Radio frequency fingerprint decoupling method and apparatus based on channel non-coherence time interval

Abstract

The application discloses a radio frequency fingerprint decoupling method and device based on channel incoherent time interval, the method comprises the following steps: acquiring a frame signal of a device in one environment, marking a label and storing the frame signal as a local signal frame; acquiring another frame signal of the device in another environment and storing the frame signal as an actual signal frame, wherein the interval between the start time of the two frames is greater than the minimum incoherent time of the channel; performing synchronization operation on the local signal frame and the actual signal frame; dividing the signal frame into symbols according to the symbol length; transforming the local symbols and the actual symbols to the frequency domain after energy normalization; taking logarithm of the two frequency domain complex numbers respectively and performing zero mean processing; calculating the cyclic cross-correlation result of the complex sequence and performing low-pass filtering and transforming to the time domain; taking the mean value of the low-frequency time domain results of all actual symbols in a frame signal to obtain the radio frequency fingerprint of the device. The radio frequency fingerprint obtained by the method has robustness in resisting channel changes.

Description

Technical Field

[0001] This invention relates to the field of communication and information security technology, specifically to a radio frequency fingerprint decoupling method and apparatus based on channel incoherent time intervals. Background Technology

[0002] With the continuous development of IoT technology, various industries have widely developed various IoT solutions. The large number of IoT devices exposed in open environments brings significant uncertainty to IoT security. IoT device identification technology can effectively determine the legitimacy of IoT device identities, preventing unauthorized devices from accessing the IoT and ensuring IoT security from the network edge. However, the low cost of IoT devices means limited storage and computing resources, making the development of lightweight security solutions under resource constraints extremely challenging.

[0003] Radio frequency (RF) fingerprints reflect the hardware differences between the RF modules of IoT devices, enabling the differentiation of different IoT devices at the wireless physical layer. RF fingerprints are embedded in the signals emitted by the device. These signals are collected by external devices for further analysis and processing, thereby extracting the RF fingerprint for each device. Due to its uniqueness, non-cloning properties, and long-term stability, and because all processing is performed outside the IoT device, RF fingerprinting is considered the best solution for identifying IoT devices.

[0004] In recent years, radio frequency fingerprint extraction and recognition technologies have become increasingly sophisticated, and the accuracy of device identification has been continuously improving. Most of these methods were trained and tested under ideal conditions, such as a constant or similar channel. However, in real-world applications, channel variations are unpredictable and uncontrollable. Furthermore, the influence of the channel on the signal is unavoidable, and this influence is similar to that of radio frequency fingerprints, making the two intertwined and difficult to distinguish. When the channel changes, radio frequency fingerprints containing channel information can cause a significant drop in device identification accuracy. Therefore, how to remove channel information has become a critical challenge that radio frequency fingerprint extraction technology needs to address. Summary of the Invention

[0005] Purpose of the invention: In order to solve the above-mentioned problems existing in the current radio frequency fingerprint extraction technology, the present invention utilizes the uncorrelation between channels during the channel incoherence time, and achieves decoupling of radio frequency fingerprint from channel by calculating the cross-correlation result between two signals, thereby eliminating channel information in device features and extracting pure radio frequency fingerprint.

[0006] Technical solution: To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0007] Firstly, a radio frequency fingerprint decoupling method based on channel incoherent time intervals is provided, comprising the following steps:

[0008] In one environment, a frame of signal from the device is acquired, labeled, and stored as a local signal frame. In another environment, another frame of signal from the device is acquired and stored as an actual signal frame. The interval between the start times of the local signal frame and the actual signal frame is greater than the minimum incoherence time δ of the channel.

[0009] Synchronization operations are performed on the acquired local signal frames and the actual signal frames to compensate for carrier frequency offset and carrier phase offset;

[0010] The signal frame is divided according to the symbol length. The local signal frame and the actual signal frame are divided into multiple symbols, and the local symbols and the actual symbols are selected.

[0011] After normalizing the energy of the local symbol and the actual symbol, transform them to the frequency domain. Take the logarithm of the two frequency domain complex numbers and perform zero-mean processing to obtain the zero-mean complex number sequence of the two symbols.

[0012] Calculate the cyclic cross-correlation result of the zero-mean complex sequences of the two symbols, perform low-pass filtering to retain the low-frequency part, and then transform to the time domain;

[0013] The RF fingerprint of the device is obtained by averaging the low-frequency time-domain results of all actual symbols within a frame of signal.

[0014] Preferably, the local symbol is one of the multiple symbols divided from the local signal frame, or the average of repeated symbols in the signal preamble after being superimposed in the time domain; the actual symbol is one of the multiple symbols divided from the actual signal frame.

[0015] Preferably, energy normalization of local symbols and actual symbols includes: dividing each sample point in the symbol by the square root of the symbol's average energy to normalize the symbol energy.

[0016] Preferably, the transformation to the frequency domain is performed using a Fourier transform to obtain a complex result in the frequency domain; the transformation to the time domain is performed using an inverse Fourier transform.

[0017] Preferably, the zero-mean processing includes: calculating the mean of the sequence after complex logarithmic transformation in the frequency domain, and then subtracting the mean from the value after complex logarithmic transformation in the frequency domain to achieve zero-mean transformation of the complex logarithm, wherein the complex logarithmic operation is expressed as:

[0018] log cp (V) = log(abs(V)) + 1j*angle(V)

[0019] log is the logarithmic operation defined in the positive real number field, abs is the operation of taking the magnitude of a complex number, angle is the operation of taking the phase angle of a complex number, V represents a complex number, and j represents the imaginary unit.

[0020] Preferably, the cyclic cross-correlation operation includes:

[0021] Let R be the zero-mean sequence of complex numbers with local symbols. K =[r1,r2,…,r K The actual symbolic zero-mean complex sequence is Q. K =[q1,q2,…,q K The cyclic cross-correlation result between the two is expressed as follows:

[0022]

[0023] Where · represents complex multiplication, * represents complex conjugation, K represents sequence length, and ( )) K This represents the principal value interval of the new sequence generated after periodically extending the sequence with a period of length K.

[0024] Preferably, when performing low-pass filtering on the cyclic cross-correlation results, the selection of the filter frequency range is based on the position of the main lobe of the spectrum.

[0025] Secondly, a radio frequency fingerprint decoupling device based on channel incoherent time intervals is provided, comprising:

[0026] The signal acquisition module is used to acquire a frame of signal from the device in one environment, label it and store it as a local signal frame, and acquire another frame of signal from the device in another environment and store it as an actual signal frame. The interval between the frame start time of the local signal frame and the actual signal frame is greater than the minimum incoherence time δ of the channel.

[0027] The frame synchronization module is used to synchronize the acquired local signal frame with the actual signal frame and to compensate for carrier frequency offset and carrier phase offset.

[0028] The symbol segmentation module is used to segment signal frames according to symbol length, dividing local signal frames and actual signal frames into multiple symbols, and selecting local symbols and actual symbols.

[0029] The first processing module is used to transform the local symbol and the actual symbol into the frequency domain after energy normalization, and to take the logarithm of the two frequency domain complex numbers and perform zero-mean processing to obtain the zero-mean complex number sequence of the two symbols.

[0030] The second processing module is used to calculate the cyclic cross-correlation result of the zero-mean complex sequence of the two symbols, perform low-pass filtering, retain the low-frequency part, and then transform it to the time domain.

[0031] The fingerprint recognition module is used to take the average of the low-frequency time-domain results obtained by the first processing module and the second processing module for all actual symbols in a frame of signal to obtain the radio frequency fingerprint of the device.

[0032] Thirdly, the present invention also provides a computer device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, wherein when the programs are executed by the processors, they implement the steps of the radio frequency fingerprint decoupling method based on channel incoherent time intervals as described above.

[0033] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the radio frequency fingerprint decoupling method based on channel incoherent time intervals as described above.

[0034] Beneficial effects:

[0035] 1. This invention utilizes the lack of correlation between channels during the channel incoherence time, and removes channel information from device features through cyclic cross-correlation operations, thereby decoupling and extracting a pure radio frequency fingerprint.

[0036] 2. This invention utilizes logarithmic operations defined in the complex number domain to achieve complete preservation of signal amplitude and phase information during the transformation process, combined with low-pass filtering to suppress noise interference, thereby preserving radio frequency fingerprint information to the greatest extent.

[0037] 3. This invention utilizes 36 ZigBee devices to verify the results in four different scenarios, achieving an accuracy rate of over 99%. The decoupled RF fingerprint exhibits robustness against channel variations. Attached Figure Description

[0038] Figure 1 This is a flowchart of the radio frequency fingerprint decoupling method based on channel incoherent time intervals according to an embodiment of the present invention;

[0039] Figure 2 These are schematic diagrams illustrating four different categories of experimental scenarios in embodiments of the present invention;

[0040] Figure 3 The cross-correlation results (amplitude part) of a signal frame between two devices under different experimental scenarios are shown.

[0041] Figure 4 for Figure 3 A schematic diagram of the main lobe position (amplitude portion) of the cross-correlation results in the embodiment;

[0042] Figure 5 for Figure 3 Implementation Examples Figure 4 The cyclic cross-correlation result (amplitude part) after main lobe filtering is retained;

[0043] Figure 6 for Figure 3Example: Radio frequency fingerprint results (amplitude portion) of signal frames from two devices. Detailed Implementation

[0044] To provide a clearer understanding of the features and advantages of the technical solution of the present invention, the composition and implementation of the specific solution are described below in conjunction with the accompanying drawings.

[0045] Reference Figure 1 The radio frequency fingerprint decoupling method based on channel incoherence time interval disclosed in this invention includes the following steps:

[0046] Step 101: In one environment, acquire a frame of signal from a known device, label it, and store it as a local signal frame; in another environment, acquire another frame of signal from the known device and store it as an actual signal frame.

[0047] The acquired signal frame y(n) satisfies the following in the time domain:

[0048]

[0049] Where x(n) is an ideal signal, h TR h(n) is the RF fingerprint of the device being collected (specifically, the time-domain response of the transmit / receive equivalent filter), h(n) is the channel impulse response, and z(n) is the additive noise. This represents the convolution operation.

[0050] Furthermore, the time interval between the local signal frame y1(n) starting at time T1 and the actual signal frame y2(n) starting at time T2 is greater than the minimum incoherence time δ of the channel, satisfying:

[0051]

[0052]

[0053] |T1-T2|>δ.

[0054] The minimum incoherence time δ of the channel was obtained through field measurements.

[0055] In this embodiment of the invention, signal frames transmitted by 36 ZigBee devices were collected in four scenarios. For example... Figure 2 As shown, signal frames from 36 ZigBee devices were acquired using the USRP N210 in four scenarios: outdoor, indoor, line-of-sight, and non-line-of-sight. Each signal frame contains 266 symbols, and the data segment is random data. The length of each ZigBee symbol is 16 microseconds, and at a sampling rate of 10 MS / s, the number of sampling points per symbol is 160.

[0056] In this embodiment, since the same batch of devices acquires signal frames in four scenarios, a signal frame from a specific device in one scenario can be selected as the local signal frame, and a signal frame from any (can be the same) device in the other three scenarios can be selected as the actual signal frame. Radio frequency fingerprints reflect device characteristics, and there should be differences between the radio frequency fingerprints of different devices to distinguish them. To verify that the radio frequency fingerprints extracted by this method do indeed differ between different devices, this embodiment uses one local signal frame and two actual signal frames for calculation. For ease of explanation, a signal frame from device 1 in the indoor line-of-sight scenario is taken as the local signal frame y. lo (n), one signal frame each from device 1 and device 2 in an indoor non-line-of-sight scenario is taken as the actual signal frame y. re (n).

[0057] Step 102: Synchronize all local signal frames with the actual signal frames to compensate for carrier frequency offset and carrier phase offset.

[0058] After synchronizing the local signal frame with the actual signal frame, carrier frequency offset and carrier phase offset are estimated and compensated. Any effective frame synchronization method, frequency offset, and phase offset compensation method can be used; this invention does not impose any limitations on this.

[0059] Step 103: Segment the signal frame and select local symbols and actual symbols.

[0060] In the embodiments of this invention, the signal frame is segmented according to the symbol length, then y lo (n) and y re (n) can be represented as:

[0061]

[0062]

[0063] in, Indicates y lo The i-th local symbol in (n) Indicates y re The t-th actual symbol in (n). Typically, any symbol within the local signal frame is selected as the local symbol, or the average of repeated symbols in the preamble is calculated in the time domain to enhance the RF fingerprint information in the local symbol. All symbols in the actual signal frame are used as actual symbols to enhance the RF fingerprint characteristics.

[0064] Step 104: Normalize the energy of the two time-domain symbols and transform them to the frequency domain, then take the logarithm and perform zero-mean processing.

[0065] Energy normalization is achieved by dividing each sample point in the symbol by the square root of the symbol's average energy. Specifically:

[0066]

[0067]

[0068] in, It is the local symbol after energy normalization. It is the actual symbol after energy normalization. Here, abs is the modulus of each complex number in the complex number sequence, generating a real number sequence.

[0069] Next, Fourier transforms are performed on the local symbol and the actual symbol respectively to obtain the result in complex form, as follows:

[0070]

[0071]

[0072] Where FFT stands for Fourier Transform. yes The complex result after transformation to the frequency domain, yes The complex result after transformation to the frequency domain.

[0073] From the perspective of signal representation, the result of the Fourier transform is preserved in complex form, specifically as follows:

[0074] Y1(k)=X1(k)H TR (k)H1(k)+Z1(k),

[0075] Y2(k)=X2(k)H TR (k)H2(k)+Z2(k).

[0076] When the signal-to-noise ratio is high enough, for example, greater than 20dB, and noise is ignored, the above two equations are approximately:

[0077] Y1(k)=X1(k)H TR (k)H1(k),

[0078] Y2(k)=X2(k)H TR (k)H2(k).

[0079] Next, taking the logarithm of the complex result, we have:

[0080]

[0081]

[0082] The above is obtained after Fourier transform. and It is a sequence of complex numbers. Performing the complex logarithm operation on a sequence of complex numbers involves performing the complex logarithm operation on each of the complex numbers, ultimately generating a new sequence. Let V represent this. and A complex number, the logarithm of a complex number. cp Specifically:

[0083] log cp (V)=log(abs(V))+1j*angle(V),

[0084] Where log is the logarithm operation defined over the positive real number domain, abs is the magnitude operation for complex numbers, angle is the phase angle operation for complex numbers, and j represents the imaginary unit. In this embodiment of the invention, the final complex mean is calculated and subtracted to achieve zero mean, expressed as:

[0085]

[0086]

[0087] Here, mean represents the calculation of the mean of a complex sequence.

[0088] Step 106: Calculate the cyclic cross-correlation result of the two and filter it to retain the low-frequency part.

[0089] Cyclic cross-correlation operations are performed on complex sequences, specifically: two complex sequences R K =[r1,r2,…,r K ] and Q K =[q1,q2,…,q K If the cyclic cross-correlation result of the two is expressed as:

[0090]

[0091] Where · represents complex multiplication, * represents complex conjugation, K represents sequence length, and ( )) K This represents the principal value interval of the new sequence generated by periodically extending the sequence with a period of length K. Furthermore, cyclic cross-correlation operations possess linear properties, namely, they satisfy homogeneity and additivity, specifically:

[0092] Homogeneity:

[0093] Additivity:

[0094]

[0095] Where c is an arbitrary constant, and P(k) and O(k) are arbitrary complex number sequences of length equal to R(k) and Q(k).

[0096] by and This represents the complex logarithm of the local signal and the actual signal. and The decoupling of RF fingerprinting and channel is achieved through cyclic cross-correlation operations, specifically:

[0097]

[0098] Since the start time interval between the two acquired signal frames is greater than the minimum incoherence time δ of the channel, therefore and If the time interval is greater than the minimum incoherence time δ of the channel, then:

[0099]

[0100] therefore, It only contains radio frequency fingerprint information Not including and Channel information enables the decoupling of radio frequency fingerprinting from the channel. If and If they are unrelated, then Further:

[0101]

[0102] The above describes the calculation process of complex cyclic cross-correlation. Based on the above derivation process, in this embodiment of the invention, the result calculated according to step 105... and Obtained by complex cyclic cross-correlation operation and Cross-correlation results The results of some symbols are as follows: Figure 3 As shown, the cross-correlation result obtained by calculating the signal frames of device 1 in the indoor line-of-sight scenario and the signal frames of device 1 in the indoor non-line-of-sight scenario is significantly different from the cross-correlation result obtained by calculating the signal frames of device 1 in the indoor line-of-sight scenario and the signal frames of device 2 in the indoor non-line-of-sight scenario.

[0103] Furthermore, when performing low-pass filtering on the cyclic cross-correlation results, the selection of the filter frequency range is based on the position of the main lobe in the spectrum, as expressed as:

[0104]

[0105]

[0106]

[0107] Among them, FFT returns the complex result after Fourier transform of the sequence, the Index function returns the subscript range D of the main lobe of the complex sequence, [D] means setting the sequence elements whose subscripts are not in D to zero and generating a complex sequence of the same length, and IFFT returns the time-domain complex sequence after frequency domain filtering.

[0108] In this embodiment of the invention, the complex cross-correlation results are... Using Fourier transform to the frequency domain, we have:

[0109]

[0110] The corresponding amplitude results are as follows Figure 4 As shown. According to Figure 4 The frequency range corresponding to the main lobe position in the mid-spectral frequency spectrum is obtained and used as the passband range of the low-pass filter. The complex cross-correlation result is passed through the designed low-pass filter to filter out all other frequency components. Then, using inverse Fourier transform, the filtered complex cross-correlation result is transformed back to the time domain to obtain, as shown below. Figure 5 The low-pass filtering result shown is:

[0111]

[0112] in, It is the frequency domain result of the complex cross-correlation result after passing through a low-pass filter.

[0113] Step 107: Take the average of the results of all symbols in the actual signal frame to obtain the radio frequency fingerprint.

[0114] The decoupled RF fingerprint is obtained by averaging the complex numbers obtained from the above operations on all actual symbols, specifically:

[0115]

[0116] Where N is the total number of actual symbols, This represents the complex result obtained by performing the above operation on the t-th actual symbol within the actual signal frame, which is the decoupled RF fingerprint.

[0117] In this embodiment of the invention, the RF fingerprint of the device is obtained by averaging the results of all symbols within a frame of signal, specifically as follows:

[0118]

[0119] The results of partial signal frames are as follows Figure 6 As shown. After summing and taking the average, Figure 6 The RF fingerprint results of a single frame of signal are compared to Figure 3 The RF fingerprint results for multiple symbols within a single frame of signal are more stable, and the distinguishability is significantly improved.

[0120] Radio frequency (RF) fingerprints reflect the physical characteristics of a device. RF fingerprints extracted from signals emitted by the same device are very similar, while those between different devices vary significantly. In this embodiment, 36 ZigBee devices were used for verification in four different scenarios, achieving an accuracy rate of over 99%. The decoupled RF fingerprints are robust against channel variations.

[0121] Based on the same technical concept as the method embodiments, the present invention also provides a radio frequency fingerprint decoupling device based on channel incoherent time intervals, comprising:

[0122] The signal acquisition module is used to acquire a frame of signal from the device in one environment, label it and store it as a local signal frame, and acquire another frame of signal from the device in another environment and store it as an actual signal frame. The interval between the frame start time of the local signal frame and the actual signal frame is greater than the minimum incoherence time δ of the channel.

[0123] The frame synchronization module is used to synchronize the acquired local signal frame with the actual signal frame and to compensate for carrier frequency offset and carrier phase offset.

[0124] The symbol segmentation module is used to segment signal frames according to symbol length, dividing local signal frames and actual signal frames into multiple symbols, and selecting local symbols and actual symbols.

[0125] The first processing module is used to transform the local symbol and the actual symbol into the frequency domain after energy normalization, and to take the logarithm of the two frequency domain complex numbers and perform zero-mean processing to obtain the zero-mean complex number sequence of the two symbols.

[0126] The second processing module is used to calculate the cyclic cross-correlation result of the zero-mean complex sequence of the two symbols, perform low-pass filtering, retain the low-frequency part, and then transform it to the time domain.

[0127] The fingerprint recognition module is used to take the average of the low-frequency time-domain results obtained by the first processing module and the second processing module for all actual symbols in a frame of signal to obtain the radio frequency fingerprint of the device.

[0128] It should be understood that the radio frequency fingerprint decoupling device based on channel incoherent time interval can implement all the technical solutions in the above method embodiments. The functions of each functional module can be specifically implemented according to the methods in the above method embodiments. The specific implementation process can be referred to the relevant descriptions in the above embodiments, which will not be repeated here.

[0129] The present invention also provides a computer device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, wherein when the programs are executed by the processors, they implement the steps of the radio frequency fingerprint decoupling method based on channel incoherent time intervals as described above.

[0130] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the radio frequency fingerprint decoupling method based on channel incoherent time intervals as described above.

[0131] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus (systems), computer devices, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0132] This invention is described with reference to a flowchart of a method according to embodiments of the invention. It should be understood that each step in the flowchart and combinations thereof can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the process. Figure 1 A device for a function specified in one or more processes.

[0133] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 The function specified in one or more processes.

[0134] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 Steps of a specified function in one or more processes.

Claims

1. A radio frequency fingerprint decoupling method based on channel incoherent time intervals, characterized in that, Includes the following steps: In one environment, a frame of signal from the device is acquired, labeled, and stored as a local signal frame. In another environment, another frame of signal from the device is acquired and stored as an actual signal frame. The interval between the start times of the local signal frame and the actual signal frame is greater than the minimum incoherence time δ of the channel. Synchronization operations are performed on the acquired local signal frames and the actual signal frames to compensate for carrier frequency offset and carrier phase offset; The signal frame is divided according to the symbol length. The local signal frame and the actual signal frame are divided into multiple symbols, and the local symbols and the actual symbols are selected. After normalizing the energy of the local symbol and the actual symbol, transform them to the frequency domain. Take the logarithm of the two frequency domain complex numbers and perform zero-mean processing to obtain the zero-mean complex number sequence of the two symbols. Calculate the cyclic cross-correlation result of the zero-mean complex sequences of the two symbols, perform low-pass filtering to retain the low-frequency part, and then transform to the time domain; The RF fingerprint of the device is obtained by averaging the low-frequency time-domain results of all actual symbols within a frame of signal.

2. The method according to claim 1, characterized in that, A local symbol is one of the multiple symbols divided into a local signal frame, or it is the average of repeated symbols in the signal preamble after being superimposed in the time domain; an actual symbol is one of the multiple symbols divided into an actual signal frame.

3. The method according to claim 1, characterized in that, Energy normalization for local symbols and actual symbols includes dividing each sample point in the symbol by the square root of the symbol's average energy to normalize the symbol's energy.

4. The method according to claim 1, characterized in that, The transformation to the frequency domain is performed using a Fourier transform, yielding a complex result in the frequency domain; the transformation to the time domain is performed using an inverse Fourier transform.

5. The method according to claim 1, characterized in that, Zero-mean normalization involves: calculating the mean of the sequence after complex logarithmic transformation in the frequency domain, and then subtracting this mean from the value after complex logarithmic transformation in the frequency domain to achieve zero-mean normalization of the complex logarithm. The complex logarithmic operation is represented as: log cp (V)=log(abs(V))+1j*angle(V) log is the logarithmic operation defined in the positive real number field, abs is the operation of taking the magnitude of a complex number, angle is the operation of taking the phase angle of a complex number, V represents a complex number, and j represents the imaginary unit.

6. The method according to claim 1, characterized in that, Cyclic cross-correlation operations include: Let R be the zero-mean sequence of complex numbers with local symbols. K =[r1,r2,…,r K The actual symbolic zero-mean complex sequence is Q. K =[q1,q2,…,q K The cyclic cross-correlation result between the two is expressed as follows: Where · represents complex multiplication, * represents complex conjugate operation, K represents sequence length, (()) K This represents the principal value interval of the new sequence generated after periodically extending the sequence with a period of length K.

7. The method according to claim 6, characterized in that, When performing low-pass filtering on the cyclic cross-correlation results, the selection of the filter frequency range is based on the position of the main lobe in the spectrum.

8. A radio frequency fingerprint decoupling device based on channel incoherent time intervals, characterized in that, include: The signal acquisition module is used to acquire a frame of signal from the device in one environment, label it and store it as a local signal frame, and acquire another frame of signal from the device in another environment and store it as an actual signal frame. The interval between the frame start time of the local signal frame and the actual signal frame is greater than the minimum incoherence time δ of the channel. The frame synchronization module is used to synchronize the acquired local signal frame with the actual signal frame and to compensate for carrier frequency offset and carrier phase offset. The symbol segmentation module is used to segment signal frames according to symbol length, dividing local signal frames and actual signal frames into multiple symbols, and selecting local symbols and actual symbols. The first processing module is used to transform the local symbol and the actual symbol into the frequency domain after energy normalization, and to take the logarithm of the two frequency domain complex numbers and perform zero-mean processing to obtain the zero-mean complex number sequence of the two symbols. The second processing module is used to calculate the cyclic cross-correlation result of the zero-mean complex sequence of the two symbols, perform low-pass filtering, retain the low-frequency part, and then transform it to the time domain. The fingerprint recognition module is used to take the average of the low-frequency time-domain results obtained by the first processing module and the second processing module for all actual symbols in a frame of signal to obtain the radio frequency fingerprint of the device.

9. A computer device, characterized in that, include: One or more processors; Memory; as well as One or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, wherein when the programs are executed by the processors, they implement the steps of the radio frequency fingerprint decoupling method based on channel incoherent time intervals as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the radio frequency fingerprint decoupling method based on channel incoherent time intervals as described in any one of claims 1-7.

Images