MIMO device radio frequency fingerprinting method and apparatus in multi-user wireless communication scenarios

Abstract

The application provides a MIMO device radio frequency fingerprint identification method in a multi-user wireless communication scenario, which comprises the following steps: receiving a training signal and an arbitrary unknown signal transmitted by the other end at a wireless access point / user end, differentiating the training signal, generating a wireless channel matrix, generating a signal vector according to the arbitrary unknown signal and the wireless channel matrix, processing all the signal vectors to obtain all the link signals of the transmitting end; training a radio frequency fingerprint feature identification model using all the link signals and generating a device radio frequency fingerprint library; inputting an unidentified signal of a device to be authenticated into the model for identification to obtain a radio frequency fingerprint feature of the device to be authenticated, and comparing the radio frequency fingerprint feature with the device radio frequency fingerprint library to obtain a radio frequency fingerprint identification result of the device to be authenticated. The application adopting the above scheme can simultaneously realize separation of multiple devices and multiple link signals of each device and ensure that fingerprint aliasing does not occur, so as to realize radio frequency fingerprint identification of multiple devices.

Description

Technical Field

[0001] This application relates to the field of wireless network access authentication technology, and in particular to a method and apparatus for radio frequency fingerprinting of MIMO devices in multi-user wireless communication scenarios. Background Technology

[0002] In real-world environments, wireless networks are highly vulnerable to various eavesdropping and interference attacks, leading to a range of cybersecurity threats such as unauthorized access and data breaches. Strict access authentication for devices is one of the primary means of preventing these threats. Existing access authentication technologies are mainly divided into two categories: the first is cryptographic-based access authentication technology, and the second is radio frequency fingerprinting technology.

[0003] Cryptographic authentication technologies employ methods such as encryption, hash functions, and digital signatures to authenticate devices based on a secret value possessed by the device being authenticated. For example, in a WiFi wireless network, wireless access points (or routers) typically use the WPA or WPA2 protocol to authenticate mobile devices. The WPA or WPA2 protocol uses encryption algorithms and performs authentication and key exchange based on the same pre-shared key (i.e., secret value) used by both parties.

[0004] Radio frequency (RF) fingerprinting technology authenticates devices based on their hardware characteristics or the features of the wireless channel. It can be divided into two categories: the first is wireless channel-based RF fingerprinting, which identifies the wireless channel characteristics between the device being authenticated and the authenticating device; the second is hardware characteristic-based RF fingerprinting, which identifies the hardware characteristics of the device being authenticated. Here, hardware characteristics mainly refer to the circuit tolerances of the device's RF front-end components and related circuits that deviate from their nominal values ​​due to factors such as materials, manufacturing processes, and aging. Specific examples include clock jitter, DAC sampling errors, mixer frequency offset, power amplifier nonlinearity, and IQ imbalance.

[0005] Cryptographic authentication technologies increase message size, transmission overhead, and latency, making it difficult to meet the high-bandwidth, low-latency communication requirements of wireless devices with limited computing power. Furthermore, with the development of cryptanalysis, significant vulnerabilities in older cryptographic methods are constantly being discovered. Quantum computers can also efficiently break most cryptographic methods, posing a significant threat to cryptographic authentication technologies. While hardware-based radio frequency (RF) fingerprinting avoids these drawbacks, existing research largely focuses on RF fingerprint extraction and identification for single-antenna devices. There is still no clear solution for RF fingerprinting of all access users in multi-antenna, multi-link devices (MIMO devices) or scenarios where multiple users communicate simultaneously. Summary of the Invention

[0006] The first objective of this application is to propose a radio frequency fingerprinting method for MIMO devices in multi-user wireless communication scenarios. This method solves the technical problem of existing methods not having fingerprints for multi-antenna, multi-link devices. It achieves the simultaneous separation of multiple devices and multiple link signals of each device, and completely preserves the fingerprint information of the corresponding link during separation without fingerprint aliasing. This allows the fingerprints of multiple links of a single MIMO transmitter to be extracted and identified using a method similar to that of a single-antenna, single-link transmitter.

[0007] The second objective of this application is to propose a radio frequency fingerprint recognition device for MIMO devices in a multi-user wireless communication scenario.

[0008] To achieve the above objectives, the first aspect of this application proposes a method for radio frequency fingerprinting of MIMO devices in a multi-user wireless communication scenario, comprising: receiving training signals and arbitrary unknown signals transmitted by a user terminal / wireless access point at a wireless access point / user terminal; performing differential processing on the training signals to generate a wireless channel matrix; generating signal vectors based on the arbitrary unknown signals and the wireless channel matrix; processing all signal vectors of all user devices to obtain link signals of all user devices; wherein, if the wireless access point is the signal transmitter, the user terminal is the signal receiver, and if the user terminal is the signal transmitter, the wireless access point is the receiver; the training signals carry training sequences, and the arbitrary unknown signals carry communication data; training a radio frequency fingerprint feature recognition model using all link signals of all devices as training data to obtain a trained radio frequency fingerprint feature recognition model, and generating a device radio frequency fingerprint database, wherein the device radio frequency fingerprint database is used to store device information and corresponding radio frequency fingerprint features generated during training; acquiring the identification signal of the device to be authenticated, inputting the identification signal into the trained radio frequency fingerprint feature recognition model for identification, obtaining the radio frequency fingerprint features of the device to be authenticated; comparing the radio frequency fingerprint features of the device to be authenticated with the device radio frequency fingerprint database to obtain the radio frequency fingerprint recognition result of the device to be authenticated.

[0009] The radio frequency fingerprinting method for MIMO devices in multi-user wireless communication scenarios in this application addresses the spatiotemporal aliasing phenomenon of signals from multiple devices at the wireless access point in uplink communication scenarios. It proposes a link separation technique to distinguish the signals of multiple devices and then extract and identify fingerprints for multiple devices. The link separation technique proposed in this application uses matrix operations to simultaneously separate multiple devices and multiple links of each device, and completely preserves the fingerprint information of the corresponding link during separation without fingerprint aliasing. This allows the fingerprints of multiple links of a single MIMO transmitter to be extracted and identified using a method similar to that of a single-antenna single-link transmitter.

[0010] Optionally, in one embodiment of this application, if the wireless communication scenario is an uplink communication scenario, the method includes:

[0011] The wireless access point receives training signals transmitted by multiple user terminals and performs differential analysis on the training signals to obtain the wireless channel matrix.

[0012] At each moment within a preset time period, it receives arbitrary unknown signals transmitted by multiple user terminals through a wireless access point;

[0013] The signal vector corresponding to each moment is calculated based on the arbitrary unknown signals transmitted by multiple users and the wireless channel matrix received at each moment;

[0014] The signal vectors of all times within a preset time period are concatenated, and the link signals of all devices are obtained from the concatenated signal vectors.

[0015] Optionally, in one embodiment of this application, the training signal is differentially divided to obtain a wireless channel matrix, including:

[0016] The training signals transmitted from multiple user terminals are differentially divided, and the differentially divided data are then spliced ​​together to obtain the first spliced ​​data.

[0017] The training signal received by the wireless access point is differentially divided, and the differentially divided data is then concatenated to obtain the second concatenated data.

[0018] The wireless channel matrix is ​​calculated based on the first and second spliced ​​data.

[0019] Optionally, in one embodiment of this application, the training signals transmitted simultaneously by multiple user terminals are represented as follows:

[0020] x t0 ,x t1 ,…,x tK (K>M)

[0021] Where t0, t1, ..., tK are the transmission times, M is the total number of antennas at the receiving end, and K is the number of training signals transmitted;

[0022] The training signal received by the wireless access point is represented as follows:

[0023]

[0024] y t1 =HAx t1 +Hb+w t1

[0025] …

[0026] y tK =HAx tK+Hb+w tK

[0027] Where t0, t1, ..., tK are the transmission times, H is the wireless channel matrix between the transceiver antennas stripped of the transmitter's RF fingerprint features, and A is determined by the RF fingerprint feature parameter α. i Sure, b is derived from the radio frequency fingerprint characteristic parameter β i Determine that b = {β1, β2, ..., β} M} T w represents noise;

[0028] The training signals transmitted from multiple user terminals, after differential processing, are represented as follows:

[0029]

[0030] The first concatenated data is represented as follows:

[0031] ΔX t =[Δx t1 ,Δx t2 ,…,Δx tK ]

[0032] The training signal received by the wireless access point, after differential processing, is represented as follows:

[0033]

[0034] The second concatenated data is represented as follows:

[0035] ΔY t =[Δy t1 ,Δy t2 ,…,Δy tK ]

[0036] The wireless channel matrix is ​​represented as follows:

[0037]

[0038] Where, ΔY t This represents the second concatenated data, ΔX. t This indicates the first concatenated data;

[0039] The signal vector is represented as:

[0040]

[0041] Where y represents the number of signals received by the wireless access point, y = {y1, y2, ..., y} N} T , The wireless channel matrix is ​​represented as follows; the concatenated signal vector is represented as:

[0042] Z = [z1, z2, ...]

[0043] All link signals of all devices are represented as follows:

[0044]

[0045] in, Z represents the set of all link sequence numbers contained in the l-th device. j Let Z represent the j-th row vector.

[0046] Optionally, in one embodiment of this application, if the wireless communication scenario is a downlink communication scenario, the method includes:

[0047] The user receives multiple training signals transmitted by the wireless access point over multiple time periods, wherein the time periods are randomly selected by the wireless access point.

[0048] The wireless channel matrix is ​​obtained by differentiating multiple training signals;

[0049] The wireless channel matrix is ​​split into multiple time periods to obtain the wireless channel matrix corresponding to each time period;

[0050] The user terminal receives all unknown signals transmitted by the wireless access point within multiple time periods.

[0051] The signal vector for each time period is generated by calculating the wireless channel matrix corresponding to each time period and the unknown signals received by the user terminal within each time period. The signal vectors of all time periods are then concatenated to obtain the concatenated signal vector.

[0052] Obtain the spliced ​​signal vectors from all user terminals, and then obtain all link signals of all devices based on the spliced ​​signal vectors from all user terminals.

[0053] Optionally, in one embodiment of this application, the wireless channel matrix is ​​obtained by differentially dividing multiple training signals, including:

[0054] The first differential data is obtained by differentially analyzing multiple training signals transmitted by the wireless access point.

[0055] The second difference data is obtained by performing differential analysis on multiple training signals received by the user terminal.

[0056] The wireless channel matrix is ​​calculated based on the first and second differential data.

[0057] Optionally, in one embodiment of this application, the multiple training signals transmitted by the wireless access point are represented as follows:

[0058] x t1 ,xt2 ,…,x tK

[0059] The multiple training signals received by the user are represented as follows:

[0060]

[0061] Among them, [Y j ] N×K H represents the signal received by the user terminal in the j-th time period. j Let A be the channel state corresponding to the j-th time period, and let A be the RF fingerprint feature parameter α. i Sure, b is derived from the radio frequency fingerprint characteristic parameter β i Determine that b = {β1, β2, ..., β} M} T w represents noise;

[0062] The first difference data is represented as follows:

[0063] ΔX=[x t2 -x t1 ,x t3 -x t2 ,…,x tK -x t(K-1) ] M×(K-1)

[0064] Where M is the number of radio frequency links of the wireless access point, and K is the number of training signals;

[0065] The second difference data is represented as follows:

[0066]

[0067] Where N is the number of radio frequency links at the user end, K is the number of training signals, and P is the number of time periods;

[0068] The wireless channel matrix is ​​represented as follows:

[0069]

[0070] in,

[0071] The split wireless channel matrix is ​​represented as follows:

[0072] H1A,H2A,…,H P A

[0073] Where P is the number of time periods;

[0074] The signal vector is represented as:

[0075] zM×K =(H j A) + y

[0076] Among them, z M×K To receive signal y N×K The corresponding signal vector, j is the time period in which the signal is located, H j H represents the channel state corresponding to the j-th time period. j A represents the j-th wireless channel matrix after splitting;

[0077] The concatenated signal vector is represented as follows:

[0078] Z = [z1, z2, ...]

[0079] All link signals of all devices are represented as follows:

[0080]

[0081] in, Z represents the set of all link sequence numbers contained in the l-th device. j Let Z represent the j-th row vector.

[0082] Optionally, in one embodiment of this application, the radio frequency fingerprint feature recognition model is a deep neural network. The model is trained using all link signals from all devices as training data to establish a device radio frequency fingerprint database, including:

[0083] Obtain the preprocessed training signals corresponding to all link signals of all devices;

[0084] With the goal of minimizing the error between the actual device label in the input signal and the device label estimated by the neural network, a deep neural network model is trained to obtain a well-trained radio frequency fingerprint feature recognition model.

[0085] The obtained device information and corresponding radio frequency fingerprint features are stored in the device radio frequency fingerprint database.

[0086] To achieve the above objectives, a second aspect of the present invention provides a radio frequency fingerprinting device for MIMO devices in a multi-user wireless communication scenario, comprising a signal processing module, a model training module, and a recognition module, wherein...

[0087] The signal processing module is used to receive training signals and arbitrary unknown signals transmitted by the user terminal / wireless access point at the wireless access point / user terminal, perform differential processing on the training signals to generate a wireless channel matrix, generate signal vectors based on the arbitrary unknown signals and the wireless channel matrix, process all signal vectors of all user equipment, and obtain the link signals of all user equipment. Wherein, if the wireless access point is the signal transmitter, the user terminal is the signal receiver; if the user terminal is the signal transmitter, the wireless access point is the receiver. The training signal carries the training sequence, and the arbitrary unknown signal carries communication data.

[0088] The model training module is used to train the radio frequency fingerprint feature recognition model using all link signals of all devices as training data, to obtain the trained radio frequency fingerprint feature recognition model, and to generate a device radio frequency fingerprint library. The device radio frequency fingerprint library is used to store the device information and corresponding radio frequency fingerprint features generated during training.

[0089] The identification module is used to acquire the identification signal of the device to be authenticated, input the identification signal into the trained radio frequency fingerprint feature recognition model for identification, obtain the radio frequency fingerprint feature of the device to be authenticated, compare the radio frequency fingerprint feature of the device to be authenticated with the device radio frequency fingerprint database, and obtain the radio frequency fingerprint recognition result of the device to be authenticated.

[0090] Optionally, in one embodiment of this application, the radio frequency fingerprint feature recognition model is a deep neural network. The model is trained using all link signals from all devices as training data to establish a device radio frequency fingerprint database, including:

[0091] Obtain the preprocessed training signals corresponding to all link signals of all devices;

[0092] With the goal of minimizing the error between the actual device label in the input signal and the device label estimated by the neural network, a deep neural network model is trained to obtain a well-trained radio frequency fingerprint feature recognition model.

[0093] The obtained device information and corresponding radio frequency fingerprint features are stored in the device radio frequency fingerprint database.

[0094] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0095] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0096] Figure 1 This is a schematic diagram of multi-user MIMO uplink and downlink wireless communication according to an embodiment of this application;

[0097] Figure 2 This is a flowchart illustrating a radio frequency fingerprinting method for MIMO devices in a multi-user wireless communication scenario provided in Embodiment 1 of this application.

[0098] Figure 3 This is a schematic diagram of the structure of a radio frequency fingerprint recognition device for a MIMO device in a multi-user wireless communication scenario, provided in an embodiment of this application. Detailed Implementation

[0099] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0100] In the existing multi-user MIMO wireless communication environment, multiple terminal users and a wireless access point conduct periodic data communication, with uplink and downlink communication alternating. Figure 1 This is a schematic diagram of multi-user MIMO uplink and downlink wireless communication according to an embodiment of this application. Figure 1 (a) is an uplink communication scenario. Figure 1 (b) is a downlink communication scenario, such as Figure 1 As shown in (a), during uplink communication, multiple users simultaneously send signals to the wireless access point, such as... Figure 1 As shown in (b), during downlink communication, the wireless access point sends signals to multiple users. In existing communication standards, during multi-user MIMO communication, the total number of antennas of the wireless access point is greater than or equal to the total number of antennas of the multiple users.

[0101] In the uplink communication scenario, the signal transmitter consists of multiple users, and the signal receiver is a wireless access point. Let M be the total number of antennas at the transmitter and N be the total number of antennas at the receiver, where N ≥ M. In the downlink communication scenario, the signal transmitter is a wireless access point, and the signal receiver consists of multiple users. Let M be the total number of antennas at the transmitter and N be the total number of antennas at the receiver, where N ≤ M.

[0102] At this point, the relationship between the received signal and the transmitted signal is:

[0103] y = HAx + Hb + w

[0104] Where x is the ideal transmitted signal from the transmitter, H is the wireless channel matrix between the transmit and receive antennas stripped of the transmitter's RF fingerprint characteristics (it can be reasonably assumed that the communication channel states of each user remain basically unchanged over a short period of time), y is the multi-user aliasing signal received by the receiver, w is noise, and A and b are related to the transmitter's RF fingerprint and satisfy the following:

[0105]

[0106] b = {β1,β2,…,β} M} T

[0107] In both uplink and downlink communication scenarios, this invention proposes corresponding radio frequency link separation methods for multi-user aliasing signals received by the receiver.

[0108] Let the fingerprint feature of the i-th radio frequency link be . Its richness in frequency dependence and uniqueness in device association determine that it can be used for device radio frequency fingerprint recognition.

[0109] The following describes a method and apparatus for radio frequency fingerprinting of MIMO devices in a multi-user wireless communication scenario, according to embodiments of this application, with reference to the accompanying drawings.

[0110] Figure 2 This is a flowchart illustrating a radio frequency fingerprinting method for MIMO devices in a multi-user wireless communication scenario, as provided in Embodiment 1 of this application.

[0111] like Figure 2 As shown, the RF fingerprinting method for MIMO devices in this multi-user wireless communication scenario includes the following steps:

[0112] Step 201: Receive training signals and arbitrary unknown signals transmitted by the user terminal / wireless access point at the wireless access point / user terminal, perform differential processing on the training signals to generate a wireless channel matrix, generate signal vectors based on the arbitrary unknown signals and the wireless channel matrix, process all signal vectors of all user equipment to obtain the link signals of all user equipment. Wherein, if the wireless access point is the signal transmitter, the user terminal is the signal receiver; if the user terminal is the signal transmitter, the wireless access point is the receiver. The training signal carries the training sequence, and the arbitrary unknown signal carries communication data.

[0113] Step 202: Use all link signals of all devices as training data to train the radio frequency fingerprint feature recognition model, obtain the trained radio frequency fingerprint feature recognition model, and generate a device radio frequency fingerprint library. The device radio frequency fingerprint library is used to store the device information and corresponding radio frequency fingerprint features generated during training.

[0114] Step 203: Obtain the identification signal of the device to be authenticated, and input the identification signal into the trained radio frequency fingerprint feature recognition model for recognition to obtain the radio frequency fingerprint feature of the device to be authenticated. Compare the radio frequency fingerprint feature of the device to be authenticated with the device radio frequency fingerprint database to obtain the radio frequency fingerprint recognition result of the device to be authenticated.

[0115] The radio frequency fingerprinting method for MIMO devices in multi-user wireless communication scenarios in this application addresses the spatiotemporal aliasing phenomenon of signals from multiple devices at the wireless access point in uplink communication scenarios. It proposes a link separation technique to distinguish the signals of multiple devices and then extract and identify fingerprints for multiple devices. The link separation technique proposed in this application uses matrix operations to simultaneously separate multiple devices and multiple links of each device, and completely preserves the fingerprint information of the corresponding link during separation without fingerprint aliasing. This allows the fingerprints of multiple links of a single MIMO transmitter to be extracted and identified using a method similar to that of a single-antenna single-link transmitter.

[0116] The radio frequency fingerprinting method for MIMO devices in multi-user wireless communication scenarios in this application embodiment can be applied to uplink and downlink communication scenarios in multi-user MIMO communication environments. The process of separating the radio frequency link differs between the two scenarios, but the process of fingerprint extraction and recognition is the same.

[0117] Optionally, in one embodiment of this application, if the wireless communication scenario is an uplink communication scenario, the method includes:

[0118] Multiple users simultaneously transmit K training signals to a wireless access point, and the training signals are differentially analyzed. The wireless access point then calculates the wireless channel matrix.

[0119] Subsequently, at some point, several users transmit arbitrary unknown signals x (these signals carry communication data but not training sequences);

[0120] A wireless access point receives N signals and can then use the received signals and the calculated wireless channel matrix to determine the appropriate channels. Calculate the signal vector z M×1 ;

[0121] The wireless access point concatenates signal vectors from several time points and, based on the MAC layer protocol in multi-user MIMO communication, directly queries the radiation source attribution information in the signal to obtain the set of all link numbers contained in the l-th device. Taking multi-user MIMO uplink communication in the IEEE 802.11ax standard as an example, the uplink communication process is controlled by the wireless access point through trigger frames. The trigger frame explicitly defines the USER INFO field for each user, including each user's 12-bit identity information (Associate Identifier, AID), resource unit (RU) allocation, encoding, modulation information, etc., so that all link signals of all devices can be obtained.

[0122] Optionally, in one embodiment of this application, the training signal is differentially divided to obtain a wireless channel matrix, including:

[0123] The training signals transmitted from multiple user terminals are differentially divided, and the differentially divided data are then spliced ​​together to obtain the first spliced ​​data.

[0124] The training signal received by the wireless access point is differentially divided, and the differentially divided data is then concatenated to obtain the second concatenated data.

[0125] The wireless access point calculates the wireless channel matrix based on the first and second spliced ​​data.

[0126] Optionally, in one embodiment of this application, the training signals transmitted simultaneously by multiple user terminals are represented as follows:

[0127] x t0 ,x t1 ,…,x tK (K>M)

[0128] Where the transmission times are t0, t1, ..., tK, M is the total number of antennas at the receiving end, and K is the number of training signals transmitted;

[0129] The training signal received by the wireless access point is represented as follows:

[0130]

[0131] y t1 =HAx t1 +Hb+w t1

[0132] …

[0133] y tK =HAx tK +Hb+w tK

[0134] Where t0, t1, ..., tK are the transmission times, H is the wireless channel matrix between the transceiver antennas after the transmitter's RF fingerprint characteristics have been removed, A and b are related to the transmitter's RF fingerprint, and w is noise;

[0135] For the transmitted training signal x t0 ,x t1 ,…,x tK Difference:

[0136]

[0137] splicing Δx t1 ,Δx t2 ,…Δx tK We can obtain:

[0138] ΔX t =[Δx t1 ,Δx t2 ,…,Δx tK ]

[0139] Differentiate the signals received at different times:

[0140]

[0141] Wherein, ΔW tj The noise term is represented by H, which is the wireless channel matrix between the transceiver antennas after the transmitter's RF fingerprint characteristics have been removed, and A is related to the transmitter's RF fingerprint.

[0142] splicing Δy t1 ,Δy t2 ..., we can obtain:

[0143] ΔY t =[Δy t1 ,Δy t2 ,…,Δy tK ]=HAΔX t +ΔW t

[0144] Where H is the wireless channel matrix between the transceiver antennas after the transmitter's RF fingerprint characteristics have been removed, A is related to the transmitter's RF fingerprint, and ΔW t Indicates the noise term;

[0145] Wireless access point calculation

[0146]

[0147] in,(*) + Let be the generalized inverse of a matrix, ignoring the noise term ΔW. t The influence of ΔY t This represents the second concatenated data, ΔX. t This indicates the first concatenated data;

[0148] The wireless access point receives N signals y = {y1, y2, ..., y} N} T Signal vector z M×1 Represented as:

[0149]

[0150] Where y represents the number of signals received by the wireless access point, y = {y1, y2, ..., y} N} T w represents noise;

[0151] The concatenated signal vector is represented as follows:

[0152] Z = [z1, z2, ...]

[0153] All link signals of all devices are represented as follows:

[0154]

[0155] in, Z represents the set of all link sequence numbers contained in the l-th device. j Let Z represent the j-th row vector.

[0156] Optionally, in one embodiment of this application, in a downlink communication scenario, it is necessary for a user terminal with fewer antennas to perform radio frequency fingerprinting on a wireless access point with more antennas. Due to the underrank characteristic of the wireless channel matrix, the radio frequency link separation method in the uplink communication scenario is no longer applicable to downlink communication. If the wireless communication scenario is a downlink communication scenario, the method includes:

[0157] The wireless access point selects P time periods, and the channel states H1, H2, ..., H1 corresponding to the P time periods are... P They are all different. K training signals are transmitted to the user terminal in each time period.

[0158] The user terminal performs differential analysis on the K training signals to obtain the wireless channel matrix.

[0159] By splitting The user can obtain matrices H1A, H2A, ..., H for P time periods respectively. P A;

[0160] Subsequently, for all unknown signals x transmitted by the wireless access point within the past P time periods, the corresponding received signals y at the user end are obtained. N×1 ;

[0161] The user terminal calculates and generates the signal vector for each time period based on the wireless channel matrix corresponding to each time period and the unknown signals received in each time period. The signal vectors of all time periods are then concatenated to obtain the concatenated signal vector.

[0162] Obtain the spliced ​​signal vectors from all user terminals, and then obtain all link signals of all devices based on the spliced ​​signal vectors from all user terminals.

[0163] Optionally, in one embodiment of this application, the wireless channel matrix is ​​obtained by differentially dividing multiple training signals, including:

[0164] The first differential data is obtained by differentially analyzing the K training signals transmitted by the wireless access point.

[0165] The second difference data is obtained by differentially analyzing the K training signals received by the user terminal.

[0166] The wireless channel matrix is ​​calculated based on the first and second differential data.

[0167] Optionally, in one embodiment of this application, the multiple training signals transmitted by the wireless access point are represented as follows:

[0168] x t1 ,x t2 ,…,x tK

[0169] The signal received by the user terminal in the j-th time period (1≤j≤P) is:

[0170]

[0171] Among them, H j This represents the channel state corresponding to the j-th time period;

[0172] Calculate the differences between the K training signals:

[0173] ΔX=[x t2 -x t1 ,x t3 -x t2 ,…, tK -x t(K-1) ] M×(K-1)

[0174] and the difference of the received signal in the j-th time period:

[0175]

[0176] It can be known

[0177] ΔY j =H j AΔX+ΔW j

[0178] definition

[0179]

[0180]

[0181] Then there is

[0182]

[0183] The user terminal estimates the wireless channel matrix using the following formula.

[0184]

[0185] The matrix representation of the P time periods after splitting is as follows:

[0186] H1A,H2A,…,H P A

[0187] For the received signal y N×1 Calculate the signal vector z N×1 :

[0188] z M×K =(H j A) + y

[0189] ≈x+A -1 b

[0190] Where j represents the time period in which the signal occurs, and based on the short-time invariance assumption of the radio frequency fingerprint matrix A and vector b, each row z in vector z... i Only the signal transmitted by the i-th RF link at the transmitter, and the RF fingerprint characteristic parameter α i ,β i And related to receiver noise;

[0191] The user end concatenates several signal vectors:

[0192] Z = [z1, z2, ...]

[0193] Unlike the uplink communication scenario, all link signals in Z belong to one device, namely the wireless access point;

[0194] All link signals of all devices are represented as follows:

[0195]

[0196] in, Z represents the set of all link sequence numbers contained in the l-th device. j Let Z represent the j-th row vector.

[0197] Optionally, in one embodiment of this application, the radio frequency fingerprint feature recognition model is a deep neural network. The model is trained using all link signals from all devices as training data to establish a device radio frequency fingerprint database. This includes: extracting device radio frequency fingerprints, training the model, and storing the trained data to obtain the device radio frequency fingerprint database. Specifically:

[0198] All link signals of all devices {z1,z2,…,z qThe training set of the self-supervised denoising autoencoder aims to minimize the error between the input signal and the denoised and reconstructed signal, training the autoencoder to learn the signal to be recognized {z1, z2, ..., z}. q The latent hidden features and encoder / decoder neural network parameters.

[0199] The device information and the corresponding trained RF fingerprint features are stored in the RF fingerprint database to generate the device RF fingerprint database.

[0200] To achieve the above embodiments, this application also proposes a radio frequency fingerprint recognition device for MIMO devices in a multi-user wireless communication scenario.

[0201] Figure 3 This is a schematic diagram of the structure of a radio frequency fingerprint recognition device for a MIMO device in a multi-user wireless communication scenario, provided in an embodiment of this application.

[0202] like Figure 3 As shown, the RF fingerprinting device for MIMO devices in this multi-user wireless communication scenario includes a signal processing module, a model training module, and a recognition module.

[0203] The signal processing module is used to receive training signals and arbitrary unknown signals transmitted by the user terminal / wireless access point at the wireless access point / user terminal, perform differential processing on the training signals to generate a wireless channel matrix, generate signal vectors based on the arbitrary unknown signals and the wireless channel matrix, process all signal vectors of all user equipment, and obtain the link signals of all user equipment. Wherein, if the wireless access point is the signal transmitter, the user terminal is the signal receiver; if the user terminal is the signal transmitter, the wireless access point is the receiver. The training signal carries the training sequence, and the arbitrary unknown signal carries communication data.

[0204] The model training module is used to train the radio frequency fingerprint feature recognition model using all link signals of all devices as training data, to obtain the trained radio frequency fingerprint feature recognition model, and to generate a device radio frequency fingerprint library. The device radio frequency fingerprint library is used to store the device information and corresponding radio frequency fingerprint features generated during training.

[0205] The identification module is used to acquire the identification signal of the device to be authenticated, input the identification signal into the trained radio frequency fingerprint feature recognition model for identification, obtain the radio frequency fingerprint feature of the device to be authenticated, compare the radio frequency fingerprint feature of the device to be authenticated with the device radio frequency fingerprint database, and obtain the radio frequency fingerprint recognition result of the device to be authenticated.

[0206] Optionally, in one embodiment of this application, the radio frequency fingerprint feature recognition model is a deep neural network. The model is trained using all link signals from all devices as training data to establish a device radio frequency fingerprint database, including:

[0207] Obtain the preprocessed training signals corresponding to all link signals of all devices;

[0208] With the goal of minimizing the error between the actual device label in the input signal and the device label estimated by the neural network, a deep neural network model is trained to obtain a well-trained radio frequency fingerprint feature recognition model.

[0209] The obtained device information and corresponding radio frequency fingerprint features are stored in the device radio frequency fingerprint database.

[0210] It should be noted that the foregoing explanation of the MIMO device radio frequency fingerprint recognition method embodiment in multi-user wireless communication scenario also applies to the MIMO device radio frequency fingerprint recognition device in multi-user wireless communication scenario of this embodiment, and will not be repeated here.

[0211] 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, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0212] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0213] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0214] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0215] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0216] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0217] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0218] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for radio frequency fingerprint recognition of MIMO devices in a multi-user wireless communication scenario, characterized in that, Includes the following steps: The system receives training signals and arbitrary unknown signals transmitted by user terminals / wireless access points, performs differential processing on the training signals to generate a wireless channel matrix, generates signal vectors based on the arbitrary unknown signals and the wireless channel matrix, processes all signal vectors of all user equipment, and obtains the link signals of all user equipment. Wherein, if the wireless access point is the signal transmitter, the user terminal is the signal receiver; if the user terminal is the signal transmitter, the wireless access point is the receiver. The training signals carry training sequences, and the arbitrary unknown signals carry communication data. The radio frequency fingerprint feature recognition model is trained using all link signals of all devices as training data to obtain a trained radio frequency fingerprint feature recognition model and generate a device radio frequency fingerprint database. The device radio frequency fingerprint database is used to store the device information and corresponding radio frequency fingerprint features generated during training. The identification signal of the device to be authenticated is obtained and input into the trained radio frequency fingerprint feature recognition model for identification to obtain the radio frequency fingerprint feature of the device to be authenticated. The radio frequency fingerprint feature of the device to be authenticated is compared with the radio frequency fingerprint database of the device to obtain the radio frequency fingerprint recognition result of the device to be authenticated. Wherein, if the wireless communication scenario is an uplink communication scenario, the method includes: The wireless access point receives training signals transmitted by multiple user terminals and performs differential processing on the training signals to obtain the wireless channel matrix. During each moment of a preset time period, the wireless access point receives arbitrary unknown signals transmitted by multiple user terminals. The signal vector corresponding to each moment is calculated based on the arbitrary unknown signals transmitted by multiple users received at each moment and the wireless channel matrix. The signal vectors of all times within the preset time period are concatenated, and all link signals of all devices are obtained based on the concatenated signal vectors. The step of differentiating the training signal to obtain the wireless channel matrix includes: The training signals transmitted by the multiple user terminals are differentially divided, and the differentially divided data are spliced ​​together to obtain the first spliced ​​data; The training signal received by the wireless access point is differentially divided, and the differentially divided data is then concatenated to obtain the second concatenated data. The wireless channel matrix is ​​calculated based on the first spliced ​​data and the second spliced ​​data; The training signals transmitted by the multiple user terminals are represented as follows: in, For the launch time, This represents the total number of antennas at the receiving end. K represents the number of training signals transmitted, where K > M; The training signal received by the wireless access point is represented as follows: in, For the launch time, To create a wireless channel matrix between the transceiver antennas after the transmitter's radio frequency fingerprint characteristics have been removed. From radio frequency fingerprint feature parameters Sure, , From radio frequency fingerprint feature parameters Sure, , For noise; The training signals transmitted by the multiple user terminals, after differential processing, are represented as follows: The first concatenated data is represented as follows: The training signal received by the wireless access point, after differential processing, is represented as follows: The second concatenated data is represented as follows: The wireless channel matrix is ​​represented as follows: in, This represents the second concatenated data. This represents the first concatenated data; The signal vector is represented as: in, This indicates that the wireless access point received... road signal, , This represents the wireless channel matrix; The concatenated signal vector is represented as follows: All link signals of all the devices are represented as follows: in, Indicates the first The set of all link sequence numbers contained in each device. express The Row vectors.

2. The method as described in claim 1, characterized in that, If the wireless communication scenario is a downlink communication scenario, the method includes: The user terminal receives multiple training signals transmitted by the wireless access point within multiple time periods, wherein the time periods are randomly selected by the wireless access point. The wireless channel matrix is ​​obtained by differentially analyzing the multiple training signals. The wireless channel matrix is ​​split according to the multiple time periods to obtain the wireless channel matrix corresponding to each time period; The user terminal receives all unknown signals transmitted by the wireless access point during the multiple time periods; The signal vector for each time period is generated by calculating the wireless channel matrix corresponding to each time period and the unknown signals received by the user terminal within each time period. The signal vectors of all time periods are then concatenated to obtain the concatenated signal vector. Obtain the spliced ​​signal vectors from all user terminals, and based on the spliced ​​signal vectors from all user terminals, obtain all link signals of all devices.

3. The method as described in claim 2, characterized in that, The step of differentiating the plurality of training signals to obtain the wireless channel matrix includes: The multiple training signals transmitted by the wireless access point are differentially divided to obtain the first differential data; The multiple training signals received by the user terminal are differentially analyzed to obtain the second differential data; The wireless channel matrix is ​​calculated based on the first differential data and the second differential data.

4. The method as described in claim 3, characterized in that, The multiple training signals transmitted by the wireless access point are represented as follows: The multiple training signals received by the user terminal are represented as follows: in, For the user terminal in the first Signals received within a time period This represents the channel state corresponding to the j-th time period. From radio frequency fingerprint feature parameters Sure, , From radio frequency fingerprint feature parameters Sure, , For noise; The first difference data is represented as follows: Where M is the number of radio frequency links of the wireless access point, and K is the number of training signals; The second difference data is represented as follows: Where N is the number of radio frequency links at the user end, K is the number of training signals, and P is the number of time periods; The wireless channel matrix is ​​represented as follows: in, ; The split wireless channel matrix is ​​represented as follows: in, This represents the number of time periods; The signal vector is represented as: in, To receive signals The corresponding signal vector, The time period in which the signal is located. This represents the channel state corresponding to the j-th time period. This represents the j-th wireless channel matrix after splitting; The concatenated signal vector is represented as follows: All link signals of all the devices are represented as follows: in, Indicates the first The set of all link sequence numbers contained in each device. express The Row vectors.

5. The method as described in claim 1, characterized in that, The radio frequency fingerprint feature recognition model is a deep neural network. All link signals from all devices are used as training data to train the radio frequency fingerprint feature recognition model and establish a device radio frequency fingerprint database, including: Obtain the preprocessed training signals corresponding to all link signals of all devices; With the goal of minimizing the error between the actual device label in the input signal and the device label estimated by the neural network, a deep neural network model is trained to obtain a well-trained radio frequency fingerprint feature recognition model. The obtained device information and corresponding radio frequency fingerprint features are stored in the device radio frequency fingerprint database.

6. A radio frequency fingerprint recognition device for MIMO devices in a multi-user wireless communication scenario, characterized in that, It includes a signal processing module, a model training module, and a recognition module, among which, The signal processing module is used to receive training signals and arbitrary unknown signals transmitted by the user terminal / wireless access point at the wireless access point / user terminal, perform differential processing on the training signals to generate a wireless channel matrix, generate signal vectors based on the arbitrary unknown signals and the wireless channel matrix, process all signal vectors of all user equipment, and obtain the link signals of all user equipment. Wherein, if the wireless access point is the signal transmitter, the user terminal is the signal receiver; if the user terminal is the signal transmitter, the wireless access point is the receiver. The training signals carry training sequences, and the arbitrary unknown signals carry communication data. The model training module is used to train the radio frequency fingerprint feature recognition model using all link signals of all devices as training data, to obtain the trained radio frequency fingerprint feature recognition model, and to generate a device radio frequency fingerprint library, wherein the device radio frequency fingerprint library is used to store the device information and corresponding radio frequency fingerprint features generated during training. The identification module is used to acquire the identification signal of the device to be authenticated, input the identification signal into the trained radio frequency fingerprint feature recognition model for identification, obtain the radio frequency fingerprint feature of the device to be authenticated, and compare the radio frequency fingerprint feature of the device to be authenticated with the device radio frequency fingerprint database to obtain the radio frequency fingerprint recognition result of the device to be authenticated. Wherein, if the wireless communication scenario is an uplink communication scenario, the method includes: The wireless access point receives training signals transmitted by multiple user terminals and performs differential processing on the training signals to obtain the wireless channel matrix. During each moment of a preset time period, the wireless access point receives arbitrary unknown signals transmitted by multiple user terminals. The signal vector corresponding to each moment is calculated based on the arbitrary unknown signals transmitted by multiple users received at each moment and the wireless channel matrix. The signal vectors of all times within the preset time period are concatenated, and all link signals of all devices are obtained based on the concatenated signal vectors. The step of differentiating the training signal to obtain the wireless channel matrix includes: The training signals transmitted by the multiple user terminals are differentially divided, and the differentially divided data are spliced ​​together to obtain the first spliced ​​data; The training signal received by the wireless access point is differentially divided, and the differentially divided data is then concatenated to obtain the second concatenated data. The wireless channel matrix is ​​calculated based on the first spliced ​​data and the second spliced ​​data; The training signals transmitted by the multiple user terminals are represented as follows: in, For the launch time, This represents the total number of antennas at the receiving end. K represents the number of training signals transmitted, where K > M; The training signal received by the wireless access point is represented as follows: in, For the launch time, To create a wireless channel matrix between the transceiver antennas after the transmitter's radio frequency fingerprint characteristics have been removed. From radio frequency fingerprint feature parameters Sure, , From radio frequency fingerprint feature parameters Sure, , For noise; The training signals transmitted by the multiple user terminals, after differential processing, are represented as follows: The first concatenated data is represented as follows: The training signal received by the wireless access point, after differential processing, is represented as follows: The second concatenated data is represented as follows: The wireless channel matrix is ​​represented as follows: in, This represents the second concatenated data. This represents the first concatenated data; The signal vector is represented as: in, This indicates that the wireless access point received... road signal, , This represents the wireless channel matrix; The concatenated signal vector is represented as follows: All link signals of all the devices are represented as follows: in, Indicates the first The set of all link sequence numbers contained in each device. express The Row vectors.

7. The apparatus as claimed in claim 6, characterized in that, The radio frequency fingerprint feature recognition model is a deep neural network. All link signals from all devices are used as training data to train the radio frequency fingerprint feature recognition model and establish a device radio frequency fingerprint database, including: Obtain the preprocessed training signals corresponding to all link signals of all devices; With the goal of minimizing the error between the actual device label in the input signal and the device label estimated by the neural network, a deep neural network model is trained to obtain a well-trained radio frequency fingerprint feature recognition model. The obtained device information and corresponding radio frequency fingerprint features are stored in the device radio frequency fingerprint database.

Images