CN114036972A - Radio frequency signal gene feature extraction method based on differential equi-potential star map - Google Patents

Abstract

Translated fromChinese

本发明公开了一种基于差分等势星球图的射频信号基因特征提取方法，包括：S01：采集待识别的通信辐射源个体的射频基带信号；S02：将射频基带信号的进行差分和归一化处理得到差分星座图；S03：根据差分星座图中点密度分布情况，对不同密度区域进行颜色区分，得到差分等势星球图；S04：利用预先训练好的神经网络对等势星球图进行识别，得到基于差分等势星球图特征的通信辐射源个体识别结果。本发明的实质性效果包括：通过神经网络对差分等势星球图进行特征提取和识别，使得通信辐射源个体识别成功率较高，在使用相同的深度卷积神经网络模型架构下，相较于传统的基于星座图的统计图域方法，可在计算效率不降低的前提下一定程度上提高识别准确率。

The invention discloses a radio frequency signal gene feature extraction method based on a differential equipotential planetary map, comprising: S01: collecting radio frequency baseband signals of individual communication radiation sources to be identified; S02: differing and normalizing the radio frequency baseband signals Process to obtain a differential constellation map; S03: According to the density distribution of points in the differential constellation map, color different density areas to obtain a differential equipotential planetary map; S04: Use a pre-trained neural network to identify the equipotential planetary map, The individual identification results of the communication radiation source based on the features of the differential isopotential planetary map are obtained. The substantial effects of the present invention include: extracting and identifying differential isopotential planetary maps through neural networks, so that the success rate of individual identification of communication radiation sources is higher, and using the same deep convolutional neural network model architecture, compared with The traditional statistical graph domain method based on constellation diagram can improve the recognition accuracy to a certain extent without reducing the computational efficiency.

Description

Radio frequency signal gene feature extraction method based on differential equi-potential star map

Technical Field

The invention relates to the field of signal feature extraction, in particular to a radio frequency signal gene feature extraction method based on a differential equipotential sphere diagram.

Background

With the continuous emergence of information security problems brought by wireless communication networks, especially the problems of user identity impersonation, replay attack, equipment cloning and the like. In recent years, attack events (such as load shedding, line overload disconnection, cascading failure and the like of a power grid caused by intelligent data attack) occurring in various countries gradually expose various hidden dangers of the power grid in the aspect of information security. How to accurately identify and authenticate the internet of things object is a first problem faced by the power internet of things and is also a basis for the application of the power internet of things.

The traditional authentication mechanism is realized at an application layer, a numerical result which is difficult to counterfeit by a third party is generated by using a cryptographic algorithm, but the mechanism has the risks of protocol security holes and key leakage. Data attack of illegal access equipment of the power internet of things causes serious interference and threat to the whole network, and the safety of a communication system is difficult to guarantee only by means of a traditional application layer password authentication method, so that the design of an effective physical layer authentication system has important significance. The star map can represent vector end points (symbol points) projected by the modulation signals under specific base vectors, can express two kinds of basic information of amplitude and phase of the signals relative to a carrier at a certain moment, and the projections of the star map on two coordinate axes are two paths of baseband signals at the current moment. The number of symbol points of the digitally modulated signal is limited and all symbol points are represented in the same vector diagram, i.e. constitute a constellation diagram. However, since the constellation map is a binary map, the statistical characteristics are submerged by noise at low snr, and thus, the constellation map cannot be directly used for physical layer authentication.

Disclosure of Invention

Aiming at the problem that the prior art lacks of effective physical layer authentication means, the invention provides a radio frequency signal gene feature extraction method based on a differential equipotential sphere diagram.

The technical scheme of the invention is as follows.

A radio frequency signal gene feature extraction method based on a differential equipotential sphere diagram comprises the following steps:

s01: collecting radio frequency baseband signals of communication radiation source individuals to be identified;

s02: carrying out difference and normalization processing on the radio frequency baseband signals to obtain a difference constellation diagram;

s03: according to the point density distribution condition in the differential constellation diagram, carrying out color distinction on different density areas to obtain a differential equipotential sphere diagram;

s04: and (4) identifying the equipotential star map by utilizing a pre-trained neural network to obtain a communication radiation source individual identification result based on the characteristics of the differential equipotential star map.

Preferably, the process of step S02 includes:

for radio frequency baseband signals

Where t is the sampling point location, x (t) is the transmitter baseband signal, f_tCarrier frequency for the transmitter;

the difference process is represented as:

wherein: d (t) is the differentially processed signal, y^*A conjugate value of y; n is 1; e.g. of the type^-j2πθnIs a phase rotation factor;

r (t) is the signal received by the receiver and r (t) s (t), f_rFor the carrier frequency of the receiver, the frequency of the receiver,

is the phase error when the receiver receives the signal.

Preferably, the process of step S01 includes: and respectively collecting the radio frequency baseband signals of the communication radiation source individuals to be identified by using a frequency spectrograph according to the preset sampling frequency and sampling time.

Preferably, the step S01 further includes a step of performing data expansion on the acquired rf baseband signal: and filtering an effective data transmission section of the signal noise section by a variance track variable point detection algorithm, and slicing the effective data transmission section to obtain a plurality of sliced radio frequency baseband signals.

Preferably, the neural network is a deep convolutional neural network, and the training process is as follows: and (3) separating a part from the differential equipotential planet map as a training sample, using the rest as an identification sample, setting a parameter structure of the deep convolutional neural network, and then introducing the training sample for training to obtain the trained deep convolutional neural network.

Preferably, the color distinguishing the different density areas includes: the shading is performed according to the density difference of the points, and then the brightness of the shading is adjusted according to the density difference, wherein the brightness is reduced along with the increase of the density.

The substantial effects of the invention include: the method has the advantages that the characteristic extraction and identification are carried out on the differential equipotential star map through the neural network, so that the success rate of individual identification of the communication radiation source is high, and under the condition that the same deep convolution neural network model architecture is used, compared with the traditional statistical map domain method based on the constellation map, the identification accuracy can be improved to a certain extent on the premise that the calculation efficiency is not reduced.

Drawings

FIG. 1 is a schematic diagram of a differential equipotential sphere generating process according to an embodiment of the present invention;

FIG. 2 is a feature diagram of the recognition result based on an equipotential sphere map without differential processing according to an embodiment of the present invention;

FIG. 3 is a feature diagram of an identification result based on a difference constellation trajectory diagram according to an embodiment of the present invention;

FIG. 4 is a feature diagram of the recognition result based on the differential equi-potential planet map according to the embodiment of the invention;

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions will be clearly and completely described below with reference to the embodiments, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It should be understood that, in various embodiments of the present invention, the sequence numbers of the processes do not mean the execution sequence, and the execution sequence of the processes should be determined by the functions and the internal logic of the processes, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

It should be understood that in the present application, "comprising" and "having" and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

The technical solution of the present invention will be described in detail below with specific examples. Embodiments may be combined with each other and some details of the same or similar concepts or processes may not be repeated in some embodiments.

Example (b):

a radio frequency signal gene feature extraction method based on a differential equipotential sphere diagram comprises the following steps:

s01: and acquiring radio frequency baseband signals of the communication radiation source individuals to be identified.

And respectively collecting the radio frequency baseband signals of the communication radiation source individuals to be identified by using a frequency spectrograph according to the preset sampling frequency and sampling time. The data expansion step is carried out on the collected radio frequency baseband signals: and filtering an effective data transmission section of the signal noise section by a variance track variable point detection algorithm, and slicing the effective data transmission section to obtain a plurality of sliced radio frequency baseband signals.

S02: and carrying out difference and normalization processing on the radio frequency baseband signals to obtain a difference constellation diagram.

For radio frequency baseband signals

Where t is the sampling point location, x (t) is the transmitter baseband signal, f_tCarrier frequency for the transmitter;

the difference process is represented as:

wherein: d (t) is the differentially processed signal, y^*A conjugate value of y; n is 1; e.g. of the type^-j2πθnIs a phase rotation factor;

r (t) is the signal received by the receiver and r (t) s (t), f_rFor the carrier frequency of the receiver, the frequency of the receiver,

is the phase error when the receiver receives the signal.

S03: and according to the point density distribution condition in the differential constellation diagram, carrying out color discrimination on different density areas to obtain a differential equi-potential star diagram.

And carrying out color differentiation on areas with different densities, wherein the areas with different densities are subjected to gradual coloring according to the density difference of the points, and then the brightness of coloring is adjusted according to the density difference, wherein the brightness is reduced along with the increase of the density.

S04: and (4) identifying the equipotential star map by utilizing a pre-trained neural network to obtain a communication radiation source individual identification result based on the characteristics of the differential equipotential star map.

Wherein, neural network is the deep convolution neural network, and the training process is: and (3) separating a part from the differential equipotential planet map as a training sample, using the rest as an identification sample, setting a parameter structure of the deep convolutional neural network, and then introducing the training sample for training to obtain the trained deep convolutional neural network.

Taking the example of identifying 20 Wi-Fi network card devices of the same manufacturer, the same model, and the same batch, the testing process is as follows.

The baseband signal acquisition equipment is an FSW26 type frequency spectrograph, and the acquisition environment is a laboratory scene. 20 Wi-Fi network card devices are collected, and each device collects 50 samples; the signal sampling frequency is 80MHz, 1.75ms is acquired each time, that is, the number of points per sample is 140000 (taking a single path as an example), wherein the number of points of an effective data transmission section for removing a signal noise section through a variance trajectory point change detection algorithm is 80000 (both are steady-state signals), and then the effective data transmission section is sliced (the number of points is 10000 is a new sample), so that 8 effective data transmission section segments are cut out from each sample, and each segment is taken as a sample, and then each device becomes a total of 50 × 8 ═ 400 samples. At this time, there are a total of 20 × 400 to 8000 samples (after the generation of the equipotential sphere map, 6400 samples are randomly selected to generate training for the deep convolutional neural network, and the remaining 1600 samples are subjected to the identification test, where the number of training samples is 320 and the number of test samples is 80 for each wireless device).

Assuming that the communication radiation source emits radio frequency signals individually

Where t is the sampling point location, x (t) is the transmitter baseband signal, f_tThe carrier frequency is transmitted. If the radio frequency circuit of the communication radiation source is ideal, and the channel is also ideal, the signal r (t) s (t) received by the receiver is obtained.

The receiver carries out down-conversion on the signal to obtain a baseband signal

Wherein is f_rThe carrier frequency of the receiver is selected,

is the phase error when the receiver receives the signal.

When f is_r≠f_tThen, the baseband signal obtained by down-conversion of the receiver is the

Wherein θ ═ f_r-f_t. Since the demodulated signal contains a residual frequency deviation theta, each sample point of the baseband signal has a phase rotation factor e^j2πθt. Since the phase rotation factor varies with the sampling point position t, the constellation diagram as a whole is rotated.

In most coherent demodulation communication systems, estimating the frequency offset and phase offset yields the estimated frequency offset

And phase deviation

The receiver performs frequency deviation and phase deviation compensation on the received signal by using the estimated result, thereby obtaining a stable constellation diagram. In the radio frequency fingerprint extraction method based on the constellation diagram, because the purpose of the receiver is not to accurately demodulate each received signal modulation symbol, the received signals can be subjected to differential processing according to a certain interval n to obtain a stable constellation diagram. The differential processing method comprises

In the formula: d (t) is the signal after the difference processing; y is^*A conjugate value of y; n is 1. The differentially processed signal d (t) also contains a phase rotation factor e^-j2πθn(ii) a However, the phase rotation factor is a constant value and does not change with the change of the positions of the sampling points, so that the new I, Q signals after the difference processing only contain the phase rotation factor with a constant value, and a stable constellation diagram can be obtained without estimating and compensating the carrier frequency deviation and the phase deviation of the receiver. In fig. 1, according to the difference of the point density of the two-dimensional difference constellation diagram, different colors are assigned to different regions, and a one-dimensional signal is converted into a two-dimensional color image (like an ultra high definition X-ray film), which is more fully describedThe subtle features of the signal.

The designed deep convolutional neural network structure is shown in table 1.

Network layer	Parameter structure
		Input layer	227×227×3
Convolutional layer	55×55×96
		Pooling layer	27×27×96
Normalization layer	27×27×96
		Convolutional layer	27×27×256
Pooling layer	13×13×256
		Normalization layer	13×13×256
Convolutional layer	13×13×384
		Convolutional layer	13×13×384
Convolutional layer	13×13×256
		Pooling layer	6×6×256
Normalization layer	6×6×256
		Full connection layer	9 216
Full connection layer	4 096
		Full connection layer	4 096
Output layer	20

TABLE 1 deep convolutional neural network architecture

Through the identification and authentication of the deep convolutional neural network, a communication radiation source individual identification result based on the equipotential star map features without differential processing, a communication radiation source individual identification result based on the differential constellation trajectory map features, and a communication radiation source individual identification result based on the method provided herein are respectively obtained, as shown in fig. 2, 3, and 4.

It can be seen that the success rate of individual identification of the communication radiation source based on the equipotential star map features without differential processing is 90.4%, the success rate of individual identification of the communication radiation source based on the differential constellation trace map features is 88.6%, and the success rate of identification based on the method proposed in this section is 98.6%, which indicates that compared with the conventional statistical map domain method based on the constellation map, the method proposed in this document can greatly improve the identification accuracy without reducing the calculation efficiency (when a 4.0GHz dual-processor notebook computer is used, the average calculation time of each identification of the deep convolutional neural network model does not exceed 20 ms).

Through the description of the above embodiments, those skilled in the art will understand that, for convenience and simplicity of description, only the division of the above functional modules is used as an example, and in practical applications, the above function distribution may be completed by different functional modules according to needs, that is, the internal structure of a specific device is divided into different functional modules to complete all or part of the above described functions.

In the embodiments provided in this application, it should be understood that the disclosed structures and methods may be implemented in other ways. For example, the above-described embodiments with respect to structures are merely illustrative, and for example, a module or a unit may be divided into only one logic function, and may have another division manner in actual implementation, for example, a plurality of units or components may be combined or may be integrated into another structure, or some features may be omitted or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, structures or units, and may be in an electrical, mechanical or other form.

Units described as separate parts may or may not be physically separate, and parts displayed as units may be one physical unit or a plurality of physical units, may be located in one place, or may be distributed to a plurality of different places. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application may be essentially or partially contributed to by the prior art, or all or part of the technical solutions may be embodied in the form of a software product, where the software product is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, or the like) or a processor (processor) to execute all or part of the steps of the methods of the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims (6)

Translated fromChinese

1.一种基于差分等势星球图的射频信号基因特征提取方法，其特征在于，包括以下步骤：1. a radio frequency signal gene feature extraction method based on differential isopotential star map, is characterized in that, comprises the following steps:S01：采集待识别的通信辐射源个体的射频基带信号；S01: Collect the radio frequency baseband signal of the individual communication radiation source to be identified;S02：将射频基带信号的进行差分和归一化处理得到差分星座图；S02: Differential and normalize the radio frequency baseband signal to obtain a differential constellation diagram;S03：根据差分星座图中点密度分布情况，对不同密度区域进行颜色区分，得到差分等势星球图；S03: According to the point density distribution in the differential constellation map, color distinction is made between different density regions, and the differential isopotential planetary map is obtained;S04：利用预先训练好的神经网络对差分等势星球图进行识别，得到基于差分等势星球图特征的通信辐射源个体识别结果。S04: Use the pre-trained neural network to identify the differential isopotential planetary map, and obtain the individual identification result of the communication radiation source based on the feature of the differential isopotential planetary map.2.根据权利要求1所述的一种基于差分等势星球图的射频信号基因特征提取方法，其特征在于，步骤S02的过程包括：2. a kind of radio frequency signal gene feature extraction method based on differential isopotential star map according to claim 1, is characterized in that, the process of step S02 comprises:对于射频基带信号

其中t为采样点位置，x(t)为发射机基带信号，f_t为发射机载波频率；For RF baseband signals

Where t is the sampling point position, x(t) is the baseband signal of the transmitter, and f_t is the carrier frequency of the transmitter;差分处理表示为：Differential processing is expressed as:

其中：d(t)为差分处理后的信号，y^*为y的共轭值；n为1；e^-j2πθn为相位旋转因子；

r(t)为接收机接收到的信号且r(t)＝s(t)，f_r为接收机载波频率，

为接收机接收信号时的相位误差。Where: d(t) is the differentially processed signal, y^* is the conjugate value of y; n is 1; e^-j2πθn is the phase rotation factor;

r(t) is the signal received by the receiver and_r (t)=s(t), fr is the receiver carrier frequency,

is the phase error when the receiver receives the signal.3.根据权利要求1所述的一种基于差分等势星球图的射频信号基因特征提取方法，其特征在于，步骤S01的过程包括：利用频谱仪以预先设定的采样频率和采样时间，分别对待识别的通信辐射源个体的射频基带信号进行采集。3. a kind of radio frequency signal gene feature extraction method based on differential equipotential planetary map according to claim 1, is characterized in that, the process of step S01 comprises: utilize spectrum analyzer with preset sampling frequency and sampling time, respectively. The radio frequency baseband signal of the individual communication radiation source to be identified is collected.4.根据权利要求3所述的一种基于差分等势星球图的射频信号基因特征提取方法，其特征在于，步骤S01中，还包括对采集到的射频基带信号进行数据扩充步骤：通过方差轨迹变点检测算法过滤出信号噪声段的有效数据传输段，再对有效数据传输段进行切片处理，得到切片后的若干射频基带信号。4. a kind of radio frequency signal gene feature extraction method based on differential equipotential planetary map according to claim 3, is characterized in that, in step S01, also comprises carrying out data expansion step to the radio frequency baseband signal collected: by variance trajectory The change point detection algorithm filters out the effective data transmission section of the signal noise section, and then slices the effective data transmission section to obtain several radio frequency baseband signals after slicing.5.根据权利要求1所述的一种基于差分等势星球图的射频信号基因特征提取方法，其特征在于，所述神经网络为深度卷积神经网络，训练过程为：从差分等势星球图中分出部分作为训练样本，其余作为识别样本，设置深度卷积神经网络的参数结构，随后导入训练样本进行训练，得到训练后的深度卷积神经网络。5. a kind of radio frequency signal gene feature extraction method based on differential equipotential planetary map according to claim 1, is characterized in that, described neural network is deep convolutional neural network, and the training process is: from differential equipotential planetary map Part of it is used as training samples, and the rest are used as identification samples. The parameter structure of the deep convolutional neural network is set, and then the training samples are imported for training to obtain the trained deep convolutional neural network.6.根据权利要求1所述的一种基于差分等势星球图的射频信号基因特征提取方法，其特征在于，所述对不同密度区域进行颜色区分，包括：根据点的密度差别进行渐变着色，然后再根据密度差别调整着色的亮度，其中亮度随密度增加而降低。6. A radio frequency signal gene feature extraction method based on a differential isopotential planetary map according to claim 1, wherein the color distinction of different density regions comprises: performing gradient coloring according to the density difference of points, The brightness of the shading is then adjusted according to the density difference, where the brightness decreases as the density increases.

Images