一对应，未详尽之处参考上述方法，不再赘述。
以上所揭露的仅为本发明一种较佳实施例而已，不能以此来限定本发明之权利范围，因此依本发明权利要求所作的等同变化，仍属本发明所涵盖的范围。

Claims (10)

  1. 一种噪声增强射频指纹识别方法，其特征在于该方法包括：
    (1)采集高信噪比环境中若干已知设备型号的无线设备的输出信号；
    (2)对每一无线设备的输出信号进行预处理；
    (3)从每一无线设备的预处理后信号中选取若干目标符号进行周期叠加和展开，得到每一无线设备的特征向量；其中，所述目标符号为预处理后信号中重复发送的符号0；
    (4)计算每一无线设备的特征向量的均值，作为对应的无线设备型号的模板，并计算所有无线设备特征向量的协方差均值，作为马氏距离基础参数；
    (5)采集待识别无线设备的输出信号，对该输出信号进行信噪比估计，并按照步骤(2)和步骤(3)处理得到待识别无线设备的特征向量；
    (6)根据估计的信噪比和所述马氏距离基础参数计算得到马氏距离参数；
    (7)根据所述马氏距离参数计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离，选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
  2. 根据权利要求1所述的噪声增强射频指纹识别方法，其特征在于：步骤(2)中所述预处理包括：依次进行的下变频、过采样、信号检测和截取、能量归一化、信号频偏和相偏的估计与补偿和I/Q路信号提取。
  3. 根据权利要求1所述的噪声增强射频指纹识别方法，其特征在于：步骤(3)具体包括：
    (3.1)从每一无线设备的预处理后信号中选取若干目标符号，所述目标符号为预处理后信号中重复发送的符号0；
    (3.2)将选取的目标符号按符号周期进行叠加，其具体公式为：
    Figure PCTCN2020115323-appb-100001
    式中，y k()表示从无线设备k的预处理后信号中选取的目标符号，y′ k(t)表示叠加后的信号，T表示符号周期，N是叠加的数量，k＝1，...，K，K表示无线设备的个数；
    (3.3)对叠加后的信号按I/Q进行展开，堆叠成特征向量：
    F k＝[real(y′ k(T S))，imag(y′ k(T S))，real(y′ k(2T S))，imag(y′ k(2T S))，...，real(y′ k(T))，imag(y′ k(T))]
    式中，F k表示无线设备k的特征向量，T S为采样时间，real()表示取实部，imag()表示取虚部，k＝1，...，K。
  4. 根据权利要求1所述的噪声增强射频指纹识别方法，其特征在于：步骤(6)具体包括：
    (6.1)根据估计的信噪比按照下式计算噪声方差：
    Figure PCTCN2020115323-appb-100002
    式中，SNR a表示估计的信噪比，
    Figure PCTCN2020115323-appb-100003
    表示噪声方差；
    (6.2)根据噪声方差更新所述马氏距离基础参数，得到马氏距离参数：
    Figure PCTCN2020115323-appb-100004
    式中，∑ *表示马氏距离参数，∑表示马氏距离基础参数，I表示单位矩阵，N表示叠加的数量。
  5. 根据权利要求4所述的噪声增强射频指纹识别方法，其特征在于：步骤(7)具体包括：
    (7.1)根据所述马氏距离参数按照下式计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离：
    Figure PCTCN2020115323-appb-100005
    式中，Mah(F test，μ k)表示待识别无线设备的特征向量F test与无线设备型号k的模板μ k的马氏距离，
    Figure PCTCN2020115323-appb-100006
    (7.2)选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
  6. 一种噪声增强射频指纹识别装置，其特征在于包括：
    采集模块，用于在训练阶段采集高信噪比环境中若干已知设备型号的无线设备的输出信号，以及在识别阶段采集待识别无线设备的输出信号；
    预处理模块，用于对采集模块采集的信号进行预处理；
    特征向量获取模块，用于从预处理后信号中选取若干目标符号进行周期叠加和展开，得到对应无线设备的特征向量；其中，所述目标符号为预处理后信号中重复发送的符号0；
    第一参数获取模块，用于计算每一无线设备的特征向量的均值，作为对应的无线设备型号的模板，并计算所有无线设备特征向量的协方差均值，作为马氏距离基础参数；
    信噪比估计模块，用于对第二采集模块采集的信号进行信噪比估计；
    第二预处理模块，用于对识别阶段采集模块采集的信号进行预处理；
    第二参数获取模块，用于根据信噪比估计模块估计的信噪比和第一参数获取模块得到的马氏距离基础参数计算得到马氏距离参数；
    设备识别模块，用于根据第二参数获取模块得到的马氏距离参数计算待识别无线设备的特征向量与第一参数获取模块得到的所有无线设备型号的模板的马氏距离，选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
  7. 根据权利要求6所述的噪声增强射频指纹识别装置，其特征在于：所述预处理模块进行的预处理包括：依次进行的下变频、过采样、信号检测和截取、能量归一化、信号频偏和相偏的估计与补偿和I/Q路信号提取。
  8. 根据权利要求6所述的噪声增强射频指纹识别装置，其特征在于：所述特征向量获取模块具体包括：
    目标符号获取单元，用于从每一无线设备的预处理后信号中选取若干目标符号，所述目标符号为预处理后信号中重复发送的符号0；
    符号叠加单元，用于将选取的目标符号按符号周期进行叠加，其具体公式为：
    Figure PCTCN2020115323-appb-100007
    式中，y k()表示从无线设备k的预处理后信号中选取的目标符号，y′ k(t)表示叠加后的信号，T表示符号周期，N是叠加的数量，k＝1，...，K，K表示无线设备的个数；
    符号展开单元，用于对叠加后的信号按I/Q进行展开，堆叠成特征向量：
    F k＝[real(y′ k(T S))，imag(y′ k(T S))，real(y′ k(2T S))，imag(y′ k(2T S))，...，real(y′ k(T))，imag(y′ k(T))]
    式中，F k表示无线设备k的特征向量，real()表示取实部，imag()表示取虚部，T S为采样时间，k＝1，...，K。
  9. 根据权利要求6所述的噪声增强射频指纹识别装置，其特征在于：所述第一参数获取模块具体包括：
    噪声方差获取单元，用于根据估计的信噪比按照下式计算噪声方差：
    Figure PCTCN2020115323-appb-100008
    式中，SNR a表示估计的信噪比，
    Figure PCTCN2020115323-appb-100009
    表示噪声方差；
    参数获取单元，用于根据噪声方差更新所述马氏距离基础参数，得到马氏距离参数：
    Figure PCTCN2020115323-appb-100010
    式中，∑ *表示马氏距离参数，∑表示马氏距离基础参数，I表示单位矩阵，N表示叠加的数量。
  10. 根据权利要求9所述的噪声增强射频指纹识别装置，其特征在于：所述设备识别模块具体包括：
    马氏距离计算单元，用于根据所述马氏距离参数按照下式计算待识别无线设备的特征向量与所有无线设备型号的模板的马氏距离：
    Figure PCTCN2020115323-appb-100011
    式中，Mah(F test，μ k)表示待识别无线设备的特征向量F test与无线设备型号k的模板μ k的马氏距离，
    Figure PCTCN2020115323-appb-100012
    设备型号确认单元，用于选择最小马氏距离对应的无线设备型号作为待识别无线设备的设备型号。
PCT/CN2020/115323 2019-12-05 2020-09-15 一种噪声增强射频指纹识别方法及装置 WO2021109672A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911232231.4A CN111027614B (zh) 2019-12-05 2019-12-05 一种噪声增强射频指纹识别方法及装置
CN201911232231.4 2019-12-05

Publications (1)

Publication Number Publication Date
WO2021109672A1 true WO2021109672A1 (zh) 2021-06-10

Family

ID=70204298

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/115323 WO2021109672A1 (zh) 2019-12-05 2020-09-15 一种噪声增强射频指纹识别方法及装置

Country Status (2)

Country Link
CN (1) CN111027614B (zh)
WO (1) WO2021109672A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114598518A (zh) * 2022-03-02 2022-06-07 中国人民解放军军事科学院战争研究院 低信噪比环境中物联网射频指纹的识别方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111027614B (zh) * 2019-12-05 2023-07-04 南京东科优信网络安全技术研究院有限公司 一种噪声增强射频指纹识别方法及装置
CN115840104B (zh) * 2023-02-24 2023-04-28 威海赛宝工业信息技术研究院有限公司 一种基于电磁兼容实验的干扰信号识别方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110177786A1 (en) * 2009-03-30 2011-07-21 Stefano Marano Method and system for identification and mitigation of errors in non-line-of-sight distance estimation
CN107392123A (zh) * 2017-07-10 2017-11-24 电子科技大学 一种基于相参积累消噪的射频指纹特征提取和识别方法
CN108809355A (zh) * 2018-07-04 2018-11-13 南京东科优信网络安全技术研究院有限公司 一种在低信噪比情况下提取设备物理指纹特征的方法
CN109921886A (zh) * 2019-01-28 2019-06-21 东南大学 一种鲁棒的低功耗设备射频指纹识别方法
CN109948731A (zh) * 2019-03-29 2019-06-28 成都大学 一种通信电台个体识别方法、系统、存储介质及终端
CN111027614A (zh) * 2019-12-05 2020-04-17 南京东科优信网络安全技术研究院有限公司 一种噪声增强射频指纹识别方法及装置

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040066752A1 (en) * 2002-10-02 2004-04-08 Hughes Michael A. Radio frequency indentification device communications systems, wireless communication devices, wireless communication systems, backscatter communication methods, radio frequency identification device communication methods and a radio frequency identification device
CN102186239B (zh) * 2011-04-13 2014-01-22 福建星网锐捷网络有限公司 射频指纹定位方法、装置及网络设备
CN107368732B (zh) * 2017-07-14 2019-07-23 南京东科优信网络安全技术研究院有限公司 一种基于设备物理指纹特征的目标识别定位系统及方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110177786A1 (en) * 2009-03-30 2011-07-21 Stefano Marano Method and system for identification and mitigation of errors in non-line-of-sight distance estimation
CN107392123A (zh) * 2017-07-10 2017-11-24 电子科技大学 一种基于相参积累消噪的射频指纹特征提取和识别方法
CN108809355A (zh) * 2018-07-04 2018-11-13 南京东科优信网络安全技术研究院有限公司 一种在低信噪比情况下提取设备物理指纹特征的方法
CN109921886A (zh) * 2019-01-28 2019-06-21 东南大学 一种鲁棒的低功耗设备射频指纹识别方法
CN109948731A (zh) * 2019-03-29 2019-06-28 成都大学 一种通信电台个体识别方法、系统、存储介质及终端
CN111027614A (zh) * 2019-12-05 2020-04-17 南京东科优信网络安全技术研究院有限公司 一种噪声增强射频指纹识别方法及装置

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
XU QIANG; ZHENG RONG; SAAD WALID; HAN ZHU: "Device Fingerprinting in Wireless Networks: Challenges and Opportunities", IEEE COMMUNICATIONS SURVEYS & TUTORIALS, IEEE, USA, vol. 18, no. 1, 1 January 1900 (1900-01-01), USA, pages 94 - 104, XP011597192, DOI: 10.1109/COMST.2015.2476338 *
YU, JIABAO ET AL.: "RF Fingerprinting Extraction and Identification of Wireless Communication Devices", JOURNAL OF CRYPTOLOGIC RESEARCH, vol. 3, no. 5, 31 October 2016 (2016-10-31), pages 433 - 446, XP055509585 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114598518A (zh) * 2022-03-02 2022-06-07 中国人民解放军军事科学院战争研究院 低信噪比环境中物联网射频指纹的识别方法

Also Published As

Publication number Publication date
CN111027614A (zh) 2020-04-17
CN111027614B (zh) 2023-07-04

Similar Documents

Publication Publication Date Title
WO2021109672A1 (zh) 一种噪声增强射频指纹识别方法及装置
Xie et al. A generalizable model-and-data driven approach for open-set RFF authentication
CN106130942B (zh) 一种基于循环谱的无线通信信号调制识别及参数估计方法
CN106161297B (zh) Ofdm系统中基于独立分量分析的抗导频欺骗攻击信道估计和识别方法
CN110855591A (zh) 一种基于卷积神经网络结构的qam和psk信号类内调制分类方法
CN110166387B (zh) 一种基于卷积神经网络识别信号调制方式的方法及系统
CN109921886B (zh) 一种鲁棒的低功耗设备射频指纹识别方法
CN113014524B (zh) 一种基于深度学习的数字信号调制识别方法
CN110113288A (zh) 一种基于机器学习的ofdm解调器的设计和解调方法
CN113051628B (zh) 基于残差学习的芯片侧信道攻击降噪预处理方法
CN111601307A (zh) 一种基于瞬态-稳态的部分叠加射频指纹方法
Tian et al. Transfer learning-based radio frequency fingerprint identification using ConvMixer network
Hou et al. DASECount: Domain-agnostic sample-efficient wireless indoor crowd counting via few-shot learning
Peng et al. Specific emitter identification via squeeze-and-excitation neural network in frequency domain
Wang et al. Specific emitter identification based on deep adversarial domain adaptation
CN106341360B (zh) 一种多输入单输出空时分组码系统的分层调制识别方法
CN108834043B (zh) 基于先验知识的压缩感知多目标无源定位方法
CN115664905A (zh) 基于多域物理层指纹特征的Wi-Fi设备识别系统及方法
CN105848200B (zh) 一种td-scdma系统中上行能量测量方法及装置
CN114584441A (zh) 一种基于深度学习的数字信号调制识别方法
CN113242201A (zh) 基于生成分类网络的无线信号增强解调方法及系统
Song et al. A lightweight radio frequency fingerprint extraction scheme for device identification
Zhang et al. Wi-Fi device identification based on multi-domain physical layer fingerprint
CN111917674A (zh) 一种基于深度学习的调制识别方法
CN112153648A (zh) 基于d2d通信技术的移动群智感知可靠数据收集方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20895088

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20895088

Country of ref document: EP

Kind code of ref document: A1