Large-scale MTC unlicensed channel estimation and user detection method and system against PCA

Abstract

The present invention discloses a large-scale MTC unlicensed channel estimation and user detection method and system for combating PCA: a millimeter wave uplink multi-input multi-output system model is constructed based on the unlicensed mMTC network, and a three-dimensional transmission model with time-correlated two-dimensional sparsity is constructed based on the three-dimensional received signal matrix and the three-dimensional CVR matrix; based on the constructed three-dimensional transmission model, joint channel estimation, active user detection, and pilot pollution attack detection can be converted into a sparse signal recovery problem of recovering the sparse CVR matrix from the received signal in T consecutive time slots, and the problem is expressed as the following three-dimensional multi-measurement vector compressed sensing problem; the problem is solved by a parallel multi-measurement vector approximate message passing algorithm based on expectation maximization to obtain an estimated active LU set, PCA attacker set, and LU channel matrix. The present invention improves the accuracy of user activity detection and data decoding, improves the correctness of pilot pollution attack detection, and obtains better system performance.

Description

Technical Field

[0001] The present invention belongs to the field of random access and data transmission in large-scale MTC networks, and particularly relates to a large-scale MTC unlicensed channel estimation and user detection method and system for combating PCA. Background Art

[0002] Massive machine-type communications (mMTC) is one of the three key application scenarios of the fifth-generation mobile communication technology (5G). It features large-scale connections, small data packets, and sporadic transmission. To address the significant signaling overhead and access latency in traditional random access, unauthorized access has attracted considerable research interest. This approach allows users to directly transmit data without requiring authorization from the base station, improving system communication efficiency. However, since the base station lacks user identity authentication during the unauthorized access process and distinguishes users solely by identifying received pilot sequences, mMTC networks are vulnerable to pilot contamination attacks (PCA). This allows attackers to impersonate legitimate users (LUs) by monitoring and transmitting identical pilots, thereby accessing the base station and severely degrading the base station's LU activity detection and channel estimation performance. Designing a joint active user detection and channel estimation scheme that is resistant to PCA remains a pressing technical challenge. Summary of the Invention

[0003] In order to solve the problems existing in the prior art, the present invention provides a large-scale MTC unlicensed channel estimation and user detection method against PCA, which distinguishes legitimate users from attackers by utilizing the sparsity of millimeter wave channels to resist pilot pollution attacks.

[0004] The present invention can also provide a large-scale MTC unlicensed millimeter wave uplink multiple-input multiple-output system to combat PCA, and perform joint channel estimation and active user detection based on the large-scale MTC unlicensed channel estimation and user detection method to combat PCA.

[0005] The present invention may also provide a computer device, including a processor and a memory, the memory being used to store a computer executable program, the processor reading the computer executable program from the memory and executing it, and the processor being able to implement the large-scale MTC unlicensed channel estimation and user detection method against PCA described in the present invention when executing the computer executable program.

[0006] A computer-readable storage medium is also provided. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the large-scale MTC unlicensed channel estimation and user detection method against PCA of the present invention can be implemented.

[0007] Compared with the existing technology, the present invention has at least the following advantages: the present invention establishes a three-dimensional transmission model with time-correlated two-dimensional sparsity to describe the sparsity of the active states of legitimate users and attackers, the sparsity of the angular domain channel, and the time correlation of the active states of legitimate users; secondly, based on the transmission model, a joint user activity detection and channel estimation scheme based on three-dimensional multi-measurement vector compressed sensing is proposed to combat pilot pollution attacks. By exploiting the differences in channel information between legitimate users and attackers, pilot pollution attacks can be detected, ensuring the security of mMTC network communications. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 Diagram of the three-dimensional transmission model of the millimeter wave unlicensed mMTC network established by the present invention.

[0009] Figure 2 The factor structure diagram of the algorithm designed for this invention.

[0010] Figure 3 2 is a curve showing the change of activity error rate with attack rate under the method of the present invention and the comparative scheme.

[0011] Figure 4 3 is a curve showing the change of the mean square error of channel estimation with the attack rate under the method of the present invention and the comparative scheme.

[0012] Figure 5 3 is a curve showing the change of the pilot pollution attack detection accuracy rate with the attack rate under the method of the present invention and the comparative scheme. DETAILED DESCRIPTION

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0014] The present invention provides a large-scale MTC unlicensed channel estimation and user detection method to combat PCA, comprising the following steps: the system constructs a millimeter wave uplink multiple-input multiple-output (MIMO) system for an unlicensed mMTC network, the system comprising a base station with M antennas, N single-antenna LUs, and multiple attackers initiating PCA. Assume that the LUs generate sporadic traffic and are fixed in position, while the attackers are variable in position. Let and They represent the LU index set and the base station antenna index set respectively.

[0015] Consider a geometric channel model with L scatterers, each of which generates a path. n The channel between time slots t can be expressed as:

[0016]

[0017] where ρ n represents the average path loss of LUn, β n，l represents the fading coefficient of the corresponding path l, θ n，l represents the physical angle-of-arrival (AoA) of path l at the base station, represents the antenna array response vector. Assuming the scatterer position is fixed, then θ n，l Since physical AoA is difficult to obtain directly, the channel virtual representation (CVR) is used to describe the millimeter wave channel. CVR can be regarded as the mapping of the actual physical channel in the virtual angle domain. Specifically, the channel vector The corresponding relationship with the channel virtual representation is:

[0018]

[0019] in represents CVR, U is the normalized discrete Fourier transform matrix, express The mth element of is equal to the gain of the mth virtual path, φ n，m Represents the virtual arrival angle corresponding to the virtual path m. Since millimeter wave communication has strong propagation loss and directivity, most of the CVR energy is concentrated on only a few virtual paths. The CVR vector It has a sparse property, which is called "virtual angle channel sparsity".

[0020] In the mmWave unlicensed mMTC network, each LU is assigned a pilot sequence. Assume that multiple attackers can impersonate LUs by sending the same pilot to the base station and launch pilot pollution attacks at will. At time slot t, the pilot signals of all active LUs and attackers received by the base station can be expressed as:

[0021]

[0022] in and Represents the set of active LUs and the set of LUs that suffer from pilot pollution attacks; represents the channel coefficient between the attacker attacking LUn′ and the base station; S = [s1, s2, ..., s N ] T represents the pilot matrix, W [t] represents the noise matrix, G [t] and G′ [t] Represents the channel matrix between active LU and attacked LU. Define two binary variables and To represent the state of each LU in time slot t. or 0 indicates that LUn is activated or inactivated in time slot t, Or 0 indicates whether LUn is attacked by PCA in time slot t. The channel matrices of the active LU and the attacked LU are: Since active LUs generate sporadic traffic, only a small number of LUs are active at each moment, that is, the matrix G [t] It has column sparse characteristics. And, based on the assumption that only a small number of attackers launch attacks at the same time at each moment, G′ [t] It also has the property of column sparsity, which is called "user activity state sparsity".

[0023] In order to increase the probability of successful attack, the pilot pollution attacker can monitor and impersonate the LU when the LU is not sending a signal, making it difficult for the base station to distinguish the attacker's identity. In addition, due to the sporadic transmission characteristics of mMTC, most LUs are inactive in each time slot, which makes it easier for the attacker to attack inactive LUs. Therefore, assuming that the attacker only attacks inactive LUs, that is, The LU state that is inactive and not attacked is defined as "idle". In each time slot, the LU state can only be "active", "attacked", or "idle".

[0024] A 3D transport model with time-dependent 2D sparsity

[0025] According to CVR, the received signal Y [t] Converted to the virtual angle domain, we get

[0026]

[0027] in represents the CVR matrix of the active LU in time slot t, represents the CVR matrix of the attacked LU in time slot t, represents the unified form of the CVR matrix, where the elements are defined as

[0028]

[0029] Considering the transmission structure of T consecutive time slots, the received signal of T time slots is Combined, we can get a three-dimensional receiving signal matrix Specifically, Contains T layers, where the tth layer represents the received signal Right now Similarly, the CVR matrix of T time slots Combined, we can get a three-dimensional CVR matrix in Therefore, the single-slot transmission model can be converted into the following three-dimensional transmission model:

[0030]

[0031] CVR Matrix It has both user active state sparsity and virtual angle signal sparsity, such as Figure 1 In addition, in actual mMTC networks, user activity may exhibit time-related characteristics, that is, active users will transmit continuously in multiple adjacent time slots with high probability. Therefore, it is assumed that the state of each active user remains unchanged for T time slots, that is, for active LUn The effective path position of the CVR is the same for each time slot

[0032]

[0033] Then, since the attacker can randomly launch a pilot pollution attack in any time slot and the attacker's position is variable, the state of the attacked LU has no time correlation, that is, for the attacked It changes with t.

[0034] Constructing 3D multi-measurement vector compressed sensing problem

[0035] Based on the constructed three-dimensional transmission model, joint channel estimation, active user detection, and pilot pollution attack detection can be converted into a continuous T time slot from the received signal Recover the sparse CVR matrix The sparse signal recovery problem can be expressed as the following three-dimensional multi-measurement vector compressed sensing problem:

[0036]

[0037] in Express The estimated results of is a high-dimensional sparse matrix and is difficult to reconstruct directly. The internal time-dependent two-dimensional sparse structure can help simplify the problem.

[0038] make represents the CVR at the mth virtual AoA within T time slots. It can be expressed as where α n =[α n，1 , α n，2 ,…,α n，T ] T ∈{0, 1} T×1 represents the LU state variable, represents the path gain variable at the virtual AoAm. Specifically, α n，t =1 means LUn is active or attacked (non-idle) in time slot t, α n，t =0 means user n is in idle state. n The probability distribution of can be expressed as:

[0039]

[0040] where λ n，t represents the probability that LUn is active or attacked in time slot t. Let the activation probability be The attack rate is Then there is The probability distribution of can be written as:

[0041]

[0042] where ω c，m，n =[ω c，m，n，1 ,ω c，m，n，2 ,…,ω c，m，n，T ] T ,σ 1，m，n，t Represents α n，t =1, the probability that the path of the virtual AoAm of the active LUn is a valid path, ω 2，m，n，t Represents α n，t =1, the probability that the path of the attacker attacking LUn at AoAm is a valid path. It represents the distribution of path gain at AoAm in different states of LUn.

[0043] Then we can get The probability distribution of

[0044]

[0045] After introducing auxiliary variables, we can further obtain the equivalent form of the original conditional probability problem:

[0046]

[0047] in Since the noise is assumed to follow a complex Gaussian distribution get Therefore, the original three-dimensional multi-measurement vector compressed sensing problem is equivalent to:

[0048]

[0049] In this way, the original problem has been transformed into estimating MN L-length sparse vectors problem.

[0050] 4 Parallel Multi-Measurement Vector Approximate Message Passing Algorithm Based on Expectation Maximization

[0051] To solve the constructed optimization problem, a parallel expectation-maximization vector approximate message passing with MMV (Parrallel-EM-VAMP-MMV) algorithm is designed. The algorithm jointly estimates a set of sparse vectors through M parallel message passing processes.

[0052] 4.1 Factor Graph

[0053] In order to illustrate the message passing process of the algorithm, the algorithm factor graph is shown as Figure 2 As shown in the figure, factor nodes and variable nodes are represented by squares and circles respectively. The variable nodes include α n and and Represents two variables of the same length T, represented by The definition of factor nodes is shown in Table 1. The information exchanged between different nodes represents the estimated probability distribution of the variable, also known as "belief". Figure 2 In the mth message passing process, all variables involved are located in the mth plane, which is called the mth frame. All M frames are superimposed together to form the vector The connection between M frames is composed of variable nodes and Constructed.

[0054] Table 1 Definition of factor nodes in factor graph

[0055]

[0056]

[0057] 4.2 Message Passing Process

[0058] The message passing process shown in the factor graph can be divided into three phases, called "into", "within" and "out". The within phase is implemented by M parallel message passing processes. The in and out phases are responsible for exchanging and updating information between each message passing process to improve the estimation accuracy. The three phases are executed by three nested iterative processes, namely "inner iteration", "middle iteration" and "outer iteration". The inner and middle iteration processes are used to execute the within phase of the algorithm, while the outer iteration processes are used to execute the into and out phases of the algorithm. Specifically, in the within phase, the vector AMP (VAMP) algorithm is executed to obtain the vector within the m frame. Since the prior parameters involved in the VAMP algorithm, called "local prior parameters", cannot be obtained in practice, the expectation-maximization (EM) algorithm is used to estimate the local prior parameters in the intermediate iterations. After the internal iteration and the intermediate iteration converge, the external iteration executes the into and out phases to update all M frames. Correspondingly, the EM algorithm is also used in the external iteration to learn the prior parameters related to the user state variables, called "global prior parameters", and the noise variance σ w .

[0059] Let i, j, k represent the iteration index of external iteration, intermediate iteration, and internal iteration respectively. and represents the messages of the incoming and outgoing frames in the i-th external iteration. The specific forms of the messages in each stage are derived below.

[0060] In the into phase, the message Represents the variable α n The estimated probability distribution is passed into each frame in the external iteration. By using the factor graph and product criterion, It can be expressed as:

[0061]

[0062] in Indicates that in the i-1th external iteration, frame m is passed to node α n News

[0063] In the within phase, based on the VAMP algorithm, the message is passed inside each frame through internal iteration to obtain the For frame m, the mth VAMP process will change the variable Separate into two identical variables and The estimation is performed through the denoising process and the linear minimum mean-squared error (LMMSE) process respectively. The specific algorithm process is shown in Table 2. In the denoising process, represents the denoising function in the standard AMP algorithm,<g′1(r,γ)> Denotes the derivative at r. In the LMMSE stage, Express The LMMSE estimation function. In the mth VAMP process, the variable node Prior parameters of If it cannot be obtained directly, the EM algorithm is used for estimation. Specifically, in the j+1th internal iteration of the i-th external iteration, the update method of the local prior parameters is:

[0064]

[0065] in

[0066]

[0067] Table 2 VAMP algorithm

[0068]

[0069]

[0070] In the out phase, the message It is transmitted from frame m and is expressed as Used to update the pair in external iteration The probability distribution of . Specifically, It can be expressed as:

[0071]

[0072] in

[0073]

[0074] In the i-th outer iteration, the reconstructed CVR matrix can be obtained by Get, among them In order to estimate the global prior parameters and noise variance The EM algorithm is also used for estimation:

[0075]

[0076] in Finally, when the outer iteration converges, the reconstructed CVR matrix is ​​obtained Where I represents the total number of external iterations. Based on the above description, the proposed Parallel EM-VAMP-MMV algorithm process is summarized as shown in Table 3.

[0077] Table 3 Parallel EM-VAMP-MMV algorithm

[0078]

[0079]

[0080] 4.3 Joint Channel Estimation, Active User Detection, and Pilot Pollution Attack Detection

[0081] Combined with the results of Parallel EM-VAMP-MMV output The following steps are used to implement joint channel estimation, active user detection, and pilot pollution attack detection.

[0082] Generate a candidate set. Define the log-likelihood ratio of the posterior LU state in time slot t as:

[0083]

[0084] represents the ratio of the probability that LUn is in the active or attacked state to the probability that it is in the idle state at time slot t, where Since only the states of active LUs show time correlation, we can generate a set of candidate active LUs Λ a and the candidate attacked LU set

[0085]

[0086] LU state identification. Since transmitters at different locations will generate different virtual AoAs in the virtual angle domain, combined with the time correlation of active LUs, the detected virtual AoAs in different time slots can be compared, i.e. To identify the active state of LU. Then the estimated LU activity indicator variable can be obtained by Determine, where n∈Λa , Then, the estimated active LU set in each T time slot can be obtained. In order to detect the attacked LU, Denotes the estimated LU attacked state indicator variable, and the following pilot pollution attack detection criterion is designed:

[0087]

[0088] in Represents the CVR energy threshold at the virtual AoA. Further, the estimated attacked LU set can be obtained

[0089] Based on the estimated active LU set The CVR matrix can be Obtain, among which express The containing set The submatrix of the columns indexed in ; finally, the channel coefficients of the active LU are

[0090] The present invention can also provide a large-scale MTC unlicensed millimeter wave uplink multiple-input multiple-output system to combat PCA, and perform joint channel estimation and active user detection based on the above method.

[0091] The present invention may also provide a computer device, including a processor and a memory, the memory being used to store a computer executable program, the processor reading the computer executable program from the memory and executing it, and the processor being able to implement the large-scale MTC unlicensed channel estimation and user detection method against PCA described in the present invention when executing the computer executable program.

[0092] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored therein. When the computer program is executed by a processor, the method for large-scale MTC unlicensed channel estimation and user detection against PCA described in the present invention can be implemented.

[0093] The computer device may be a laptop computer, a desktop computer or a workstation.

[0094] The processor may be a central processing unit (CPU), a digital signal processor (DSP), an application specific integrated circuit (ASIC), or an off-the-shelf field programmable gate array (FPGA).

[0095] The memory of the present invention may be an internal storage unit of a laptop computer, desktop computer or workstation, such as a memory or a hard disk; or an external storage unit, such as a mobile hard disk or a flash memory card.

[0096] Computer-readable storage media may include computer storage media and communication media. Computer storage media include volatile and non-volatile, removable and non-removable media implemented by any method or technology for storing information such as computer-readable instructions, data structures, program modules or other data. Computer-readable storage media may include: read-only memory (ROM), random access memory (RAM), solid-state drives (SSD) or optical disks, etc. Among them, random access memory may include resistance random access memory (ReRAM) and dynamic random access memory (DRAM).

[0097] Numerical simulation and result analysis

[0098] 1) Simulation parameter settings

[0099] The present invention evaluates the system performance of the proposed solution through simulation, and the system parameter settings are shown in Table 4:

[0100] Table 4 Simulation parameters

[0101]

[0102]

[0103] To measure the performance of the proposed scheme, comparative experiments were conducted with the following compressed sensing algorithms: EM-AMP, EM-VAMP, and Oracle-LS. EM-AMP and EM-VAMP independently estimate the CVR at M virtual AoAs and estimate the prior parameters by performing EM updates in each AMP and VAMP iteration. Oracle-LS provides an upper bound on the performance of the other algorithms, assuming that the true active LU set and the attacked LU set are perfectly known by the base station. Furthermore, the performance of the proposed algorithm is presented without considering the transmission model of the pilot contamination attacker, as well as without external iterations.

[0104] Figure 3 、 Figure 4 、 Figure 5 The performance simulation results of the Parallel EM-VAMP-MMV scheme of the present invention and other comparison schemes are presented, showing the legitimate user activity detection error rate, the normalized mean square error of legitimate user channel estimation, and the accuracy of pilot pollution attack detection as the attack rate changes. Figure 3 、 Figure 4 、 Figure 5 As shown, all algorithms degrade with increasing attack rates. However, the proposed Parallel EM-VAMP-MMV scheme exhibits significant performance advantages over the comparative schemes EM-AMP and EM-VAMP. This is because EM-AMP and EM-VAMP independently estimate the CVR coefficients at each virtual AoA, without leveraging the two-dimensional sparsity of the LU, namely the sparsity of user activity states and the sparsity of virtual angular channels. In contrast, the proposed scheme jointly estimates the CVR coefficients for all M virtual AoAs through three nested loop iterations. Inner and intermediate iterations obtain local CVR estimates at each virtual AoA. Finally, an outer iteration utilizes two-dimensional sparsity to jointly update and correct all estimated results, thereby improving the system's detection performance. The proposed scheme still achieves a certain performance gain compared to the comparative scheme without external iterations, i.e., independently estimating CVRs through inner and intermediate iterations. This gain also demonstrates the benefits of utilizing two-dimensional sparsity to the system. Furthermore, the proposed scheme's performance is evaluated without considering the attacker's transmission model, assuming no attackers exist in the network. In this case, the algorithm's performance significantly degrades, indicating that pilot pollution attacks can severely degrade the receiver's performance in detecting legitimate user activity and estimating the channel. However, by jointly establishing a three-dimensional transmission model of the LU and the attacker and integrating it into the designed Parrallel EM-VAMP-MMV algorithm, the proposed solution effectively counteracts pilot pollution attacks, thereby improving the accuracy of channel estimation and activity detection.

[0105] In summary, the present invention proposes a large-scale, unlicensed MTC channel estimation and user detection method to combat PCA. This method leverages the sparsity of the millimeter-wave channels between legitimate users and attackers in the virtual angle and user domains, as well as the temporal correlation of legitimate users' activation states, achieving superior system performance. Simulation results demonstrate that, compared with existing multi-user detection schemes, the present invention's joint channel estimation and active user detection method to combat pilot contamination attacks effectively reduces the user activity detection error rate and the mean squared error of channel estimation, improves the accuracy of pilot contamination attack detection, and achieves superior system performance.

[0106] The above content is a detailed description of the present invention, and it cannot be considered that the present invention is limited to this. For ordinary technicians in the technical field to which the present invention belongs, they can make several simple deductions or substitutions without departing from the concept of the present invention, which should be regarded as belonging to the scope of protection of the present invention determined by the submitted claims.

Claims

1. A large-scale MTC unlicensed channel estimation and user detection method against PCA, characterized in that: The following steps are involved: Based on the unlicensed mMTC network, a millimeter wave uplink multi-input multi-output system model is constructed, considering the continuous The transmission structure of time slots is The received signal of the time slot Combined, we get a three-dimensional receiving signal matrix ,Will CVR matrix of time slots Combined, we get a three-dimensional CVR matrix, and according to the three-dimensional received signal matrix and the three-dimensional CVR matrix Constructing a 3D transport model with time-dependent 2D sparsity; Based on the constructed three-dimensional transmission model, joint channel estimation, active user detection, and pilot pollution attack detection can be converted into continuous In the time slot, the received signal matrix Recover the sparse CVR matrix The sparse signal recovery problem is formulated as the following three-dimensional multi-measurement vector compressed sensing problem: in, Express The estimated results of The internal time-dependent two-dimensional sparse structure simplifies the problem and converts the three-dimensional multi-measurement vector compressed sensing problem into an estimation problem. indivual Long sparse vectors The problem, , is the antenna index set of the base station, Index set for legitimate users; solve the estimation problem by using parallel multi-measurement vector approximate message passing algorithm based on expectation maximization indivual Long sparse vectors The problem of 3D MMV compressed sensing is solved , and get the estimated active legal user LU set , pilot pollution attack PCA attacker set , and the channel matrix of the legal user LU ; The system model includes a preparatory Base stations with antennas, The attackers attack a single-antenna legitimate user LU and a variable number of pilot pollution attacks PCA attackers. Due to the sporadic transmission characteristics of mMTC, no more than 10% of the legitimate user LUs are active at any given moment. Each legitimate user LU is fixed relative to the base station, while the attacker's position moves within the cell. The legal user LU index set is expressed as , the antenna index set of the base station is expressed as ; Each legal user LU is assigned a unique pilot sequence , the length is Consider the scenario where a legitimate user LU and a pilot pollution attack PCA attacker coexist. The attacker randomly sends the same pilot sequence as a legitimate user LU to pollute the pilot channel. The pilot signals received from all active legitimate users LU and attackers are expressed as ,in and They represent the channel matrices of active legitimate users LU and attackers respectively, represents the pilot matrix, represents noise; construct a three-dimensional transmission model with time-dependent two-dimensional sparsity, expressed as , Indicates time slot The pilot received signal in the virtual angle domain, represents the received signal in the virtual angle domain, represents the normalized discrete Fourier transform matrix, represents the conjugate transpose of the matrix, Indicates time slot The CVR matrix, and represents the CVR matrix of active legitimate users LU and attackers, Indicates time slot Noise in the virtual angle domain; the three-dimensional transmission model jointly describes the sparsity of user activity status and millimeter wave virtual channels, as well as the temporal correlation of the activity status of legitimate users LU in multiple time slots.

2. The PCA-resistant large-scale MTC unlicensed channel estimation and user detection method according to claim 1, characterized in that: Signal Model In the channel matrix, , , where the two-dimensional indicator variable and Represents LU Active state and attack state, and Represents LU The channel coefficient vectors in active and attack states respectively; the geometric channel model is adopted, that is, ,in represents the average path loss, represents the number of scatterers in the environment, represents the fading coefficient of the corresponding path, Indicates the physical AoA of the path, Represents the corresponding vector of the antenna array; CVR is introduced to describe the characteristics of the millimeter wave channel and establish the channel coefficient Its CVR Conversion relationship , express No. element, describing the The channel gain of the virtual path, Indicates the virtual AoA corresponding to the virtual path.

3. The PCA-resistant large-scale MTC unlicensed channel estimation and user detection method according to claim 1, characterized in that: The three-dimensional transmission model with time-dependent two-dimensional sparsity In the example, the legal user LU is The time correlation within a time slot is ,in express The set of non-zero element indices of , Representative time slot Internal LU The effective virtual AoA set, Represents LU In the time slot The CVR vector.

4. The large-scale MTC unlicensed channel estimation and user detection method against PCA according to claim 1, characterized in that: use The internal time-dependent two-dimensional sparse structure simplifies the problem and converts the three-dimensional multi-measurement vector compressed sensing problem into an estimation problem. indivual Long sparse vectors ( ) include: Indicates Virtual AoA within the time slot The CVR vector at the location is introduced by introducing the legal user LU state variable and the variables describing the virtual path gain ,get , the original three-dimensional problem is equivalently expressed as a two-dimensional problem: in, , as well as ; The details are as follows: make express In the time slot CVR at a virtual AoA, Expressed as ,in represents the LU state variable, Indicates virtual AoA The path gain variable at Represents LU In the time slot Active or under attack, Represents a user In idle state, The probability distribution of is: in, Represents LU In the time slot The probability of being active or attacked under the current state, let the activation probability be , the attack rate is , then , The probability distribution of is: in , express Active LU Virtual AoA The probability that the path is a valid path, express When attacking LU The attacker is in AoA The probability that the path at is a valid path, Represents LU AoA in different states The distribution of path gain at The probability distribution of is: 。 5. The large-scale MTC unlicensed channel estimation and user detection method against PCA according to claim 1, characterized in that: Parallel multi-measurement vector approximate message passing algorithm based on expectation maximization to solve estimation indivual Long sparse vectors ( ) issues, including: A parallel message passing process to jointly estimate the sparse vector ,Each message passing process involves message passing and updating between multiple ,variable nodes and factor nodes. The messages between nodes represent the ,estimation results of the probability distribution of the relevant ,variables; The algorithm steps are divided into three stages: "into", "within" and "out". In the "into" stage, the message carries the user status indicator variable. The estimated results of the probability distribution and the Passed to In the message passing process, in the within phase, the message is iteratively updated within each message passing process to obtain the CVR vector. and variables The estimated result of the probability distribution; in the out phase, the message is The message passing process is passed back to the variable node to achieve Correction of probability distribution estimation results; repeat the above three stages until convergence, and output the estimation results of CVR ; Based on the output of the algorithm , the joint channel estimation, active user detection and PCA detection are achieved through the following steps: First, the candidate user set is generated and defined in the time slot The log-likelihood ratio of the inner posterior LU state represents LU In the time slot The ratio of the probability of being in an active or attacked state to the probability of being in an idle state. Since only the state of active LUs shows time correlation, a candidate active LU set is generated. and the candidate attacked LU set ; LU state identification: Transmitters at different locations will generate different virtual AoAs in the virtual angle domain. Combined with the time correlation of active LUs, the detected virtual AoAs in different time slots are compared, i.e. Identify the active state of LU, then the estimated LU activity indicator variable is given by OK, among them , Represents the ratio threshold; get each The estimated active LU set in a time slot ;make It represents the estimated LU attacked state indicator variable and the pilot pollution attack detection criterion: in , Represents the CVR energy threshold at the virtual AoA; get the estimated attacked LU set ; Based on the estimated active LU set , the CVR matrix is ​​obtained by Obtain, among which express The containing set The submatrix of the columns indexed in ; finally, the channel coefficients of the active LU are .

6. A large-scale MTC unlicensed millimeter wave uplink multiple-input multiple-output system to combat PCA, characterized by: The system includes Base stations with antennas, A single-antenna legitimate user LU, and multiple attackers launching pilot pollution attacks PCA, perform joint channel estimation and active user detection based on the large-scale MTC unlicensed channel estimation and user detection method against PCA according to any one of claims 1-5.

7. A computer device, characterized in that: The invention comprises a processor and a memory, the memory is used to store a computer executable program, the processor reads the computer executable program from the memory and executes it, and when the processor executes the computer executable program, it can implement the large-scale MTC unlicensed channel estimation and user detection method against PCA as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that A computer program is stored in a computer-readable storage medium. When the computer program is executed by a processor, the method for large-scale MTC unlicensed channel estimation and user detection against PCA according to any one of claims 1 to 5 can be implemented.

Images