A physical layer device identity authentication method and system based on channel prediction

Abstract

A physical layer device authentication method and system based on channel prediction is disclosed. The method includes the following steps: predicting the channel state information of a legitimate device at the authentication time using a periodically updated Long Short-Term Memory (LSTM) network; constructing a receiving domain around the channel state information of the legitimate device at the authentication time using Bayesian parameter estimation; estimating the channel state information of the received signal; if the channel state information of the received signal falls within the receiving domain, the received signal is determined to originate from a legitimate device; otherwise, the received signal is determined to originate from an illegitimate device. The periodically updated LSTM network can adaptively learn from changes in the learning environment. This invention can adapt well to dynamic scenarios and does not require obtaining the attacker's channel information during implementation. When using multiple subcarriers for joint authentication, the method of this invention employs a subcarrier spacing selection strategy to further improve the overall authentication performance.

Description

Technical Field

[0001] This invention belongs to the field of communication technology and relates to a physical layer device authentication method and system based on channel prediction. Background Technology

[0002] Communication technology is rapidly evolving towards more connected devices, massive connections, and wider application scenarios. However, the openness of wireless channels makes communication devices vulnerable to various attacks, such as eavesdropping, interference, and spoofing attacks. The transmission of large amounts of private and confidential information in wireless networks places higher demands on their security. Device authentication, as the first step in ensuring communication security, is therefore extremely important.

[0003] Traditional device authentication typically utilizes upper-layer key-based security mechanisms, such as challenge-response mechanisms. This mechanism assumes limited computing power for the attacker and relies on encryption techniques based on mathematical calculations. By distributing encryption keys, it prevents the attacker from decrypting within a reasonable timeframe, thus ensuring security. However, with the development of technologies like quantum computing, computing power has made breakthroughs, and abundant computing resources have made keys easily vulnerable to attack and theft. Furthermore, key-based upper-layer security mechanisms not only require complex encoding and decoding operations at the upper layer but also necessitate key sharing and management. These complex upper-layer operations lead to high complexity, while key sharing and management result in higher costs. In networks such as the Internet of Things (IoT), connected vehicles, and drones, terminal devices are typically small and low-power. This makes traditional key-based security mechanisms difficult to implement in such large-scale, heterogeneous, and distributed wireless networks.

[0004] Physical layer authentication (PLE) is widely studied as an effective supplement to upper-layer authentication schemes due to its low complexity and high accuracy. PLE primarily utilizes the unique physical layer characteristics between wireless communication devices as their natural fingerprints for authentication. These physical layer features, due to their inherent nature, do not require special generation, and their uniqueness and difficulty in imitation allow for high recognition accuracy while maintaining low complexity. Currently, some research has utilized machine learning and deep learning algorithms for PLE. For example, there are schemes that extract two-dimensional channel features based on channel state information and use extreme learning machines for authentication; and schemes that use adaptive matrices and convolutional neural networks to automatically extract features for device authentication. However, deep learning-based schemes require channel information from unauthorized attackers as training samples and cannot adapt to dynamic scenarios. Gaussian process regression-based authentication schemes can adapt well to dynamic scenarios and do not require channel information from unauthorized attackers for training; however, the single kernel function form of Gaussian process regression limits its performance. Summary of the Invention

[0005] The purpose of this invention is to address the problems in the prior art by providing a physical layer device authentication method and system based on channel prediction, which can achieve authentication effects that do not require channel information from unauthorized attackers and are adaptable to dynamic environments.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A physical layer device authentication method based on channel prediction includes the following steps:

[0008] The channel state information of the legitimate device at the authentication time is predicted by a regularly updated long short-term memory network;

[0009] Using Bayesian parameter estimation, a receiver domain is constructed around the channel state information of the legitimate device at the authentication time;

[0010] The channel state information of the received signal is estimated. If the channel state information of the received signal falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

[0011] As a preferred embodiment, the periodically updated Long Short-Term Memory network uses the channel state information from d consecutive time steps to predict the channel state information at the next time step;

[0012] The input x(n) of a Long Short-Term Memory (LSTM) network is represented as:

[0013]

[0014] In the formula, This represents the actual channel state information on the m-th subcarrier at time n;

[0015] The labels for the input x(n) are:

[0016]

[0017] The label y(n) represents the channel state information on the m-th subcarrier at the next time step;

[0018] The entire training set is:

[0019] X = (x(1), x(2), ..., x(N))

[0020] Where N represents the size of the training set, the corresponding label set is Y = (y(1), y(2), ..., y(N)), and the network output value is denoted as... Minimize the loss function using gradient descent. Update network parameters to complete the training process;

[0021] After training the network, predict the channel state information of legitimate devices at detection time t+1:

[0022] Will

[0023]

[0024] The input is given to the trained network, and the network output is the predicted value of the legitimate channel at time t+1.

[0025] As a preferred embodiment, the periodically updated Long Short-Term Memory network uses the optimal weights with the smallest training loss value from the previous training process as the initial weights for each retraining, randomly initializes the biases, and performs training. This allows the network to periodically update and track changes in the environment, with the retraining interval determined based on changes in the wireless channel.

[0026] As a preferred embodiment, the Bayesian parameter estimation estimates the prediction error range of the legitimate channel state information at the detection time, thereby constructing a legitimate receiving region; the step of constructing a receiving domain around the channel state information of the legitimate device at the authentication time includes:

[0027] The prediction error is:

[0028]

[0029] in, Let e ​​represent the predicted and actual channel state information on the m-th subcarrier at time t+1, respectively; the prediction error e (m) (t+1) is modeled as having a mean of μ (m) The variance is φ (m) Gaussian distribution, Bayesian estimation obtains the prediction error estimate μ from the training set during the training phase. (m) and φ (m) ;

[0030] μ (m) and φ (m) The Bayesian estimates are as follows:

[0031]

[0032]

[0033] in:

[0034]

[0035]

[0036]

[0037] In the formula, Let N be an initialization constant, and N be the size of the training set.

[0038] The receiving domain is constructed as follows:

[0039]

[0040] Here, q is a parameter that determines the size of the receiving domain, and the value of q is selected according to the false alarm rate requirement; if the channel state information estimated from the received signal falls within the receiving domain, the received signal is considered to come from a legitimate device, otherwise it is considered to come from an attacker.

[0041] As a preferred embodiment, the physical layer device authentication method based on channel prediction of the present invention utilizes channel state information on multiple subcarriers for joint authentication, and the authentication process is as follows:

[0042] The channel state information of the legitimate device at the authentication time is predicted on multiple subcarriers through a periodically updated long short-term memory network; and a receiving domain is constructed around the channel state information of the legitimate device at the authentication time using Bayesian parameter estimation.

[0043] The channel state information of the received signals on multiple subcarriers is estimated. If the channel state information of the received signals on all subcarriers falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

[0044] As a preferred embodiment, the multiple subcarriers are selected for authentication at a fixed interval ΔD, where the interval ΔD is determined based on the correlation between the subcarriers.

[0045] A physical layer device authentication system based on channel prediction, comprising:

[0046] The legitimate device channel state information prediction module is used to predict the channel state information of the legitimate device at the time of authentication through a periodically updated long short-term memory network.

[0047] The legitimate reception domain construction module is used to construct the reception domain around the channel state information of the legitimate device at the time of authentication by utilizing Bayesian parameter estimation.

[0048] The determination module is used to estimate the channel state information of the received signal. If the channel state information of the received signal falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

[0049] As a preferred embodiment, the physical layer device authentication system based on channel prediction of the present invention utilizes channel state information on multiple subcarriers for joint authentication, and the authentication process is as follows:

[0050] The channel state information of the legitimate device at the authentication time is predicted on multiple subcarriers through a periodically updated long short-term memory network; and a receiving domain is constructed around the channel state information of the legitimate device at the authentication time using Bayesian parameter estimation.

[0051] The channel state information of the received signals on multiple subcarriers is estimated. If the channel state information of the received signals on all subcarriers falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

[0052] An electronic device, comprising:

[0053] A memory for storing at least one instruction; and a processor for executing the instructions stored in the memory to implement the channel prediction-based physical layer device authentication method.

[0054] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the channel prediction-based physical layer device authentication method.

[0055] Compared with the prior art, the present invention has at least the following beneficial effects:

[0056] By utilizing a periodically updated Long Short-Term Memory (LSTM) network, the channel state information of the legitimate device at the authentication time is predicted. The periodically updated LSTM network can adaptively learn from changes in the learning environment. By using Bayesian parameter estimation, a receiving domain is constructed around the channel state information of the legitimate device at the authentication time. By estimating the channel state information of the received signal, if the channel state information of the received signal falls within the receiving domain, the received signal is determined to come from the legitimate device; otherwise, the received signal is determined to come from the illegitimate device. This achieves good authentication performance in dynamic environments without requiring channel information training from an attacker.

[0057] Furthermore, this invention utilizes channel state information on multiple subcarriers for joint authentication. When using multiple subcarriers for authentication, the subcarrier spacing selection strategy can further improve the authentication accuracy of the system. Attached Figure Description

[0058] Figure 1 The architecture diagram of the physical layer device authentication method based on channel prediction applied to a wireless network according to an embodiment of the present invention;

[0059] Figure 2The channel prediction performance diagram based on a periodically updated long short-term memory network in this embodiment of the invention;

[0060] Figure 3 A comparison chart of the authentication performance of the embodiments of the present invention and the Gaussian process regression-based scheme;

[0061] Figure 4 The effect diagram of the subcarrier spacing selection strategy in an embodiment of the present invention. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, those skilled in the art can obtain other embodiments without creative effort.

[0063] See Figure 1 This invention proposes a physical layer device authentication method based on channel prediction. When applied to a wireless network architecture, a legitimate transmitting device Alice and a legitimate receiving device Bob attempt to establish a legitimate communication connection. However, signals transmitted in the wireless channel can also be captured by Eve. Eve can extract information such as Alice's MAC address from these signals and impersonate Alice to attack Bob. Bob extracts physical layer information from the received signals to determine the source of the signals, classifying them as either legitimate or illegitimate.

[0064] Specifically, the physical layer device authentication method based on channel prediction of the present invention includes the following steps:

[0065] 1) Predict the channel state information of legitimate devices at the detection time using a periodically updated Long Short-Term Memory network:

[0066] A periodically updated Long Short-Term Memory (LSTM) network uses channel state information from d consecutive time steps to predict the channel state information at the next time step, where d represents the number of time steps of the channel state information used. The network input x(n) can be expressed as:

[0067]

[0068] in, This represents the channel state information on the m-th subcarrier at time n. The label corresponding to x(n) is:

[0069]

[0070] The label y(n) represents the channel state information on the m-th subcarrier at the next moment.

[0071] Therefore, the entire training set can be denoted as X = (x(1), x(2), ..., x(N)), where N represents the size of the training set.

[0072] The corresponding label set is Y = (y(1), y(2), ..., y(N)). The network output value is denoted as... Minimize the loss function using gradient descent. Update network parameters to complete the training process.

[0073] After training the network, the channel state information of the legitimate devices at detection time t+1 needs to be predicted, that is:

[0074]

[0075] The input to the trained network yields the network output value, which is the predicted value of the legitimate channel at time t+1.

[0076] Due to factors such as device movement, wireless channels change over time. Therefore, the aforementioned network requires periodic training to prevent network parameters from becoming outdated and failing to keep up with changes in the wireless channel. To reduce training time, this embodiment of the invention employs an incremental update approach during network retraining. Each time the network is retrained, the optimal weights with the lowest training loss value from the previous training process are used as the initial weights for this retraining. Bias is randomly initialized during training, allowing the network to periodically update and track environmental changes. The retraining interval can be determined based on the changes in the wireless channel.

[0077] 2) Authentication is performed by constructing a receiver domain around the channel state information of legitimate devices predicted at the detection time using Bayesian estimation:

[0078] When constructing the receiving domain, Bayesian estimation is used to estimate the prediction error range of legitimate channel state information at the detection time, thereby constructing the legitimate receiving area. The prediction error is denoted as:

[0079]

[0080] in, Let represent the predicted and actual channel state information on the m-th subcarrier at time t+1, respectively. Prediction error e (m) (t+1) is modeled as having a mean of μ (m) The variance is φ (m) Gaussian distribution, Bayesian estimation estimates the prediction error μ based on the training set obtained during the training phase. (m) and φ (m) μ (m) and φ (m)The Bayesian estimate is:

[0081]

[0082]

[0083] in:

[0084]

[0085]

[0086]

[0087] In the formula, Let N be an initialization constant, and N be the size of the training set.

[0088] Therefore, the receiving domain can be constructed as follows:

[0089]

[0090] Here, q is a parameter that determines the size of the receiving domain, and an appropriate q value can be selected according to the false alarm rate requirements. If the channel state information estimated from the received signal falls within the receiving domain, the signal is considered to originate from a legitimate device; otherwise, it is considered to originate from an attacker.

[0091] In one possible implementation, the physical layer device authentication method based on channel prediction of the present invention utilizes channel state information on multiple subcarriers for joint authentication, and the authentication process is as follows:

[0092] 3.1) Repeat steps 1) and 2) on multiple subcarriers respectively to perform channel prediction and construct the receiver domain.

[0093] 3.2) The authentication results on multiple subcarriers are combined for judgment. If the channel state information estimated from the received signal on all subcarriers falls within the receiving domain, it is considered valid; otherwise, it is considered invalid.

[0094] Selecting multiple subcarriers for authentication can further improve the system's authentication accuracy by employing a subcarrier spacing selection strategy, which involves selecting more independent subcarriers for authentication at a fixed interval ΔD.

[0095] The interval ΔD can be determined based on the correlation of the subcarriers used. If the subcarrier correlation is high, the interval ΔD should be a larger value; if the subcarrier correlation is low, the interval ΔD can be a smaller value.

[0096] Simulation results show that the method of the present invention can achieve good device authentication results in dynamic scenarios. Furthermore, the authentication accuracy of the system can be further improved by employing a subcarrier spacing selection strategy.

[0097] To verify the performance of the physical layer device authentication method based on channel prediction of the present invention, the following simulation was performed:

[0098] Simulation conditions

[0099] This simulation experiment uses Quasi Deterministic Radio Channel Generator (QuaDRiGa) software to generate a dynamic channel that more closely resembles the real-world scenario. The center frequency is set to 2.53 GHz, the baseband bandwidth to 20 MHz, and the number of subcarriers to 128. The scenario used is BERLIN_UMa_LOS, with 4 samples per half wavelength. The legitimate transmitting device, Alice, starts 10 m away from Bob and moves at a constant speed eastward. The attacker, Eve, moves at a constant speed in a circle around Bob with a radius of 10 m. The Long Short-Term Memory (LSTM) network uses a two-layer structure with 80 and 100 neurons respectively. The Adam optimizer is used, with an initial learning rate of 1e-3. The network training stops after 50 epochs, with a batch size of 32.

[0100] Simulation content and result analysis

[0101] Depend on Figure 2 It can be seen that using a periodically updated Long Short-Term Memory (LSTM) network for channel prediction results in a very close match between the normalized Channel State Information (CSI) true value and the predicted value, enabling good prediction of channels for legitimate devices. A comparison of the authentication performance of the method in this invention with that based on a Gaussian process regression scheme is provided. Figure 3 As shown. Authentication performance is reflected by the false alarm rate / false negative rate curve; the lower the curve, the better the authentication performance. From... Figure 3 It can be seen that the authentication performance of the method of this invention is superior to that of the authentication scheme based on Gaussian process regression in dynamic scenarios. This is because the authentication scheme based on Gaussian process regression uses a fixed kernel function for prediction in the channel prediction stage, which may not accurately represent the current environment. The method of this invention uses a periodically updated Long Short-Term Memory network for channel prediction, which can adaptively learn the current environment, achieving better channel prediction performance and thus obtaining better authentication results. Figure 4 This demonstrates the effectiveness of a subcarrier spacing selection strategy used for joint authentication leveraging channel state information across multiple subcarriers. Figure 4The paper compares the authentication performance with and without a subcarrier spacing selection strategy when using two subcarriers (M=2) and three subcarriers (M=3). It shows that, regardless of whether two or three subcarriers are used for authentication, the subcarrier spacing selection strategy is superior to not using it. This is because the subcarrier spacing selection strategy allows for the selection of more independent subcarriers for authentication, resulting in better authentication performance.

[0102] Therefore, as summarized above, the physical layer device authentication method based on channel prediction proposed in this invention can adapt well to dynamic scenarios and does not require obtaining the attacker's channel information during implementation. When using multiple subcarriers for joint authentication, the subcarrier spacing selection strategy employed in this invention can further improve the overall authentication performance.

[0103] Another embodiment of the present invention also proposes a physical layer device authentication system based on channel prediction, comprising:

[0104] The legitimate device channel state information prediction module is used to predict the channel state information of the legitimate device at the time of authentication through a periodically updated long short-term memory network.

[0105] The legitimate reception domain construction module is used to construct the reception domain around the channel state information of the legitimate device at the time of authentication by utilizing Bayesian parameter estimation.

[0106] The determination module is used to estimate the channel state information of the received signal. If the channel state information of the received signal falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

[0107] In one possible implementation, the channel prediction-based physical layer device authentication system uses channel state information on multiple subcarriers for joint authentication, and the authentication process is as follows:

[0108] The channel state information of the legitimate device at the authentication time is predicted on multiple subcarriers through a periodically updated long short-term memory network; and a receiving domain is constructed around the channel state information of the legitimate device at the authentication time using Bayesian parameter estimation.

[0109] The channel state information of the received signals on multiple subcarriers is estimated. If the channel state information of the received signals on all subcarriers falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

[0110] This invention also proposes an electronic device, comprising:

[0111] A memory for storing at least one instruction; and a processor for executing the instructions stored in the memory to implement the channel prediction-based physical layer device authentication method.

[0112] This invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the channel prediction-based physical layer device authentication method.

[0113] For example, the instructions stored in the memory can be divided into one or more modules / units. These modules / units are stored in a computer-readable storage medium and executed by the processor to complete the physical layer device authentication method based on channel prediction according to the present invention. The one or more modules / units can be a series of computer-readable instruction segments capable of performing specific functions, which describe the execution process of the computer program in the server.

[0114] The electronic device may be a smartphone, laptop, PDA, or cloud server, among other computing devices. It may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the electronic device may also include more or fewer components, or combinations of certain components, or different components; for example, it may also include input / output devices, network access devices, buses, etc.

[0115] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0116] The memory can be an internal storage unit of the server, such as a hard drive or RAM. Alternatively, it can be an external storage device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory can include both internal and external storage units. The memory is used to store computer-readable instructions and other programs and data required by the server. It can also be used to temporarily store data that has been output or will be output.

[0117] It should be noted that the information interaction and execution process between the above-mentioned module units are based on the same concept as the method embodiment. For details on their specific functions and technical effects, please refer to the method embodiment section. They will not be repeated here.

[0118] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0119] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0120] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0121] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A physical layer device authentication method based on channel prediction, characterized in that, Includes the following steps: The channel state information of the legitimate device at the authentication time is predicted by a regularly updated long short-term memory network; Using Bayesian parameter estimation, a receiver domain is constructed around the channel state information of the legitimate device at the authentication time; The channel state information of the received signal is estimated. If the channel state information of the received signal falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device. The periodically updated Long Short-Term Memory network utilizes d Channel state information at consecutive time steps can be used to predict channel state information at the next time step. Input to Long Short-Term Memory Network Represented as: In the formula, express n Time of the first m Real channel state information on each subcarrier; Corresponding input The tags are: Label For the next moment m Channel state information on each subcarrier; The entire training set is: in, N This represents the size of the training set, and the corresponding label set is... Y The network output value is denoted as Minimize the loss function using gradient descent. Update network parameters to complete the training process; After training the network, the detection timing for legitimate devices will be... t Prediction is performed using channel state information of +1: Will The input to the trained network is the network output value. t Predicted value of the legitimate channel at time +1 ; The Bayesian parameter estimation estimates the prediction error range of the legitimate channel state information at the detection time, thereby constructing a legitimate receiving region; the step of constructing a receiving domain around the channel state information of the legitimate device at the authentication time includes: The prediction error is: in, , They represent t The predicted and actual channel state information on the m-th subcarrier at time +1; prediction error. Modeled as mean variance is Gaussian distribution, Bayesian estimation obtains the prediction error estimate from the training set during the training phase. and ; and The Bayesian estimates are as follows: in: In the formula, , , , Let N be an initialization constant, and N be the size of the training set. , , ; The receiving domain is constructed as follows: Here, q is a parameter that determines the size of the receiving domain, and the value of q is selected according to the false alarm rate requirement; if the channel state information estimated from the received signal falls within the receiving domain, the received signal is considered to come from a legitimate device, otherwise it is considered to come from an attacker.

2. The physical layer device authentication method based on channel prediction according to claim 1, characterized in that, The periodically updated Long Short-Term Memory network uses the optimal weights with the smallest training loss value from the previous training process as the initial weights for each retraining, and randomly initializes the biases for training. This allows the network to periodically update and track changes in the environment. The retraining interval is determined based on changes in the wireless channel.

3. The physical layer device authentication method based on channel prediction according to claim 1, characterized in that, Joint authentication is performed using channel state information from multiple subcarriers. The authentication process is as follows: The channel state information of the legitimate device at the authentication time is predicted on multiple subcarriers through a periodically updated long short-term memory network; and a receiving domain is constructed around the channel state information of the legitimate device at the authentication time using Bayesian parameter estimation. The channel state information of the received signals on multiple subcarriers is estimated. If the channel state information of the received signals on all subcarriers falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

4. The physical layer device authentication method based on channel prediction according to claim 3, characterized in that, The multiple subcarriers are spaced at fixed intervals. D Subcarriers are selected for authentication, with intervals between them. D It is determined based on the magnitude of subcarrier correlation.

5. A physical layer device authentication system based on channel prediction, characterized in that, include: The legitimate device channel state information prediction module is used to predict the channel state information of the legitimate device at the time of authentication through a periodically updated long short-term memory network. The legitimate reception domain construction module is used to construct the reception domain around the channel state information of the legitimate device at the time of authentication by utilizing Bayesian parameter estimation. The determination module is used to estimate the channel state information of the received signal. If the channel state information of the received signal falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device. The periodically updated Long Short-Term Memory network utilizes d Channel state information at consecutive time steps can be used to predict channel state information at the next time step. Input to Long Short-Term Memory Network Represented as: In the formula, express n Time of the first m Real channel state information on each subcarrier; Corresponding input The tags are: Label For the next moment m Channel state information on each subcarrier; The entire training set is: in, N This represents the size of the training set, and the corresponding label set is... Y The network output value is denoted as Minimize the loss function using gradient descent. Update network parameters to complete the training process; After training the network, the detection timing for legitimate devices will be... t Prediction is performed using channel state information of +1: Will The input to the trained network is the network output value. t Predicted value of the legitimate channel at time +1 ; The Bayesian parameter estimation estimates the prediction error range of the legitimate channel state information at the detection time, thereby constructing a legitimate receiving region; the step of constructing a receiving domain around the channel state information of the legitimate device at the authentication time includes: The prediction error is: in, , They represent t The predicted and actual channel state information on the m-th subcarrier at time +1; prediction error. Modeled as mean variance is Gaussian distribution, Bayesian estimation obtains the prediction error estimate from the training set during the training phase. and ; and The Bayesian estimates are as follows: in: In the formula, , , , Let N be an initialization constant, and N be the size of the training set. , , ; The receiving domain is constructed as follows: Here, q is a parameter that determines the size of the receiving domain, and the value of q is selected according to the false alarm rate requirement; if the channel state information estimated from the received signal falls within the receiving domain, the received signal is considered to come from a legitimate device, otherwise it is considered to come from an attacker.

6. The physical layer device authentication system based on channel prediction according to claim 5, characterized in that, Joint authentication is performed using channel state information from multiple subcarriers. The authentication process is as follows: The channel state information of the legitimate device at the authentication time is predicted on multiple subcarriers through a periodically updated long short-term memory network; and a receiving domain is constructed around the channel state information of the legitimate device at the authentication time using Bayesian parameter estimation. The channel state information of the received signals on multiple subcarriers is estimated. If the channel state information of the received signals on all subcarriers falls within the receiving domain, the received signal is determined to come from a legitimate device; otherwise, the received signal is determined to come from an illegitimate device.

7. An electronic device, characterized in that, include: Memory, storing at least one instruction; and The processor executes instructions stored in the memory to implement the physical layer device authentication method based on channel prediction as described in any one of claims 1 to 4.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the physical layer device authentication method based on channel prediction as described in any one of claims 1 to 4.

Images