Massive MIMO communication system physical layer authentication method

Abstract

The invention provides a physical layer authentication method for a massive MIMO communication system. The method comprises the following steps: an expected receiver determines a test statistic Lx used for evaluating channel state information difference between adjacent time intervals through a log-likelihood ratio; and when the sender is the legal user equipment and assumed H0, when the sender is the attacker and assumed H1, and when Lx is smaller than or equal to a threshold value delta, the expected receiver accepts the assumed H0 and judges that the sender is the legal user equipment, otherwise, the expected receiver accepts the assumed H1 and judges that the sender is the attacker. According to the authentication scheme, modeling is carried out on a channel in a massive MIMO antenna communication scene, and the influence of antenna coupling, antenna space correlation and the like caused by antenna element spacing reduction and antenna data increase on the physical layer authentication performance is measured. And the PCA and the SVD are used for carrying out spatial domain and frequency domain decorrelation on the high-dimensional channel information, so that the storage space occupied by the channel information at a base station end can be reduced, and the authentication overhead is greatly reduced under the condition that the same authentication performance is ensured.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and more specifically to a physical layer authentication method for a massive MIMO communication system. Background Technology

[0002] The fundamental goal of wireless communication systems is to adapt to the significant increase in mobile data traffic, improve spectrum efficiency and quality of service, and achieve reliable communication. To meet these demands, advanced disruptive technologies are required. Massive Multiple-Input Multiple-Output (MIMO) is one of the key implementation technologies for wireless communication systems. Specifically, massive MIMO systems utilize large-scale antenna arrays on base stations to support the ever-increasing demand for data traffic. Furthermore, massive MIMO systems have the ability to significantly reduce energy consumption and hardware power consumption. However, the broadcast nature of wireless communication in massive MIMO systems makes them sensitive to spoofing attacks, where an unknown transmitter impersonates another legitimate transmitter to inject malicious information. For secure communication, the identity of unknown transmitters needs to be verified to reject malicious impersonators.

[0003] Existing physical layer authentication schemes typically use channel state information as the authentication feature. However, in massive MIMO systems, base stations often need to deploy dozens or even hundreds of antenna elements within a limited space, leading to spatial correlation and coupling between antenna elements. Furthermore, the increased number of antennas significantly increases the dimensionality of the channel vector. Due to the spatial correlation of antenna elements, the channel information vector elements also exhibit correlation. Storing all these channel vectors in the content space in the authentication model would undoubtedly be a significant overhead. Simultaneously, in massive MIMO systems, the expensive ultralinear 50-watt amplifiers used in traditional systems are replaced by hundreds of low-cost amplifiers with output power in the milliwatt range. This reduces the limitations on the accuracy and linearity of each individual amplifier and RF chain, resulting in high cross-polarization isolation not always being achievable. Therefore, massive MIMO systems need to consider the antenna coupling and antenna spatial correlation issues caused by the reduced antenna element spacing and increased antenna data.

[0004] Existing methods do not consider the impact of these characteristics in massive MIMO systems on the performance of physical layer authentication schemes. Given the increasing prevalence of massive MIMO systems, there is an urgent need to assess the impact of these factors on physical layer authentication performance and to provide an authentication method that reduces authentication overhead. Summary of the Invention

[0005] This invention addresses the problems of existing channel state information failing to characterize the effects of antenna correlation coupling and the high overhead of existing physical layer authentication schemes. It provides a physical layer authentication method for massive MIMO communication systems. When modeling the channel, the impact of antenna cross-polarization isolation, antenna coupling, antenna spatial correlation, channel depolarization effects, and channel drop fading on information transmission must be considered. Furthermore, since Principal Component Analysis (PCA) and Singular Value Decomposition (SVD) have significant advantages in information extraction and matrix dimensionality reduction, PCA and SVD are used to decorrelate the channel matrix, thereby reducing the matrix dimensionality.

[0006] This invention provides a physical layer authentication method for a massive MIMO communication system. The uplink massive MIMO system includes a legitimate user equipment, a desired receiver, and an attacker. The desired receiver communicates with the legitimate user equipment through massive MIMO. The desired receiver is a base station with N antennas. Expected recipient in the first t -1 receives the authenticated channel matrix from the legitimate user equipment. And store; Expected recipient in the first t The channel matrix of the sender to be determined is received at each time step. ; The time interval between time t-1 and time t is less than the channel correlation time; The receiver is expected to determine the test statistic used to evaluate the difference in channel state information between adjacent time intervals using the log-likelihood ratio. , ; When the sender is a legitimate user device, assuming When the sender is an attacker, assuming ,when Less than or equal to the threshold At that time, it is expected that the recipient will accept the assumption. The sender is determined to be a legitimate user equipment; otherwise, the receiver is expected to accept the assumption. The sender was determined to be an attacker. .

[0007] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, ; in, This represents the time-varying correlation of the channel, obtained from the antenna base station; T To change the order; To use method X The reduced-dimensional channel matrix, To use method X The reduced-dimensional channel matrix; Used in the data feature library at time t-1 X Theoretical channel information of legitimate user equipment under dimensionality reduction method variance , A Representing legitimate users, theoretical channel information for matrix, The number of subcarriers, H Follows a mean of 0 and a variance of The complex Gaussian distribution; After dimensionality reduction using method X , When the receiver receives legitimate user equipment , Channel estimation error caused by receiver noise variance ; Channel matrix for: , , .

[0008] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, When using PCA for dimensionality reduction, the receiver is expected to use the channel state information covariance matrix... Determined KLT matrix This makes the original channel state information from Each relevant component is compressed into M The uncorrelated components yield the channel matrix after PCA dimensionality reduction. , M The dimension of the compressed channel matrix is ​​( ).

[0009] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, ; in, To decompose from eigenvalues Selected from Let D be a diagonal matrix, and the diagonal elements of D are the covariance matrix. The eigenvalues, the number of non-zero eigenvalues ​​is , The column vector is eigenvectors.

[0010] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, The recipient is expected to select Principal components generate sparse vectors to indicate Then use a binary matrix from Choose the most important M One piece of information: .

[0011] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, For a binary matrix, select S and extract the channel information corresponding to the first M eigenvalues, arranged in descending order of eigenvalues. ; The covariance matrix is: ; The theoretical channel matrix after PCA dimensionality reduction The covariance matrix is: ; for: ; When using PCA for dimensionality reduction, the compression ratio of the channel information is: .

[0012] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred method, requires the receiver to perform singular value decomposition on the channel matrix to obtain a sparse representation matrix when using SVD dimensionality reduction, thus obtaining the SVD-reduced channel matrix. ; Channel matrix The singular value decomposition is ,in, It is a diagonal matrix, and the values ​​on the diagonal are singular values. The number of non-zero singular values ​​is . , and These are the left singular matrix and the right singular matrix, respectively; The SVD sparse representation matrix is: ; Expected recipient to use Select Principal components generate sparse vectors. to indicate : ; Then, the receiver is expected to use a binary matrix. from Choose the most important M One piece of information: ; The covariance matrix is: ; The covariance matrix is: ; The covariance matrix of the channel estimation error after SVD dimensionality reduction is: .

[0013] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, The compression ratio of the channel information reduced by the SVD method is: .

[0014] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, During the initial communication, the receiver is expected to use the upper-layer authentication scheme to determine whether the sender of the signal is a legitimate user equipment. If so, the receiver will estimate the channel frequency response of the legitimate communication link and store it in the feature database for subsequent authentication. If not, an alarm will be triggered.

[0015] The physical layer authentication method for a massive MIMO communication system described in this invention, as a preferred embodiment, Legitimate user equipment deploys a single dual-polarized antenna, and antenna base station deployment... One dual-polarized antenna.

[0016] This invention is a physical layer authentication technology based on compressed sensing for large-scale antenna arrays.

[0017] The present invention has the following advantages: The authentication scheme of this invention models the channel in massive MIMO antenna communication scenarios, measuring the impact of reduced antenna element spacing, increased antenna data leading to antenna coupling, and antenna spatial correlation on physical layer authentication performance. Furthermore, it uses PCA and SVD to decorrelate high-dimensional channel information in both the spatial and frequency domains, reducing the storage space occupied by channel information at the base station and significantly lowering authentication overhead while maintaining the same authentication performance. Attached Figure Description

[0018] Figure 1 A schematic diagram of a massive MIMO communication system, illustrating a physical layer authentication method for a massive MIMO communication system. Figure 2 This is a flowchart illustrating the authentication process of a physical layer authentication method for a massive MIMO communication system. Figure 3 A comparison of authentication performance under dimensionality reduction using PCA and SVD for a physical layer authentication method in a massive MIMO communication system. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Example 1

[0020] like Figures 1-3 As shown, a physical layer authentication method for a massive MIMO communication system considers a legitimate user equipment (UE) with a single antenna, an N-antenna base station (BS) (the intended receiver), and an attacker's uplink massive MIMO system, such as... Figure 1 As shown in the diagram. Assume both the UE and the BS use co-located dual-polarized antennas, a common feature in massive MIMO systems. Assume a legitimate UE, referred to as Alice, communicates with the intended receiver, Bob, while an attacker, referred to as Eve, may eavesdrop on the radio channel and attempt to impersonate Alice to inject harmful information into the network. Assume Eve possesses some reusable, publicly known information, such as training sequences, pilot sequences, and the polarization type of the transmit antenna.

[0021] Our security objective is for Bob to authenticate the sender's identity in the presence of Eve. Assume Bob receives signals at time intervals t-1 and t, where the time intervals are... Let be the channel coherence time. Bob uses the upper-layer authentication scheme to authenticate the signal sender in the (t-1)th time interval, determining that it originates from Alice. Then, Bob stores the channel state information at this time. Since channel state information from the same transmitter is highly correlated within the coherence time, while channel state information from different transmitters is independent, in the t-th time interval, Bob identifies the unknown sender by comparing the similarity of the channel state information in the t-th and (t-1)th time intervals.

[0022] This embodiment uses, for example Figure 1 The communication system shown in the diagram involves a legitimate party communicating with a base station that has N antennas, while an attacker attempts to impersonate a legitimate communicator and communicate with the base station.

[0023] In Massive MIMO systems, the use of numerous antennas and multiple carriers significantly increases the dimensionality of the channel state information matrix obtained by the BS (Browser Controller), and the matrix elements exhibit substantial correlation. On one hand, multi-carrier communication causes correlation between channel information from different subcarriers. On the other hand, the use of numerous antennas causes correlation between channel information from different antenna elements. Directly storing the channel state information matrix obtained by the BS in memory for authentication would undoubtedly lead to high memory consumption and significant information redundancy. To reduce memory consumption and improve authentication efficiency, it is necessary to reduce the dimensionality of the high-dimensional channel matrix and decorrelate its elements. Here, we introduce two typical channel matrix dimensionality reduction methods: PCA (Proportional Constraint A) and SVD (Sparse Channel Decorrelation). PCA selects a sparse matrix based on statistical channel information (the covariance matrix of channel state information) to decorrelate the channel information, while SVD selects a sparse matrix based on instantaneous channel information (channel state information) to decorrelate the channel information.

[0024] Assume Bob has already estimated the channel matrix using the uplink pilot sequence. It can be represented as: (1) in, The channel information is derived theoretically, and it is... matrix, Indicates the number of subcarriers. N Indicates the number of dual-polarized antennas. H It follows a mean of 0 and a variance of . The complex Gaussian distribution, i.e. ; The channel estimation error caused by receiver noise has a mean of 0 and a variance of . The complex Gaussian distribution, i.e. ;therefore ,in .

[0025] From containing Of the relevant components, PCA selects through signal transformation. M ( We use 1) uncorrelated principal components, i.e., dimensionality reduction of the matrix with correlated components. For the transformation, we employ a matrix composed of the channel state information covariance matrix. The determined KLT matrix compresses the original channel state information into uncorrelated components. Specifically, the KLT matrix is: (2) in, From eigenvalue decomposition Selected from Here, D is a diagonal matrix, and its diagonal elements are... covariance matrix The eigenvalues, the number of non-zero eigenvalues ​​is , The column vector is eigenvectors.

[0026] use BS can be selected The principal components generate a sparse vector. to indicate ,Right now: (3) Then, using a binary matrix BS can be obtained from Choose the most important M One piece of information, namely: (4) because Each element in S is selected at most once, therefore each row or column of S contains at most one "1". In this paper, the first M elements of the binary matrix S are selected. The channel information corresponds to ) feature values, which are assumed to be arranged in descending order. The compression ratio of the channel information is then: (5) in M The dimensions of the compressed channel matrix. This represents the dimension of the channel matrix before compression.

[0027] Channel matrix after dimensionality reduction The covariance matrix is: (6) Theoretical channel matrix after dimensionality reduction The covariance matrix is: (7) The covariance matrix of the channel estimation error after dimensionality reduction is: (8) Unlike PCA, SVD directly performs singular value decomposition on the channel matrix to obtain a sparse representation matrix. Specifically, the channel matrix... The singular value decomposition is ,in It is a diagonal matrix, and the values ​​on the diagonal are the singular values. The number of non-zero singular values ​​is . , and These are the left singular matrix and the right singular matrix, respectively. The SVD sparse representation matrix is: (9) use BS can be selected The principal components generate a sparse vector to represent them. ,Right now: (10) Then, using a binary matrix BS can be obtained from Choose the most important M One piece of information, namely: (11) In this embodiment, the binary matrix S is selected from the first... M ( We have channel information corresponding to ) singular values, assumed to be arranged in descending order. The compression ratio of the channel information is then: (12) Channel matrix after dimensionality reduction The covariance matrix is: (13) Theoretical channel matrix after dimensionality reduction The covariance matrix is: (14) The covariance matrix of the channel estimation error after dimensionality reduction is: (15) The test statistics after dimensionality reduction of the channel matrix are similar to those before dimensionality reduction. The difference is that after dimensionality reduction, the channel information of different antennas and frequencies is independent of each other, and the dimension of the channel matrix is ​​much smaller than the dimension of the matrix before dimensionality reduction.

[0028] For massive MIMO systems, the movement of the UE and the scattering surface causes correlated temporal variations in the radio channel. Assuming each channel gain is constant within a frame, independent and continuous variations occur from one frame to another, where the frame duration is... ,and , Let be the channel correlation time. A first-order Gaussian Markov process is used to approximate the time-varying nature of the channel. Let Let Gaussian noise have a mean of 0 and a variance of 1 be represented. Then, the theoretical channel information for the t-th time interval is: (16) in and These represent the (t-1)th and tth time intervals, respectively. . This represents the time-varying correlation of the channel, which can be obtained at BS.

[0029] Bob used the log-likelihood ratio to determine the test statistic used to evaluate the differences in channel state information between adjacent time intervals. It can be represented as: (17) in, , The sender's channel matrix to be judged at time t is the dimension-reduced matrix using method X. The channel matrix of the legitimate user equipment stored by the expected receiver at time t-1 after dimensionality reduction using the X method; Indicates use X The variance of Alice's channel information H under the dimensionality reduction method. Indicates use X Alice under dimensionality reduction method .

[0030] Following typical statistical decision-making methods, Bob uses a hypothesis testing model to determine whether the sender is still Alice at time interval t. Hypothesis The sender is Alice, a legitimate sender; assuming... The sender is the attacker, Eve. When Less than or equal to the threshold At that time, Bob accepted the hypothesis. Otherwise, Bob accepts the hypothesis. .

[0031] (18) The detection probabilities are given below. With compression ratio We can see the relationship. along with It increases as the value increases. This is because... The larger the value, the more channel elements are used for authentication, and the more channel information is included. We can also see that when... It can be achieved in time. ,at this time In other words, Bob only needs to store 41 channel elements to achieve a detection probability of 98.7%, compared to storing... Each channel element significantly saves memory space. Furthermore, in the same... Under these conditions, using PCA for channel matrix dimensionality reduction yields higher efficiency compared to using SVD. Therefore, in practice, we can choose PCA as the compression method for the channel matrix.

[0032] To explain the certification scheme in detail, we Figure 2 The certification process of the proposed scheme is illustrated in the figure. The relationship between the detection probability and the compression ratio in this embodiment is as follows: Figure 3 As shown.

[0033] During the initial communication, Bob uses the upper-layer encryption mechanism to determine if the sender of the signal is Alice. If Alice is identified as the sender, the channel frequency response of the legitimate communication link is estimated and stored in the feature database for subsequent authentication. If Eve is the sender, an alarm is triggered.

[0034] In subsequent communications, Bob uses the physical layer authentication mechanism to determine whether the sender of the signal is still Alice. Bob estimates the channel response matrix using the least squares method based on the received signal, and then uses PCA or SVD to decorrelate the high-dimensional channel response matrix, reducing the memory required for Bob to store the channel response matrix. Then, Bob extracts the channel response matrix of the previous time interval from the feature database and calculates the test statistic based on equation (17). Finally, the sender of the signal is determined based on equation (18). If If so, the sender is assumed to be Alice, and the channel polarization response in the feature database is replaced with the channel polarization response. If If so, it is assumed that the sender is Eve, and an alarm is triggered.

[0035] In massive MIMO systems, the deployment of numerous antennas within a limited space leads to reduced antenna spacing, resulting in mutual coupling effects between antenna elements. This means that the radiated signal from one antenna can be affected by electromagnetic coupling from other antennas. Furthermore, spatial correlation exists between antenna elements, causing channel information to correlate across different elements. Due to the use of numerous low-cost antennas and RF front-ends, the cross-polarization isolation of dual-polarized antennas is no longer infinite; horizontally polarized antennas may exhibit coupling signals from vertically polarized antennas. Additionally, depolarization effects, large-scale fading, and small-scale fading exist in the wireless channel. Therefore, channel modeling for massive MIMO systems requires consideration of antenna coupling, antenna spatial correlation, antenna cross-polarization isolation, channel depolarization effects, and both large-scale and small-scale fading.

[0036] We consider deploying a single dual-polarized antenna for the UE and a BS deployment. One dual-polarized antenna, with a communication frequency range of [missing information]. It contains There are subcarriers, and the bandwidth of each subcarrier is . Since the UE deploys a single dual-polarized antenna, its antenna coupling and spatial correlation effects are not considered. Here, we only consider the antenna coupling and spatial correlation effects of the large-scale antenna array at the BS. Therefore, the channel vector between the UE and the BS is: (19) in This means stacking all the columns of the matrix in order into a column vector. Indicates the first The channel information at each subcarrier can be represented as: (20) in , These represent the channel information transmitted by the vertically polarized antenna for vertically polarized reception and the channel information transmitted by the vertically polarized antenna for horizontally polarized reception, respectively. , These represent the channel information transmitted by the horizontally polarized antenna and received by the horizontally polarized antenna, respectively, and the channel information transmitted by the horizontally polarized antenna and received by the vertically polarized antenna. . express The identity matrix. This represents the cross-polarization isolation matrix for the transmitting and receiving dual-polarized antennas. To receive the mutual interference generated by vertically polarized waves, To mitigate the mutual interference generated by horizontally polarized waves, for ease of analysis, we assume that the mutual interference generated by horizontally and vertically polarized waves is the same, i.e. . Here is the cross-polarization discrimination matrix of the channel, where Indicates cross-polarization discrimination. This represents the power imbalance coefficient. This is the polarization correlation matrix of the channel. , Let X be the path loss coefficient. Various measurements show that in non-line-of-sight scenarios, the received XPC is close to 0, therefore we assume... In other words, the fading experienced by vertical polarization and horizontal polarization through the channel is independent of each other. . Its variance is It follows a normal distribution. This represents the coupling matrix between the vertically polarized antennas at the BS end. This represents the coupling matrix between the horizontally polarized antennas at the BS end. Since the horizontally polarized antennas and the vertically polarized antennas have the same antenna architecture, therefore... In equation (20), we use . This represents the channel correlation matrix when the BS (Browser / Server) uses a vertically polarized antenna and the UE (User Equipment) uses a vertically polarized antenna. This represents the channel correlation matrix when the BS (Browser / Base Station) uses a horizontally polarized antenna and the UE (User Equipment) uses a vertically polarized antenna. This represents the channel correlation matrix when the BS (Browser / Server) uses a vertically polarized antenna and the UE (User Equipment) uses a horizontally polarized antenna. This represents the channel correlation matrix when both the BS and UE use horizontally polarized antennas. Since the horizontal and vertical polarized antennas at the BS are co-located, they are subject to the same channel environment. . This represents the channel gain, i.e., the large-scale fading factor. Therefore, the channel vector... It follows a mean of 0 and a variance of . The normal distribution, i.e. ,in This represents the transpose of the matrix. Since the fading of V-polarized waves and H-polarized waves in the channel is independent of each other, therefore... , it is The anti-diagonal element.

[0037] (twenty one) (twenty two) in, Let be the frequency correlation matrix of the channel, the first... The and the first The correlation coefficient between each subcarrier is , Indicates the arrival time of the multipath.

[0038] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A physical layer authentication method for a massive MIMO communication system, characterized in that: The uplink massive MIMO system includes a legal user equipment, an expected receiver and an attacker, the expected receiver communicates with the legal user equipment through massive MIMO, and the expected receiver is an antenna base station with a number of N; The desired receiver receives at a t -1 time a channel matrix from the legitimate user equipment that is authenticated and stores; The desired receiver receives the channel matrix of the sending party to be judged at the first t time point ; A time interval between the t-1th moment and the tth moment is less than a channel correlation time; The desired receiver determines a test statistic for assessing the difference in channel state information for adjacent time intervals through a log-likelihood ratio , ; When the sender is the legitimate user device, assume When the sender is the attacker, assume When is less than or equal to a threshold , the intended receiver assumes that the sender is the legitimate user device, otherwise, the intended receiver assumes that the sender is the attacker: 。 2. The physical layer authentication method of the massive MIMO communication system according to claim 1, characterized in that: ； wherein, is a time-varying correlation of the channel, obtained from the antenna base station; T is a rank conversion; for using the X method on the reduced dimension channel matrix, for using the X method on the reduced dimension channel matrix; theoretical channel information of the legitimate user equipment at time t-1 stored in the data feature library X dimension reduction method the variance of , A representing the legitimate user, the theoretical channel information is a matrix, is the number of subcarriers, H obeys a complex Gaussian distribution with mean 0 and variance ; after dimension reduction using the X method , is the theoretical channel information of the legitimate user equipment received by the receiver , is the variance of the channel estimation error generated by the receiver noise ; ; Channel matrix For: , , . 3.The physical layer authentication method for a massive MIMO communication system according to claim 2, wherein: When using PCA dimension reduction, the desired receiver compresses the original channel state information into a lower dimension channel matrix according to the KLT matrix determined by the channel state information covariance matrix Rxx , M M ​​​​​ 4. The physical layer authentication method of the massive MIMO communication system according to claim 3, characterized in that: ； wherein is the eigenvalue decomposition of is the eigenvalue decomposition of , D is a diagonal matrix, the diagonal elements of D are the eigenvalues of the covariance matrix , the number of non-zero eigenvalues is , the column vectors of are the eigenvectors of 5. The method of claim 4, wherein: The desired recipient selects Principal components generate sparse vectors to indicate Then use a binary matrix from Choose the most important M One piece of information: 。 6. The physical layer authentication method for a massive MIMO communication system according to claim 5, characterized in that: binary... The matrix selects channel information corresponding to the first M characteristic values, and the characteristic values are arranged from large to small. ; The covariance matrix is: ； The covariance matrix of the theoretical channel matrix after PCA dimensionality reduction is: ​ ； To: ； When the PCA dimension reduction is used, a compression ratio of the channel information is: 。 7. The method of claim 2, wherein the method further comprises: When using SVD dimension reduction, the desired receiver singular value decomposes the channel matrix to obtain a sparse representation matrix, resulting in an SVD dimension-reduced channel matrix ; Singular value decomposition of the channel matrix where is a diagonal matrix with the singular values on the diagonal and the number of non-zero singular values is , and are the left and right singular matrices, respectively;​ The SVD sparse representation matrix for is: ； The desired recipient is identified by using The principal components are selected The sparse vector is generated to represent : ； Then, the desired recipient uses the binary matrix to select the most important pieces of information: M ​ ； The covariance matrix of the is: ； The covariance matrix of the is: ； A covariance matrix of the channel estimation error after the SVD dimension reduction is: 。 8.The method of claim 7, wherein: A compression ratio of the channel information after the SVD dimension reduction is: 。 9.The physical layer authentication method for massive MIMO communication system of claim 1, wherein: In the initial communication, the expected receiver uses an upper layer authentication scheme to determine whether the signal sender is the legal user equipment, if yes, the channel frequency response of the legal communication link is estimated and stored in a feature database, so as to facilitate subsequent authentication, if not, an alarm is triggered. 10.The physical layer authentication method for massive MIMO communication system of claim 1, wherein: The legitimate user equipment deploys a single dual-polarized antenna, and the antenna base station deploys a dual-polarized antenna.

Images