Radio frequency fingerprint hiding method based on feature compensation

By compensating for carrier frequency offset, I/Q imbalance, and PA nonlinear distortion of the signal to be transmitted at the transmitter end, a joint compensation model is constructed, which solves the problem that it is difficult to completely eliminate radio frequency fingerprints at the receiver end, realizes radio frequency fingerprint hiding, and improves the privacy protection and anti-identification capability of the communication system.

CN120897195APending Publication Date: 2025-11-04XIDIAN UNIV

Patent Information

Application Number
CN202511004959.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-21
Publication Date
2025-11-04

AI Technical Summary

Technical Problem

Existing technologies that rely on signal compensation methods at the receiver end are insufficient to completely eliminate radio frequency fingerprints. Attackers can still identify them using unaffected features, and this also affects the normal transmission performance of the communication system.

Method used

At the transmitter end, a joint compensation model is used to compensate for carrier frequency offset, I/Q imbalance and PA nonlinear distortion of the signal to be transmitted. The joint compensation model is constructed by carrier frequency offset estimation and compensation, I/Q imbalance estimation and compensation and PA nonlinear distortion estimation to achieve radio frequency fingerprint hiding.

Benefits of technology

It effectively reduces the classification accuracy of RFFI technology, enhances the privacy protection and anti-identification capabilities of the transmitter, and ensures the normal transmission performance of the communication system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120897195A_ABST
    Figure CN120897195A_ABST
Patent Text Reader

Abstract

The invention discloses a radio frequency fingerprint hiding method based on feature compensation. The method comprises the following steps: at a transmitter end, carrying out carrier frequency offset compensation, I / Q imbalance compensation and PA nonlinear distortion compensation on a to-be-transmitted signal in sequence by using a preset joint compensation model to obtain a transmitted signal after radio frequency fingerprint hiding; the joint compensation model is constructed by the following steps: carrying out carrier frequency offset estimation and compensation, I / Q imbalance estimation and compensation and PA nonlinear distortion estimation on a received signal of a receiver in sequence, and constructing the joint compensation model according to a carrier frequency offset estimation result, an I / Q imbalance estimation result and a PA nonlinear distortion estimation result. According to the method, a scientific and reasonable parameter estimation and compensation sequence is designed, an illegal receiver is effectively prevented from extracting radio frequency fingerprints from received signals, the classification precision of the RFFI technology can be effectively reduced, and the privacy protection and recognition resistance of a transmitter are remarkably enhanced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of wireless device communication security technology, and specifically relates to a radio frequency fingerprint hiding method based on feature compensation. Background Technology

[0002] Radio Frequency Fingerprint Identification (RFFI) is an identification technology based on physical layer feature extraction and classification, which has received widespread attention in the field of wireless communication security in recent years. Its basic principle is to identify the wireless transmitter by utilizing inherent impairments caused by its hardware characteristics, such as carrier frequency offset, in-phase / quadrature (I / Q) imbalance, and power amplifier (PA) nonlinear distortion. These impairments are unique to different devices and can be considered as the "radio frequency fingerprint" of the wireless device. By extracting and classifying features from the received wireless signal, RFFI technology can identify the transmitter without the need for additional equipment, thereby achieving security functions such as device authentication and intrusion detection at the physical layer. Compared to traditional cryptographic authentication methods, RFFI has advantages such as no need for key management and difficulty in forgery, thus showing broad application prospects in scenarios such as the Internet of Things (IoT) and drone communication.

[0003] However, because RFFI technology can effectively distinguish between different transmitters, its widespread application brings potential security risks. Since RFFI relies solely on the inherent radio frequency characteristics of a device, this means that not only legitimate devices can use this technology for authentication, but attackers can also use similar methods for device tracking, identity spoofing, and even eavesdropping. Furthermore, in scenarios with high privacy and security requirements, such as anonymous networks and privacy-preserving communications, devices may not want their radio frequency fingerprints to be tracked or identified. Therefore, how to hide radio frequency fingerprint information has become an important research problem. The proposed radio frequency fingerprint hiding technology aims to address this challenge. Its core idea is to weaken or erase the device's inherent radio frequency fingerprint, making it difficult for the receiver to accurately identify the transmitter, thereby improving the concealment and security of communication.

[0004] Currently, radio frequency (RF) fingerprint processing methods mainly focus on the receiver end, employing signal compensation or feature perturbation to reduce the impact of RF fingerprints. For example, some research has compensated for distortions introduced by transmitter hardware damage by equalizing, filtering, or using neural networks to process the received signal at the receiver. However, this approach has significant limitations: first, it relies on the receiver's processing power, while attackers can still use unaffected features for identification; second, since the generation mechanism of RF fingerprints mainly originates from inherent damage at the transmitter end, post-processing methods at the receiver end are insufficient to completely eliminate this information. Therefore, relying solely on receiver-end processing cannot fundamentally solve the problem of RF fingerprint concealment.

[0005] Therefore, in order to overcome the above limitations, there is an urgent need for a radio frequency fingerprint hiding method applied to the transmitter end, which can effectively reduce the classification accuracy of RFFI technology without affecting the normal transmission performance of the communication system. Summary of the Invention

[0006] To address the aforementioned problems in the prior art, this invention provides a radio frequency fingerprint hiding method based on feature compensation.

[0007] The technical problem to be solved by this invention is achieved through the following technical solution: This invention provides a radio frequency fingerprint hiding method based on feature compensation, comprising: At the transmitter end, a preset joint compensation model is used to sequentially perform carrier frequency offset compensation, I / Q imbalance compensation and PA nonlinear distortion compensation on the signal to be transmitted to obtain the transmitted signal after the radio frequency fingerprint is hidden. The joint compensation model is constructed in the following manner: The received signal of the receiver is sequentially subjected to carrier frequency offset estimation and compensation, I / Q imbalance estimation and compensation, and PA nonlinear distortion estimation. A joint compensation model is then constructed based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results.

[0008] Optionally, the step of sequentially performing carrier frequency offset estimation and compensation, I / Q imbalance estimation and compensation, and PA nonlinear distortion estimation on the received signal of the receiver, and constructing a joint compensation model based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results, includes: Based on the received signal from the receiver, the carrier frequency offset is estimated using Fast Fourier Transform. A carrier frequency offset compensation model is constructed based on the carrier frequency offset estimation result, and the carrier frequency offset compensation model is used to compensate the received signal for the carrier frequency offset to obtain the carrier frequency offset compensated signal. The I / Q imbalance is estimated for the carrier frequency offset compensated signal based on the adaptive generalized linear filtering method. An I / Q imbalance compensation model is constructed based on the I / Q imbalance estimation results. The I / Q imbalance compensation model is then used to compensate the carrier frequency offset compensated signal for I / Q imbalance, thus obtaining the I / Q imbalance compensated signal. Based on the signal after I / Q imbalance compensation, the generalized memory polynomial method is used to estimate the PA nonlinear distortion, and a PA nonlinear distortion compensation model is constructed based on the PA nonlinear distortion estimation results. The joint compensation model is constructed based on the carrier frequency offset compensation model, the I / Q imbalance compensation model, and the PA nonlinear distortion compensation model.

[0009] Optionally, the carrier frequency offset compensation model is expressed as: ; in, Represents a discrete signal to be compensated. This indicates the carrier frequency offset estimation result. Represents a discrete-time index. Indicates the sampling interval. j Represents the imaginary unit. This indicates the signal after carrier frequency offset compensation.

[0010] Optionally, the I / Q imbalance compensation model is expressed as: ; in, Represents a discrete signal to be compensated. This represents the signal after I / Q imbalance compensation. This represents the gain imbalance estimate in the I / Q imbalance estimation results. This represents the phase imbalance estimate in the I / Q imbalance estimation results. express The real part, express The imaginary part, j It represents the imaginary unit.

[0011] Optionally, the PA nonlinear distortion compensation model is expressed as: ; in, Indicates delay Discrete signal to be compensated at each sampling point , This represents the signal after PA nonlinear distortion compensation. The memory depth represents the memory polynomial. This represents the nonlinear order of the memory polynomial. Indicates the first The first delay, the first The compensation coefficients corresponding to the first-order nonlinearity.

[0012] Optionally, the memory depth of the memory polynomial The nonlinear order of the memory polynomial is set to 3. Set it to 5.

[0013] Alternatively, the joint compensation model can be expressed as: ; in, Represents a discrete signal to be compensated. This represents the carrier frequency offset compensation model. This represents the I / Q imbalance compensation model. This represents the PA nonlinear distortion compensation model. This indicates the signal after joint compensation.

[0014] Optionally, estimating the carrier frequency offset based on the received signal from the receiver using a Fast Fourier Transform includes: The received signal is demodulated to obtain a demodulated signal, and the demodulated signal is subjected to a fast Fourier transform to obtain the spectrum of the demodulated signal. The frequency corresponding to the amplitude peak in the spectrum is the estimated result of the carrier frequency offset.

[0015] Optionally, the I / Q imbalance estimation of the carrier frequency offset compensated signal based on the adaptive generalized linear filtering method includes: The output signal after I / Q imbalance compensation is expressed as: ; in, This represents the signal after carrier frequency offset compensation. This represents the conjugate of the signal after carrier frequency offset compensation. Indicates the compensation coefficient. This represents the output signal after I / Q imbalance compensation; The compensation coefficients are solved iteratively, and the estimated result of I / Q imbalance is calculated based on the solution of the compensation coefficients; the estimated result of I / Q imbalance includes the estimated value of gain imbalance and the estimated value of phase imbalance.

[0016] Optionally, the iterative solution of the compensation coefficient includes: The compensation coefficients and the output signal after I / Q imbalance compensation are iteratively updated using an adaptive algorithm until the expected value of the output signal after I / Q imbalance compensation is 0, at which point the iteration terminates, and the compensation coefficients are obtained.

[0017] This invention provides a feature-compensated RF fingerprint hiding method. First, at the receiver end, carrier frequency offset estimation, I / Q imbalance estimation, and PA nonlinear distortion estimation are performed sequentially on the received signal. A joint compensation model is obtained based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results. Then, at the transmitter end, this joint compensation model is used to sequentially compensate for carrier frequency offset, I / Q imbalance, and PA nonlinear distortion of the signal to be transmitted, thereby obtaining the transmitted signal with RF fingerprint hidden. This invention achieves a technological leap from "passive compensation at the receiver end" to "active hiding at the transmitter end," not only improving communication performance as traditional methods but, more importantly, effectively preventing unauthorized receivers from extracting RF fingerprints from the received signal. This effectively reduces the classification accuracy of RFFI technology and significantly enhances the transmitter's privacy protection and anti-identification capabilities.

[0018] Furthermore, this invention designs a scientifically sound sequence for parameter estimation and compensation, namely carrier frequency offset, I / Q imbalance, and PA nonlinear distortion. First, carrier frequency offset is compensated. If this offset is not corrected first, subsequent compensations will apply to the frequency-shifted signal, leading to poor compensation results and potentially introducing secondary distortion in the spectrum. Therefore, carrier frequency offset compensation must be prioritized. Next, I / Q imbalance compensation is performed. This is because if PA nonlinear distortion compensation is performed at this point, the original I / Q imbalance will be further amplified or superimposed on the nonlinear effect, resulting in model coupling. This not only complicates the compensation model but also reduces compensation accuracy. Therefore, I / Q imbalance needs to be corrected before PA nonlinear distortion compensation. Finally, PA nonlinear distortion compensation is performed. If it is not compensated in the last step, other distortion factors will remain in the signal, leading to inaccurate characteristic curves fitted by the PA nonlinear distortion compensation model and causing compensation failure. In summary, the joint feature compensation model proposed in this invention organically integrates compensation models for carrier frequency offset, I / Q imbalance, and PA nonlinear distortion, achieving good feature compensation results in a scientifically sound order.

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0020] Figure 1 This is a flowchart of constructing a joint compensation model in a feature-compensated radio frequency fingerprint hiding method provided in an embodiment of the present invention; Figure 2This is a constellation diagram showing the compensation effect of a joint compensation model on the received signal compensation in a feature-compensated radio frequency fingerprint hiding method provided in an embodiment of the present invention. Figure 3 This is a constellation diagram showing the compensation effect of a joint compensation model used in a feature-compensated radio frequency fingerprint hiding method provided in this embodiment of the invention to compensate for the transmitted signal; Figure 4 This is a spectrum diagram of the compensation effect of the carrier frequency offset compensation model in a feature-compensated radio frequency fingerprint hiding method provided in an embodiment of the present invention; Figure 5 This is a constellation diagram showing the compensation effect of the I / Q imbalance compensation model in a feature-compensated radio frequency fingerprint hiding method provided in an embodiment of the present invention. Figure 6 This is a power spectral density diagram of the compensation effect of the PA nonlinear distortion compensation model in a feature-compensated radio frequency fingerprint hiding method provided in an embodiment of the present invention. Figure 7 This is a diagram illustrating the impact of a feature-compensated radio frequency fingerprint hiding method on transmitter classification accuracy, as provided in an embodiment of the present invention. Detailed Implementation

[0021] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0022] In order to prevent unauthorized receivers from extracting radio frequency fingerprints from received signals without affecting the normal transmission performance of the communication system, thereby reducing the classification accuracy of RFFI technology and enhancing the privacy protection and anti-identification capabilities of the transmitter, this invention provides a radio frequency fingerprint hiding method based on feature compensation.

[0023] The following is a detailed description of a feature-compensated radio frequency fingerprint hiding method provided by an embodiment of the present invention. The method includes: At the transmitter, a preset joint compensation model is used to sequentially perform carrier frequency offset compensation, I / Q imbalance compensation, and PA nonlinear distortion compensation on the signal to be transmitted, so as to obtain the transmitted signal after the radio frequency fingerprint is hidden. Among them, see Figure 1 The joint compensation model is constructed in the following way: The received signal of the receiver is sequentially subjected to carrier frequency offset estimation and compensation, I / Q imbalance estimation and compensation, and PA nonlinear distortion estimation. A joint compensation model is then constructed based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results.

[0024] Specifically, in the construction phase of the joint compensation model, targeting the three main features of RF fingerprints—carrier frequency offset, I / Q imbalance, and PA nonlinear distortion—this invention performs parameter estimation and compensation at the receiver end according to a specific order of carrier frequency offset, I / Q imbalance, and PA nonlinear distortion. Based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results, a joint compensation model is constructed. Using this joint compensation model at the transmitter end, carrier frequency offset compensation, I / Q imbalance compensation, and PA nonlinear distortion compensation are performed sequentially on the signal to be transmitted. After these pre-distortion processing steps, the physical layer characteristics of the transmitter during transmission are superimposed, resulting in a final transmitted signal that is the signal with the RF fingerprint hidden, thus achieving the purpose of RF fingerprint hiding.

[0025] In one implementation, such as Figure 1 As shown, carrier frequency offset estimation and compensation, I / Q imbalance estimation and compensation, and PA nonlinear distortion estimation are performed sequentially on the received signal of the receiver. A joint compensation model is then constructed based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results, including: S1. Based on the received signal from the receiver, estimate its carrier frequency offset using Fast Fourier Transform. Construct a carrier frequency offset compensation model based on the carrier frequency offset estimation result, and use the carrier frequency offset compensation model to compensate the received signal for carrier frequency offset, thereby obtaining the carrier frequency offset compensated signal.

[0026] Specifically, carrier frequency offset refers to the possibility that the receiver's local carrier frequency may not be exactly the same as the transmitter's, resulting in an additional frequency shift in the received signal in the frequency domain. Here, carrier frequency offset affects the signal... The impact can be modeled as follows: ; in, The signal representing the effect of carrier frequency offset For the carrier frequency offset to be estimated, Indicates the sampling interval. j Represents the imaginary unit. This represents a discrete-time index.

[0027] Assuming that during the construction phase of the joint compensation model, the transmitter sends a known transmitted signal to the receiver. The receiver received a signal with radio frequency impairment. sampling frequency By performing equal-interval sampling, discrete received signals are obtained. .

[0028] The carrier frequency offset estimation method based on Fast Fourier Transform estimates the carrier frequency offset by analyzing the spectral distribution of the received signal and calculating the offset of the main peak frequency relative to the ideal position.

[0029] Here, based on the received signal from the receiver, the carrier frequency offset is estimated using Fast Fourier Transform (FFT). This includes: demodulating the received signal to obtain a demodulated signal; performing an FFT on the demodulated signal to obtain its spectrum; and identifying the frequency corresponding to the amplitude peak in the spectrum as the estimated carrier frequency offset. The specific steps are as follows: (1) For discrete received signals conduct The signal is processed by exponentiation to remove the phase information of the carrier modulation, resulting in a demodulated signal.

[0030] ; in, Indicates the demodulated signal. This represents a discrete-time index.

[0031] (2) Demodulation signal Performing a Fast Fourier Transform yields its frequency domain representation: ; in, This indicates the number of points in the Fast Fourier Transform. Represents the imaginary unit. The first frequency domain signal represents the second frequency domain signal. One frequency component, .

[0032] (3) Find the frequency corresponding to the peak amplitude in the frequency domain, which is the coarse estimate of the carrier frequency offset: ; in, Indicates the symbol rate, Indicates the sampling frequency. For step (1) , Indicates the first The frequency of each sampling point .

[0033] Therefore, based on the carrier frequency offset estimation results Based on the aforementioned impact of carrier frequency offset on the signal, a carrier frequency offset compensation model can be constructed. For example, this carrier frequency offset compensation model can be expressed as: ; in, Represents a discrete signal to be compensated. This indicates the carrier frequency offset estimation result. Represents a discrete-time index. Indicates the sampling interval. This indicates the signal after carrier frequency offset compensation.

[0034] Therefore, when this carrier frequency offset compensation model is used for discrete received signals When performing carrier frequency offset compensation, the resulting carrier frequency offset compensation signal is: ; in, This indicates the signal after carrier frequency offset compensation.

[0035] S2. Based on the adaptive generalized linear filtering method, I / Q imbalance estimation is performed on the carrier frequency offset compensated signal. Based on the I / Q imbalance estimation results, an I / Q imbalance compensation model is constructed, and the I / Q imbalance compensation model is used to perform I / Q imbalance compensation on the carrier frequency offset compensated signal to obtain the I / Q imbalance compensated signal.

[0036] Specifically, I / Q imbalance refers to the imbalance between the in-phase and out-of-phase components of a signal in a wireless communication system. I Components) and orthogonal components (i.e. Q Inconsistent amplitude gains of the I / Q components, or a deviation from the ideal 90° quadrature between them, cause distortion in the representation of complex signals. Here, I / Q imbalance affects the signal... The impact can be modeled as follows: ; in, This indicates a signal affected by I / Q imbalance. and They represent the signals respectively. I Components and Q Component gain, This indicates an imbalance in the gain to be estimated. In an ideal situation, That is, the gains of the I / Q components are equal. This indicates that the Q component is subject to phase rotation. The impact, The phase imbalance to be estimated.

[0037] The adaptive generalized linear filtering method restores the signal's regularity by suppressing the mirroring effect caused by I / Q imbalance. Specifically, the adaptive generalized linear filtering method performs I / Q imbalance estimation on the carrier frequency offset compensated signal, including: (a) The output signal after I / Q imbalance compensation is expressed as: ; in, This represents the signal after carrier frequency offset compensation. This represents the conjugate of the signal after carrier frequency offset compensation. Indicates the compensation coefficient. This represents the output signal after I / Q imbalance compensation. Here, It can be represented as ,but conjugate It can be represented as ,in Indicates in-phase components, This represents orthogonal components.

[0038] (b) Iteratively solve for the compensation coefficients and calculate the I / Q imbalance estimate based on the solution of the compensation coefficients; the I / Q imbalance estimate includes the gain imbalance estimate and the phase imbalance estimate.

[0039] The iterative solution of the compensation coefficients includes using an adaptive algorithm to iteratively update the compensation coefficients and the output signal after I / Q imbalance compensation, until the expected value of the output signal after I / Q imbalance compensation is 0, at which point the iteration terminates and the compensation coefficients are obtained.

[0040] Specifically, by dynamically adjusting the compensation coefficient It can To obtain a signal that approximates the ideal without distortion. The compensation coefficient is used to best approximate the ideal, distortion-free signal, and the I / Q imbalance is estimated based on this compensation coefficient. The specific calculation steps are as follows: (i) Iteratively update the compensation coefficients according to the following formula. Output signal after I / Q imbalance compensation Until the termination condition is met: This yields the corresponding compensation coefficient; for example, the initial value of the compensation coefficient can be set to 0. ; in, s Indicates the search step size.

[0041] It is understandable that the signal after carrier frequency offset compensation When the sequence length is short, the iteration termination condition may not be reached until the end of the iteration. In this case, The compensation coefficient corresponding to the minimum expected value is taken as the final solution result.

[0042] (ii) Calculate the estimated result of I / Q imbalance, i.e., the estimated value of gain imbalance, based on the solution of the compensation coefficient. and the estimated value of phase imbalance The calculation process is as follows: Based on the results of solving for the compensation coefficients, calculate the temporary variable matrix. : ; in, This represents the real part of the compensation coefficient. This represents the imaginary part of the compensation coefficient.

[0043] Calculate temporary variables and : ; ; in, , , and All are temporary variable matrices The elements in.

[0044] Calculate temporary variables and : ; .

[0045] Calculate the estimate of the I / Q imbalance, i.e., the estimate of the gain imbalance. and the estimated value of phase imbalance : ; ; in, , and All are temporary variables The elements in.

[0046] Therefore, based on the estimation results of I / Q imbalance and the impact of I / Q imbalance on the signal mentioned above, an I / Q imbalance compensation model can be constructed.

[0047] For example, the construction of the I / Q imbalance compensation model can be expressed as: ; in, Represents a discrete signal to be compensated. This represents the signal after I / Q imbalance compensation. This represents the gain imbalance estimate in the I / Q imbalance estimation results. This represents the phase imbalance estimate in the I / Q imbalance estimation results. express The real part, express The imaginary part, j It represents the imaginary unit.

[0048] Therefore, when this I / Q imbalance compensation model is used to compensate the carrier frequency offset signal When performing I / Q imbalance compensation, at this time In fact Therefore, the signal after I / Q imbalance compensation can be expressed as: .

[0049] S3. Based on the signal after I / Q imbalance compensation, the PA nonlinear distortion is estimated using the generalized memory polynomial method, and a PA nonlinear distortion compensation model is constructed based on the PA nonlinear distortion estimation results.

[0050] Specifically, PA nonlinear distortion refers to the distortion of the signal amplitude and phase when the power amplifier operates at high power, as the input-output relationship is no longer linear. Here, PA nonlinear distortion affects the signal... The impact can be modeled as follows: ; in, To calculate the amplitude of the signal, To calculate the phase of the signal, This indicates the signal affected by the nonlinear distortion of the PA. Indicates signal Amplitude Modulation to Amplitude Modulation (AM-AM) characteristics. Indicates signal The amplitude-phase conversion characteristic (AM-PM) can typically be represented by the following two models: (1) Saleh model: ; in, Indicates small-signal gain. This represents the nonlinear compression coefficient.

[0051] ; in, Represents the phase sensitivity coefficient. This represents the phase saturation coefficient.

[0052] (2) Rapp model: ; in, This represents the smoothing factor.

[0053] The method based on Generalized Memory Polynomial (GMP) refers to using a GMP model to inversely model the nonlinear distortion characteristics of the power amplifier (PA), and pre-distorting the signal at the transmitter to maintain linearity as much as possible after passing through the power amplifier, thereby suppressing nonlinear distortion. The specific steps are as follows: (1) Modeling the inverse parameter of PA nonlinear distortion using GMP: ; in, Indicates delay Signal after I / Q imbalance compensation at each sampling point , This represents the signal after PA nonlinear distortion compensation. The memory depth represents the memory polynomial. This represents the nonlinear order of the memory polynomial. Indicates the first The first delay, the first The compensation coefficients corresponding to the first-order nonlinearity.

[0054] For example, the memory depth of a memory polynomial The nonlinear order of the memory polynomial is set to 3. Set it to 5.

[0055] (2) Generate feature vectors: ; (3) The coefficients of GMP are estimated using the recursive least squares (RLS) method. Here, the signal after PA nonlinear distortion compensation is... This can be represented by an eigenvector: .

[0056] The RLS update process is as follows: Initialize the forgetting factor and the error covariance matrix when n=0 and GMP coefficient vector And iteratively update the temporary variables. Error covariance matrix GMP coefficient vector and error The iteration terminates when n=N. The minimum value corresponding to This is the optimal GMP coefficient vector. .

[0057] ; ; ; ; in, express . conjugate.

[0058] (4) Convert the one-dimensional GMP coefficient vector The compensation coefficient matrix is ​​rearranged into a memory polynomial. :

[0059] Therefore, based on the PA nonlinear distortion estimation results and the impact of PA nonlinear distortion on the signal, a PA nonlinear distortion compensation model can be constructed.

[0060] For example, the PA nonlinear distortion compensation model is expressed as follows: ; in, Indicates delay Discrete signal to be compensated at each sampling point , This represents the signal after PA nonlinear distortion compensation. The memory depth represents the memory polynomial. This represents the nonlinear order of the memory polynomial. Indicates the first The first delay, the first The compensation coefficients corresponding to the first-order nonlinearity.

[0061] S4. Construct a joint compensation model based on the carrier frequency offset compensation model, the I / Q imbalance compensation model, and the PA nonlinear distortion compensation model.

[0062] Specifically, based on the carrier frequency offset compensation model obtained in step S1, the I / Q imbalance compensation model obtained in step S2, and the PA nonlinear distortion compensation model obtained in step S3, a joint compensation model is constructed: ; in, Represents a discrete signal to be compensated. This represents the carrier frequency offset compensation model. This represents the I / Q imbalance compensation model. This represents the PA nonlinear distortion compensation model. This indicates the signal after joint compensation.

[0063] Understandably, after the model is built at the receiver, it is sent to the transmitter. At the transmitter, the signal to be transmitted is... Then the signal to be transmitted First, carrier frequency offset compensation is performed using a carrier frequency offset compensation model. The compensated signal is then used as the input to the I / Q imbalance compensation model. I / Q imbalance compensation is performed using an I / Q imbalance compensation model. The compensated signal is then used as the input to a PA nonlinear distortion compensation model, which performs the compensation to obtain the transmitted signal with hidden RF fingerprint. Unlike receiver-side compensation, transmitter-side concealment strategies typically employ pre-distortion methods. This involves performing an inverse transformation of the RF fingerprint introduction process before signal transmission, making the signal as close as possible to the ideal signal after passing through the transmitter, thereby weakening the fingerprint features introduced by transmitter hardware defects.

[0064] This invention provides a feature-compensated RF fingerprint hiding method. First, at the receiver, the received signal is sequentially subjected to carrier frequency offset estimation, I / Q imbalance estimation, and PA nonlinear distortion estimation. A joint compensation model is obtained based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results. Then, at the transmitter, this joint compensation model is used to sequentially compensate for carrier frequency offset, I / Q imbalance, and PA nonlinear distortion of the signal to be transmitted, thereby obtaining the transmitted signal with the RF fingerprint hidden. This invention, by performing joint compensation on the signal to be transmitted, obtains the transmitted signal with the RF fingerprint hidden, enabling the signal received at the receiver to hide the RF fingerprint formed by hardware damage to the transmitter. This invention achieves a technological leap from "passive compensation at the receiver" to "active hiding at the transmitter," not only improving communication performance as traditional methods but, more importantly, effectively preventing unauthorized receivers from extracting the RF fingerprint from the received signal. It can effectively reduce the classification accuracy of RFFI technology and significantly enhance the privacy protection and anti-identification capabilities of the transmitter.

[0065] Furthermore, this invention designs a scientifically sound sequence for parameter estimation and compensation, namely carrier frequency offset, I / Q imbalance, and PA nonlinear distortion. First, carrier frequency offset is compensated. If this offset is not corrected first, subsequent compensations will apply to the frequency-shifted signal, leading to poor compensation results and potentially introducing secondary distortion in the spectrum. Therefore, carrier frequency offset compensation must be prioritized. Next, I / Q imbalance compensation is performed. This is because if PA nonlinear distortion compensation is performed at this point, the original I / Q imbalance will be further amplified or superimposed on the nonlinear effect, resulting in model coupling. This not only complicates the compensation model but also reduces compensation accuracy. Therefore, I / Q imbalance needs to be corrected before PA nonlinear distortion compensation. Finally, PA nonlinear distortion compensation is performed. If it is not compensated in the last step, other distortion factors will remain in the signal, leading to inaccurate characteristic curves fitted by the PA nonlinear distortion compensation model and causing compensation failure. In summary, the joint feature compensation model proposed in this invention organically integrates compensation models for carrier frequency offset, I / Q imbalance, and PA nonlinear distortion, achieving good feature compensation results in a scientifically sound order.

[0066] The following simulation experiment further illustrates the radio frequency fingerprint hiding method based on feature compensation provided by this invention.

[0067] See Figure 2 , Figure 2 This is a constellation diagram illustrating the compensation effect of a joint compensation model on the received signal in a feature-compensated radio frequency fingerprint concealment method provided in this embodiment of the invention. Its parameters are set as carrier frequency offset. Gain imbalance Phase imbalance Small signal gain Nonlinear compression coefficient Phase sensitivity coefficient Phase saturation coefficient . Figure 2 The horizontal axis represents the in-phase component of the signal, and the vertical axis represents the quadrature component. Red dots represent the ideal received signal, green dots represent the received signal with RF fingerprinting impairment (its distribution is scattered and not concentrated near the 16 red constellation points), yellow dots represent the signal after carrier frequency offset compensation, cyan dots represent the signal after I / Q imbalance compensation, and blue dots represent the signal after PA nonlinear distortion compensation. Figure 2 It can be seen that after sequentially performing carrier frequency offset compensation, I / Q imbalance compensation, and PA nonlinear distortion compensation on the received damaged signal, the constellation diagram of the final compensated signal (blue dots) gradually recovers to the position of the ideal received signal (red dots). This shows that the feature-compensated RF fingerprint hiding method provided by the present invention has good effect, and each compensation step has a relatively obvious effect.

[0068] like Figure 3 As shown, Figure 3 This is a constellation diagram illustrating the compensation effect of a joint compensation model used in a feature-compensated RF fingerprint concealment method according to an embodiment of the present invention to compensate the transmitted signal. Its parameter settings are consistent with... Figure 2 They are exactly the same. Figure 3 The horizontal axis represents the in-phase component of the signal, and the vertical axis represents the quadrature component. Red dots represent the ideal received signal, green dots represent the received signal with RF fingerprinting, and blue dots represent the received signal after pre-distortion processing of the transmitted signal at the transmitter using a joint compensation model. Figure 3 As can be seen, after pre-distortion processing, the received signal of the receiver (blue dot) is distributed near the ideal received signal (red dot), indicating that the RF fingerprint hiding method based on feature compensation provided by the present invention has a good effect. By using a preset joint compensation model to pre-distort the signal before signal transmission, the signal is made as close as possible to the ideal signal after passing through the transmitter, thereby reducing the RF fingerprint features introduced by the transmitter hardware damage and achieving the purpose of RF fingerprint hiding.

[0069] See Figure 4 , Figure 4 This is a spectrum of the compensation effect of the carrier frequency offset compensation model in a feature-compensated radio frequency fingerprint hiding method provided in an embodiment of the present invention, with its parameters set as carrier frequency offset. . Figure 4 The horizontal axis represents frequency in kHz, and the vertical axis represents amplitude (dB). Here, due to the carrier frequency offset, the received signal experiences an overall shift in the frequency domain, leading to constellation diagram rotation and inter-symbol interference. Figure 4 The subplot above shows the spectrum of the received signal with carrier frequency offset. Figure 4 The sub-figure below shows the spectrum of the received signal after carrier frequency offset compensation using the carrier frequency offset compensation model. Figure 4 As can be seen, the radio frequency fingerprint hiding method based on feature compensation provided by the present invention can compensate for carrier frequency offset, so that the spectrum of the compensated received signal is restored to the ideal position and exhibits characteristics of symmetry about the center frequency.

[0070] See Figure 5 , Figure 5 This is a constellation diagram showing the compensation effect of the I / Q imbalance compensation model in a feature-compensated RF fingerprint hiding method provided in this embodiment of the invention, with its parameters set to gain imbalance. Phase imbalance In this diagram, red dots represent the ideal received signal, green dots represent the received signal with I / Q imbalance, and blue dots represent the received signal after I / Q imbalance compensation using an I / Q imbalance compensation model. Due to the influence of I / Q imbalance, the received signal experiences amplitude and phase mismatch, causing constellation point shifts, affecting signal demodulation performance, and reducing signal transmission quality. Figure 5 As can be seen, the radio frequency fingerprint hiding method based on feature compensation provided by the present invention can compensate for I / Q imbalance, so that the constellation points of the compensated received signal return to the vicinity of the ideal position, and the constellation diagram presents a regular and symmetrical constellation structure again.

[0071] See Figure 6 , Figure 6 This is a power spectral density map of the PA nonlinear distortion compensation model in a feature-compensated RF fingerprint hiding method provided in this embodiment of the invention, with its parameters set as small-signal gain. Nonlinear compression coefficient Phase sensitivity coefficient Phase saturation coefficient . Figure 6 In the diagram, blue represents the ideal received signal, red represents the received signal with PA nonlinear distortion, and yellow represents the received signal after PA nonlinear distortion compensation using a PA nonlinear distortion compensation model. Due to the influence of PA nonlinear distortion, the received signal experiences amplitude and phase distortion, especially at high power output. This nonlinear distortion causes out-of-band radiation and constellation diagram distortion, severely affecting the system's spectral efficiency and leading to an increased bit error rate. Figure 6 As can be seen, the radio frequency fingerprint hiding method based on feature compensation provided by the present invention can compensate for PA nonlinear distortion and improve the linearity of the signal.

[0072] See Figure 7 , Figure 7 This diagram illustrates the impact of a feature-compensated RF fingerprint hiding method on transmitter classification accuracy, as provided in this embodiment of the invention. Six different transmitters were simulated using MATLAB, with parameters shown in Table 1. The signal-to-noise ratio (SNR) was... In this case, the Higher-Order Cumulants (HOC) feature extraction method and Support Vector Machine (SVM) classification model are used to identify the identities of these six transmitters. It can be seen that without feature compensation (i.e., ... Figure 7The classification accuracy (shown by the blue broken line) reached over 90% at all SNR levels, and the accuracy increased with increasing SNR. This indicates that without feature compensation, the receiver can accurately identify the transmitter from the received signal containing the RF fingerprint. However, when the RF fingerprint hiding method based on feature compensation provided by this invention is used to pre-distort the transmitted signal (i.e.,...), the accuracy is significantly improved. Figure 6 The orange broken line shows that at all SNR levels, the classification accuracy is lower than that without feature compensation. As SNR increases, classification accuracy initially increases slightly, then drops sharply after SNR=10. Finally, at SNR=30, the classification accuracy is only 30%. This indicates that using a joint compensation model to compensate for the three features of the transmitted signal can effectively hide the signal's RF fingerprint, making it difficult for unauthorized receivers to accurately identify the transmitter, thereby improving the concealment and security of communication.

[0073] Parameters of 16 transmitters

[0074] It should be noted that the terms "first," "second," etc., are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention.

[0075] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

[0076] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings and the disclosure in carrying out the claimed invention. In the description of the invention, the word "comprising" does not exclude other components or steps, "a" or "an" does not exclude a plurality, and "a plurality" means two or more, unless otherwise explicitly specified. Furthermore, while different embodiments may describe certain measures, this does not mean that these measures cannot be combined to produce good results.

[0077] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A radio frequency fingerprint hiding method based on feature compensation, characterized in that, include: At the transmitter end, a preset joint compensation model is used to sequentially perform carrier frequency offset compensation, I / Q imbalance compensation and PA nonlinear distortion compensation on the signal to be transmitted to obtain the transmitted signal after the radio frequency fingerprint is hidden. The joint compensation model is constructed in the following manner: The received signal of the receiver is sequentially subjected to carrier frequency offset estimation and compensation, I / Q imbalance estimation and compensation, and PA nonlinear distortion estimation. A joint compensation model is then constructed based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results.

2. The radio frequency fingerprint hiding method based on feature compensation according to claim 1, characterized in that, The process involves sequentially performing carrier frequency offset estimation and compensation, I / Q imbalance estimation and compensation, and PA nonlinear distortion estimation on the received signal of the receiver. A joint compensation model is then constructed based on the carrier frequency offset estimation results, I / Q imbalance estimation results, and PA nonlinear distortion estimation results, including: Based on the received signal from the receiver, the carrier frequency offset is estimated using Fast Fourier Transform. A carrier frequency offset compensation model is constructed based on the carrier frequency offset estimation result, and the carrier frequency offset compensation model is used to compensate the received signal for the carrier frequency offset to obtain the carrier frequency offset compensated signal. The I / Q imbalance is estimated for the carrier frequency offset compensated signal based on the adaptive generalized linear filtering method. An I / Q imbalance compensation model is constructed based on the I / Q imbalance estimation results. The I / Q imbalance compensation model is then used to compensate the carrier frequency offset compensated signal for I / Q imbalance, thus obtaining the I / Q imbalance compensated signal. Based on the signal after I / Q imbalance compensation, the generalized memory polynomial method is used to estimate the PA nonlinear distortion, and a PA nonlinear distortion compensation model is constructed based on the PA nonlinear distortion estimation results. The joint compensation model is constructed based on the carrier frequency offset compensation model, the I / Q imbalance compensation model, and the PA nonlinear distortion compensation model.

3. The radio frequency fingerprint hiding method based on feature compensation according to claim 2, characterized in that, The carrier frequency offset compensation model is expressed as follows: ; in, Represents a discrete signal to be compensated. This indicates the carrier frequency offset estimation result. Represents a discrete-time index. Indicates the sampling interval. j Represents the imaginary unit. This indicates the signal after carrier frequency offset compensation.

4. The radio frequency fingerprint hiding method based on feature compensation according to claim 2, characterized in that, The I / Q imbalance compensation model is expressed as follows: ; in, Represents a discrete signal to be compensated. This represents the signal after I / Q imbalance compensation. This represents the gain imbalance estimate in the I / Q imbalance estimation results. This represents the phase imbalance estimate in the I / Q imbalance estimation results. express The real part, express The imaginary part, j It represents the imaginary unit.

5. The radio frequency fingerprint hiding method based on feature compensation according to claim 2, characterized in that, The PA nonlinear distortion compensation model is expressed as follows: ; in, Indicates delay Discrete signal to be compensated at each sampling point , This represents the signal after PA nonlinear distortion compensation. The memory depth represents the memory polynomial. This represents the nonlinear order of the memory polynomial. Indicates the first The first delay, the first The compensation coefficients corresponding to the first-order nonlinearity.

6. The radio frequency fingerprint hiding method based on feature compensation according to claim 5, characterized in that, The memory depth of the memory polynomial The nonlinear order of the memory polynomial is set to 3. Set it to 5.

7. The radio frequency fingerprint hiding method based on feature compensation according to claim 3, 4 or 5, characterized in that, The joint compensation model is expressed as: ; in, Represents a discrete signal to be compensated. This represents the carrier frequency offset compensation model. This represents the I / Q imbalance compensation model. This represents the PA nonlinear distortion compensation model. This indicates the signal after joint compensation.

8. The radio frequency fingerprint hiding method based on feature compensation according to claim 2, characterized in that, The step of estimating the carrier frequency offset based on the received signal from the receiver using Fast Fourier Transform includes: The received signal is demodulated to obtain a demodulated signal, and the demodulated signal is subjected to a fast Fourier transform to obtain the spectrum of the demodulated signal. The frequency corresponding to the amplitude peak in the spectrum is the estimated result of the carrier frequency offset.

9. The radio frequency fingerprint hiding method based on feature compensation according to claim 2, characterized in that, The method for estimating I / Q imbalance in the carrier frequency offset compensated signal based on adaptive generalized linear filtering includes: The output signal after I / Q imbalance compensation is expressed as: ; in, This represents the signal after carrier frequency offset compensation. This represents the conjugate of the signal after carrier frequency offset compensation. Indicates the compensation coefficient. This represents the output signal after I / Q imbalance compensation; The compensation coefficients are solved iteratively, and the estimated result of I / Q imbalance is calculated based on the solution of the compensation coefficients; the estimated result of I / Q imbalance includes the estimated value of gain imbalance and the estimated value of phase imbalance.

10. The radio frequency fingerprint hiding method based on feature compensation according to claim 9, characterized in that, The iterative solution of the compensation coefficient includes: The compensation coefficients and the output signal after I / Q imbalance compensation are iteratively updated using an adaptive algorithm until the expected value of the output signal after I / Q imbalance compensation is 0, at which point the iteration terminates, and the compensation coefficients are obtained.

Citation Information

Patent Citations

  • Digital predistortion method of jointly compensating for IQ imbalance and PA non-linearity

    CN105471784A

  • Wireless device radio frequency fingerprint extraction method based on multipath fading compensation

    CN116599810A

  • Equipment identity recognition method and device, electronic equipment and storage medium

    CN119676715A

  • Nonlinear distortion compensating method foe transmission part of radio communication equipment and the equipment

    JP2000244370A

  • United estimation pre-compensation method for solving imperfection in downstream transmission system capable of transmitting a more perfect signal by compensating for a receiving signal in advance on a transmitting end

    TW201724818A

Cited By

  • Unmanned aerial vehicle active countering method based on radio frequency fingerprint identification and protocol cracking

    CN122339622A