A holographic hybrid precoding method based on reconfigurable smart metasurface

Abstract

A holographic hybrid precoding method based on a reconfigurable intelligent metasurface belongs to the technical field of precoding for mobile communications. The present invention applies the holographic principle to design a codebook, and the transmitting end simulates the precoding matrix to select a holographic pattern from the codebook to achieve dynamic beamforming and simplify the iterative steps; the characteristics of the holographic pattern are used to combine multiple feed sources for precoding, and the holographic patterns of different feed sources have different gains in the same direction, and the holographic pattern is preferentially matched to the feed source with the highest gain, thereby improving the bit error rate and spectrum efficiency performance of the communication system; a layered holographic codebook is used to roughly match the holographic pattern through the first layer of the codebook, and then a second layer of the codebook with refined sampling is generated according to the rough matching result, thereby reducing the computational complexity. The present invention is applicable to the field of communications, and is used to improve the bit error rate and spectrum efficiency performance of MIMO-OFDM communication systems assisted by reconfigurable intelligent metasurfaces, and reduce computational complexity.

Description

Technical Field

[0001] The invention relates to a holographic hybrid precoding method based on a reconfigurable intelligent super surface, belonging to the technical field of precoding of mobile communications. Background Art

[0002] Antennas are terminal devices of airborne satellite communication systems, and their working performance is crucial to the overall performance of the system. In existing satellite communication systems, ground receiving antennas are divided into two types: reflector antennas and phased array antennas. Although they can achieve high gain, they each have obvious disadvantages. The transmission direction of the reflector antenna needs to be mechanically controlled, and its deployment is limited by environmental obstacles; the feeding network of the phased array antenna consumes a lot of power, and its RF circuit design is complex and costly, which is not conducive to practical application. Therefore, a new type of antenna that is thinner, has low power consumption, low hardware cost, and good beam control capability is needed to meet the mobile communication requirements of satellites with high data rates. Among the existing antenna technologies, holographic antennas, as a small-sized, low-power planar antenna, have attracted more and more attention for their low manufacturing cost and low hardware cost multi-beam control capabilities. Specifically, holographic antennas use meta-patches to construct holographic patterns on the surface. These meta-patches record the interference between the reference wave and the desired target wave based on the principle of interference. The holographic pattern can change the radiation characteristics of the reference wave to generate a radiation pattern in the desired direction.

[0003] However, with the rapid increase in the number of mobile devices, traditional holographic antennas face great challenges because once the holographic pattern is generated, its radiation pattern is fixed. The emerging reconfigurable holographic metasurface adaptively changes the holographic pattern to the channel environment through the controllability of metamaterials. Thanks to the tunability and programmability of metamaterials, this metasurface shows great potential in making up for the shortcomings of traditional holographic antennas. It has a small-size ultra-thin planar structure suitable for transceiver integration and flexible beam steering capabilities with low hardware cost. Compared with traditional phased array antennas, this technology reduces the reliance of RHS on complex phase shifters. Therefore, RHS (Reconfigurable Holographic Surface) can achieve dynamic beamforming without heavy mechanical mobile equipment and complex phase shift circuits. It can also be installed on mobile platforms to provide high-throughput data services.

[0004] The research on RHS in academia and industry mainly focuses on hardware component design and radiation pattern control. In the literature, a parallel plate waveguide RHS based on PCB (Printed Circuit Board) is proposed, which can generate a multi-beam radiation pattern RHS, proving the feasibility of RHS to achieve dynamic multi-beam steering. In the literature "O.Yurduseven and DR Smith, "Dual-Polarization Printed Holographic Multibeam Metasurface Antenna," in IEEE Antennas and Wireless Propagation Letters, vol.16, pp.2738-2741, 2017.", a parallel plate waveguide RHS based on PCB is proposed, which can generate a multi-beam radiation pattern RHS, proving the feasibility of RHS to achieve dynamic multi-beam steering.

[0005] However, most existing works only demonstrate the feasibility of RHS to achieve dynamic multi-beam steering. There is little analysis of the performance gain of holographic beamforming technology on wireless communication systems. Holographic beamforming poses new challenges to maximizing the inflection rate. Since the traditional phased analog precoder is a complex-valued matrix with complex-domain phase constraints, existing algorithms cannot be well applied to amplitude-controlled holographic beamforming. A new beamforming scheme needs to be developed to deal with complex domain optimization problems subject to unconventional real-domain amplitude constraints and the superposition of radiated waves from different radiating units. The literature "R. Deng, B. Di, H. Zhang, Y. Tan and L. Song, "Reconfigurable Holographic Surface: Holographic Beamforming for Metasurface-Aided Wireless Communications," in IEEE Transactions on Vehicular Technology, vol. 70, no. 6, pp. 6255-6259, June 2021." The performance of the communication system is improved by designing a holographic pattern that is adaptive to the channel. The literature takes spectrum efficiency as the optimization goal and uses a convex optimization algorithm to obtain the holographic precoding matrix through multiple iterations. However, the holographic principle is deviated when designing the precoder, resulting in a high algorithm complexity. Summary of the invention

[0006] In view of the shortcomings of the existing precoding methods with high complexity and complicated iterative process, the main purpose of the present invention is to provide a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, apply the holographic principle to optimize the precoding process, and generate a holographic layered codebook for the analog precoding matrix to select codewords for transmitting end beamforming, so that the iterative steps are simple and easy to converge, thereby improving the bit error rate and spectrum efficiency performance of the MIMO (Multiple-Input Multiple-Output)-OFDM (Orthogonal Frequency Division Multiplexing) communication system assisted by the reconfigurable intelligent metasurface, and reducing the computational complexity under the same quantization accuracy.

[0007] The objective of the present invention is achieved through the following technical solutions:

[0008] The present invention discloses a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, which obtains a holographic layered codebook according to the holographic principle, and applies the codebook to the transmitting end holographic simulation precoding of a multi-feed joint SOMP (Simultaneous Orthogonal Matching Pursuit) algorithm, thereby improving the bit error rate and spectrum efficiency performance of the MIMO-OFDM communication system assisted by the reconfigurable intelligent metasurface, while reducing the computational complexity.

[0009] The present invention discloses a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, comprising the following steps:

[0010] Step 1: Generate a holographic hierarchical codebook based on the holographic principle to be applicable to a hybrid precoder in a scenario where the base station uses the RHS as an antenna, and adopt a hierarchical codebook structure to reduce computational complexity;

[0011] Step 1.1: Calculate the azimuth sampling value set of the first layer codebook And the pitch angle sampling value set Where M and N are the number of antennas of the UPA (Uniform Planar Array) in the corresponding dimensions;

[0012] Step 1.2: Generate reference wave and pointing direction based on holographic principle The object wave in the direction of , where θ is the azimuth angle, is the pitch angle; the interference pattern of the two waves is recorded as the column vector m k , the column vector represents the holographic pattern corresponding to the kth feed source, where m k The elements of correspond to the amplitude of the electromagnetic waves radiated by each array element of RHS;

[0013] Step 1.3, for θ∈T1, Any θ, The first layer of holographic layered codebook is obtained from step 1.2

[0014] Step 1.4: Obtain the optimal all-digital precoding matrix F from the result of the frequency domain channel matrix SVD (Singular Value Decomposition) opt , matched by the SOMP algorithm The closest to the full digital precoding matrix F opt The holographic pattern and obtain the index i of the holographic pattern in the codebook;

[0015] Step 1.5, calculating the indexes t1 and p1 of the azimuth and elevation angles corresponding to the matched holographic pattern from the index i;

[0016] Step 1.6: Calculate the azimuth sampling value set of the second layer codebook And the pitch angle sampling value set Then, the second-layer holographic hierarchical codebook is obtained by step 1.2

[0017] Step 2, obtaining the holographic analog precoding matrix at the transmitting end by approximating the all-digital optimal precoding matrix;

[0018] Step 2.1, perform SVD on the frequency domain channel matrix to obtain the fully digital optimal precoding matrix as the approximation target of the transmitter and encoder;

[0019] Step 2.2, using a multi-feed joint SOMP algorithm to select a holographic pattern from the holographic layered codebook to approximate the all-digital optimal precoding matrix; wherein the multi-feed joint SOMP algorithm combines the characteristics of the holographic codebook on the basis of the original SOMP algorithm, combines multiple feeds to jointly match the holographic pattern, and obtains the transmitter analog precoding matrix;

[0020] The second-layer holographic layered codebook corresponding to the k-th feed source is obtained by the layered codebook scheme proposed in step 1. Where k = {1, 2, ..., K}, K is the number of feed sources; the second layer holographic layered codebooks calculated for different k values ​​are The codebook D2 is formed as the perception matrix input to the SOMP algorithm. The holographic pattern index i corresponding to the maximum inner product value and the feed source k0 corresponding to the holographic pattern are found. After each iteration, all codeword sets corresponding to the feed source index k0 of D2(:,i) are deleted from D2. And let the holographic simulation precoding matrix M corresponding to the lth user be l Medium l(:, k0) = D2(:, i), iterate K times to get the complete M l Then the M corresponding to different users l The matrices are concatenated into matrix M.

[0021] Step 3: Apply the MMSE (Minimum Mean Square Error) criterion to the product of the frequency domain channel and the precoding matrix to obtain the fully digital optimal combiner corresponding to the lth user, and use the SOMP algorithm to find the analog combiner matrix W that is closest to the combiner. l With the digital combiner matrix P l .

[0022] Step 4: Use the zero-forcing method to obtain the transmitting end digital precoding matrix V by calculating the pseudo-inverse of the equivalent channel matrix from the receiving end baseband to the transmitting end RF chain.

[0023] The method also includes step 5, deploying the transmitting RHS and receiving array antenna of the communication transmission system according to the precoding matrix obtained in step 4, realizing dynamic beamforming, preferentially matching the holographic pattern to the feed source with the highest gain, improving the bit error rate and spectrum efficiency performance of the MIMO-OFDM communication system assisted by the reconfigurable intelligent metasurface, and reducing the computational complexity.

[0024] Beneficial effects:

[0025] 1. The present invention discloses a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, which applies the holographic principle to design a codebook. The transmitting end simulates the precoding matrix to select a holographic pattern from the codebook to achieve dynamic beamforming, which simplifies the iterative steps compared to the traditional convex optimization method.

[0026] 2. The present invention discloses a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, which utilizes the characteristics of the holographic pattern to jointly perform precoding design with multiple feed sources, and utilizes the fact that the holographic patterns of different feed sources have different gains in the same direction to preferentially match the holographic pattern to the feed source with the highest gain, thereby improving the bit error rate performance and spectrum efficiency performance of the MIMO-OFDM communication system assisted by the reconfigurable intelligent metasurface.

[0027] 3. The present invention discloses a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, which adopts a layered holographic codebook scheme. The holographic pattern is roughly matched through the first layer codebook, and then a second layer codebook with fine sampling is generated according to the rough matching result. When the bit error rate and spectrum efficiency performance are better than those of the holographic redundant codebook, the computational complexity is lower than that of the holographic redundant codebook. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1This is a model diagram of a massive MIMO-OFDM system using a holographic hybrid precoding solution in a millimeter wave scenario in this embodiment;

[0029] Figure 2 This is a flow chart of a holographic hybrid precoding method based on a reconfigurable intelligent metasurface disclosed in the present invention;

[0030] Figure 3 It is a schematic diagram of the hierarchical codebook principle in a holographic hybrid precoding method based on a reconfigurable intelligent metasurface disclosed in the present invention;

[0031] Figure 4 is the diagram of RHS,

[0032] Where (a) is the schematic structure diagram of RHS, and (b) is the geometric relationship between RHS and target beam;

[0033] Figure 5 This is the RHS deployment structure diagram in this embodiment.

[0034] The square is the antenna position and the circle is the feed position;

[0035] Figure 6 Schematic diagram of bit error rate evaluation in this embodiment;

[0036] Figure 7 FIG. 4 is a schematic diagram of spectrum efficiency evaluation in this embodiment. DETAILED DESCRIPTION

[0037] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. The technical problems solved by the technical solution of the present invention and the beneficial effects are also described. It should be noted that the described embodiments are only intended to facilitate the understanding of the present invention and do not have any limiting effect on it.

[0038] This embodiment considers a massive MIMO-OFDM system using a hybrid precoding scheme in a millimeter wave scenario. Figure 1 As shown. A multi-user hybrid MIMO system using OFDM is considered to overcome the time-selective fading on the wideband channel. Assume that both the transmitting and receiving antennas use UPA. The transmitting end uses M×N radiating elements as the antenna array as mentioned above, and K RF chains are each connected to a feed source and K<<M×N. The receiving end also uses the UPA array, and each user is equipped with J×J antennas. The number of RF chains M r The number of received data streams N s Since the RHS does not have any digital signal processing capabilities, the base station needs to process the signal at the baseband. Assume that the base station first uses a digital precoder to process N s×L different data streams are precoded, and then the processed signals are up-converted to carrier frequency through RF chains. Each RF chain is connected to the feed source of RHS, that is, the number of RF chains is the same as the number of feed sources K.

[0039] Send N s data streams, there are KL feed sources at the base station, and each K feed sources serve one user. The signal vector transmitted by the cth subcarrier can be expressed as s[c], c = {1,...,N c}, then after the baseband precoder V[c], the signal transmitted by the base station is V[c]s[c]. Then, a k-point IFFT (Inverse Fast Fourier Transform) is performed on each RF chain to transform the signal vector into the time domain. After adding a cyclic prefix, the RF chain up-converts it to the feed source to which it is connected. The feed source then converts the high-frequency current into an electromagnetic wave (also called a reference wave) propagating on the RHS. The radiation amplitude of the reference wave at each radiating element is controlled by the holographic analog precoding matrix M to generate the required directional beam.

[0040] The signal finally received by mobile user l can be expressed as:

[0041]

[0042] Among them, H l [c] is the complex channel gain matrix from RHS to the lth user antenna, V l [c] is the lth column of V[c], z l [c]~CN(0,σ 2 t J ) is additive Gaussian white noise, t J W is a J×1 dimension column vector with all element values ​​1. l is each mobile user J×N RF dimensional analog combiner, implemented using phase shifters, so that |W l (i,j)| 2 =1 is always true, N RF is the number of RF chains at the receiving end. l [c] is the digital combiner at the receiving end. Consider a uniform linear phased array at each mobile user, where d s is the distance between the antennas of the mobile users, φ l is the angle of arrival of the signal relative to the antenna array at mobile user l.

[0043] The channel models involved in this implementation are as follows:

[0044] The millimeter-wave MIMO channel for airborne communications can be viewed as a Saleh-Valenzuela (SV) cluster sparse channel with large-scale fading, that is, a LoS (Line-of-Sight) and N c It consists of clusters of scatterers, each of which contains N p There are NLoS (Non-Line-of-Sight) paths, and these NLoS paths have similar delays. Then the delay domain MIMO channel matrix of the lth user at d delay sampling points can be expressed as:

[0045]

[0046] in and are the delay domain channel coefficients of the LoS path and NLoS path corresponding to the p-th multipath in the c-th cluster, respectively.

[0047] is the complex gain, where μ is the Rice factor, which represents the energy distribution between the NLoS path and the LoS path.

[0048] P=N cl N p +1 is the total number of multipaths in this channel model, N T =M×N,N R = J × J are the total number of antennas at the transmitter and the total number of antennas at the receiver, and d is the distance between the transmitter and the receiver. p(τ) is the signal at T s is the pulse shaping filter after interval sampling, is the relative delay. and The angles in each cluster follow a normal distribution with a constant standard deviation, and their variances are and are the normalized receive and transmit array steering vectors, respectively.

[0049] Both the transmitter and the receiver use the UPA antenna array model. Here, without loss of generality, the array is assumed to be located in the yz plane and the number of antennas on the y-axis and z-axis are N respectively. y and N z , then the array steering vector can be expressed as:

[0050]

[0051] Where n and m satisfy 1≤n<N y , 1≤m<N z , are the y-coordinate and z-coordinate indices of the antenna array element, λ and They represent the wavelength and the interval between adjacent antennas respectively.

[0052] The frequency domain channel is the FFT (Fast Fourier Transform) of the time domain channel. c For an OFDM system with subcarriers, the channel response of the cth subcarrier can be expressed as Where D l is the length of the cyclic prefix.

[0053] This implementation is based on a dual-user downlink communication system modeled based on the above scenario, discloses a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, introduces in detail the generation method of a layered codebook and generates a precoder for dynamic beamforming according to a real-time channel, simulates the bit error rate and spectral efficiency of the communication system, and analyzes the simulation results and computational complexity.

[0054] The holographic hybrid precoder adopts a hybrid precoding architecture, including a transmitter analog precoder, a transmitter digital precoder, a receiver analog synthesizer and a receiver digital synthesizer. The K column vectors of the transmitter analog precoder are the first K code words selected from the holographic code book using the multi-feed joint SOMP algorithm, where K is the number of transmitter feeds; the J column vectors of the receiver analog synthesizer are the first J code words selected from the DFT (Discrete Fourier Transform) code book using the SOMP algorithm, where J is the number of receiver feeds. The transmitter digital precoder and the receiver digital combiner are both obtained by the zero-forcing method. Figure 2 As shown, the present embodiment discloses a holographic hybrid precoding method based on a reconfigurable intelligent metasurface, which specifically comprises the following steps:

[0055] Step 1: Generate a holographic hierarchical codebook based on the holographic principle to be applicable to a hybrid precoder in a scenario where the base station uses the RHS as an antenna, and adopt a hierarchical codebook structure to reduce computational complexity;

[0056] For the design method of holographic layered codebook, the first layer holographic codebook is uniformly sampled in the angle domain; the holographic pattern with the highest matching degree is selected from the first layer codebook, and the sampling range of the second layer codebook is determined by the index of the holographic pattern, so as to obtain the second layer codebook, such as Figure 3 As shown, the following steps are included:

[0057] Step 1.1, generate the holographic pattern corresponding to the RHS in the direction according to the holographic principle, where θ is the azimuth angle, is the pitch angle;

[0058] Step 1.1.1. Calculate the directional propagation vector corresponding to the target beam in free space

[0059] Step 1.1.2: Calculate the target beam where r m,n is the position vector of the (m,n)th radiation element, where 1≤n≤N, 1≤m≤M;

[0060] Step 1.1.3: Calculate the reference wave generated by feed source k where k s is the propagation vector of the reference wave, is the distance vector from feed source k to the (m,n)th radiating element;

[0061] Step 1.1.4: Calculate the interference pattern obtained by interfering the reference wave and the target beam

[0062] Step 1.1.5: Calculate the radiation amplitude of the (m,n)th radiation unit of RHS

[0063] Step 1.1.6: Repeat steps 1.1.2 to 1.1.5 to obtain the element The column vector m k , the column vector represents the holographic pattern corresponding to the kth feed source, where is the radiation amplitude of the (m,n)th radiation unit of RHS, is the distance vector from feed source k to the (m,n)th radiating element;

[0064] Step 1.2: Calculate the azimuth sampling value set of the first layer codebook And the pitch angle sampling value set Where M and N are the number of antennas of UPA in the corresponding dimensions;

[0065] Step 1.3, for θ∈T1, Any θ, Repeat step 1.1 to get the first layer of holographic layered codebook

[0066] Step 1.4: Take the frequency domain channel matrix H corresponding to the lth user l [c] = the first N of the right singular value matrix V of UΣV s The optimal full digital precoding matrix F is obtained by opt [c], and let the residual matrix F res [c]=F opt [c], where c = {1, 2, ..., N c},N c is the number of subcarriers;

[0067] Step 1.5: The first layer of holographic codebook As the perception matrix, calculate the perception matrix and residual F res [c]=F opt The inner product of [c]

[0068] Step 1.6: Find the column vector index corresponding to the maximum inner product value, that is,

[0069] Step 1.7, calculating the corresponding azimuth and elevation indexes t1 and p1 from the index i of the hologram;

[0070] Step 1.8: Calculate the azimuth sampling value set of the second layer codebook And the pitch angle sampling value set The second layer holographic layered codebook is obtained from step 1.1

[0071] Step 2, obtaining the holographic analog precoding matrix at the transmitting end by approximating the all-digital optimal precoding matrix;

[0072] Step 2.1: Initialize the transmitter RF precoding matrix

[0073] Step 2.2: Initialize the residual matrix F res [c]=F opt [c], support set index

[0074] Step 2.3: Generate the second layer holographic layered codebook corresponding to the kth feed source Where k = {1, 2, ..., K}, the second layer holographic layered codebook calculated for different k values Spell out code book D2;

[0075] Step 2.4: Use codebook D2 as the perception matrix. Calculate the perception matrix D2 and the residual F res The inner product of [c]

[0076] Step 2.5: Find the column vector index corresponding to the maximum inner product value, that is, D is the number of columns in the holographic codebook, find the feed source k0 corresponding to the codeword;

[0077] Step 2.6: Update support set index The holographic pattern set matrix M corresponding to user l l ,Right now M l (:,k0)=D2(:,i);

[0078] Step 2.7: Find the least squares solution in represents the pseudo-inverse of the matrix; updates the residual

[0079] Step 2.8: Delete all codeword sets corresponding to the feed index k0 of D2(:,i) from D2 And let the holographic simulation precoding matrix M corresponding to the lth user be l Medium l (:,k0)=D2(:,i);

[0080] Step 2.9: Repeat steps 2.4 to 2.8 for K times to obtain the complete M l ;

[0081] Step 2.10: Repeat steps 2.2 to 2.9 for l = {1, 2, ..., L}, and update M = {M, M l}Get the complete transmitter RF precoding matrix; where L is the total number of users.

[0082] Step 3: Apply the MMSE (Minimum Mean Square Error) criterion to the product of the frequency domain channel and the precoding matrix to obtain the fully digital optimal combiner corresponding to the lth user, and use the SOMP algorithm to find the analog combiner matrix W that is closest to the combiner. l With the digital combiner matrix P l [c];

[0083] Step 3.1: The conjugate matrix of the optimal all-digital MMSE synthesizer on the cth subcarrier is: in is the transmitting end baseband precoder that is closest to the optimal all-digital precoding matrix under power constraints; A transmitting end baseband precoder that is closest to the optimal all-digital precoding matrix;

[0084] Step 3.2: Generate a two-dimensional DFT codebook A of a J×J size UPA array r ; Where θ is the azimuth angle, is the elevation angle, and the azimuth angle set is The pitch angle set is 1≤n<N,1≤m<M;

[0085] Step 3.3: Initialize the RF synthesizer Let the residual matrix W res [c]=W MMSE,l [c], support set index

[0086] Step 3.4: Calculate the two-dimensional DFT codebook A r and the residual W res The inner product of [c] in is the signal received by the antenna end of user l;

[0087] Step 3.5: Find the column vector index corresponding to the maximum inner product value, that is, Where D is the number of columns in the codebook;

[0088] Step 3.6: Update support set index and the RF synthesizer matrix W l ,Right now W l = {W l ,A r (:,i)};

[0089] Step 3.7: Find the baseband synthesizer

[0090] Step 3.8: Update the residual

[0091] Step 3.9: Repeat steps 3.4 to 3.8 for a total of N RF times, and obtain the final RF synthesizer W l and baseband synthesizer P l [c].

[0092] Step 4: Use the zero forcing method to obtain the pseudo-inverse of the equivalent channel matrix from the baseband of the receiving end to the radio frequency chain of the transmitting end to obtain the digital precoding matrix V of the transmitting end;

[0093] Step 4.1. Calculation The equivalent channel stacking matrix from the transmitting end baseband to the receiving end RF of all users;

[0094] Step 4.2, calculation The baseband precoder at the transmitting end without adding power constraint;

[0095] Step 4.3: Get the baseband precoder at the transmitting end.

[0096] Step 5: Deploy the transmitting end RHS and receiving end array antenna of the communication transmission system according to the precoding scheme of steps 1 to 4, wherein the structure of the transmitting end RHS is as follows: Figure 4 shown.

[0097] Specifically, in the simulation, the RHS array size at the base station is N t =M×N=6×6 UPA array, each user receiving end is equipped with Nr =L×L=4×4 UPA array. The carrier frequency is 10GHz, and the antenna spacing is half of the wavelength corresponding to the carrier frequency, that is, 0.0075m. In this case, it can meet the requirements of complete sampling in the angle domain of π, without mirror angles and the beam is as narrow as possible, which is more conducive to refined channel estimation and beamforming. Take the number of subcarriers as N c =32, the subcarrier spacing is 15kHz. Assume the number of users L = 2, and the number of data streams transmitted to each user is N s =2, the number of feed sources serving each user at the base station, and also the number of RF chains serving each user is K = 2, so there are a total of 4 feed sources at the base station, generating 4 data streams to transmit to the user. The number of RF chains at each receiving end is 2. When calculating noise, the dB value of noise power can be expressed as σ 2 =-174dBm+10log 10 (BW), where BW is the bandwidth.

[0098] For the airborne transmission channel, the parameters are set as follows: except for one LoS path, the number of NLoS clusters is set to N cl =4, each cluster has N p = 4 propagation paths. The azimuth and elevation angles conform to The uniform distribution within the range is In the millimeter wave channel, the channel gain of the LoS path is much greater than that of the NLoS path, so the Rice factor μ used for power allocation in the simulation is set to 2.3, which means that the power of the LoS path is 23dB higher than the sum of the powers of all NLoS paths. 16QAM modulation is used in all simulations.

[0099] The RHS array and feed position at the user end are as follows Figure 5 As shown in the figure, the square is the position of the RHS radiation element, and the circle is the feed position. Assume that the coordinates of the upper left corner feed are (x1, y1), and the coordinates of the lower right corner feed are (x M ,y N ), the coordinates of the four feed sources are The electromagnetic wave emitted by the feed source is the reference wave. The reference wave propagates on the waveguide and interacts with the holographic interference pattern stored in the amplitude controller at each RHS radiation element, thereby changing the radiation characteristics of the reference wave and generating the desired target beam.

[0100] Figure 6 and Figure 7The diagrams are respectively the bit error rate evaluation diagram and the spectrum efficiency evaluation diagram, in which the comparison schemes used are the non-redundant codebook scheme, the redundant codebook scheme and the prior-assisted scheme. The cases of ρ=1 and ρ=2 are the precoding schemes obtained by using the non-redundant holographic codebook and the holographic codebook with redundancy of 2 as the sensing matrix of the multi-feed joint SOMP algorithm. Prior-assisted means using the channel multipath component steering vector as a priori condition. First, it is assumed that the transmitter and the receiver also use ordinary array antennas. The AoD corresponding to each multipath can generate the array steering vector of the transmitter. The transmitting end steering vectors generated by all multipaths are used as column vectors to form a steering vector matrix. This matrix is ​​then used as the sensing matrix of the SOMP algorithm to match the fully digital optimal precoding matrix. The corresponding azimuth and elevation angles are inverted from the matched steering vectors, and the holographic pattern in that direction is generated according to each feed position.

[0101] Figure 6 The bit error rate (BER) performance of different analog precoding schemes is compared. Among them, all precoding schemes use the codebook-based SOMP algorithm, and the all-digital optimal precoding matrix they try to approximate is the same, which is the first N right singular matrices obtained after SVD of the channel matrix. s Among the four precoding schemes, the BER performance is the worst when using the non-redundant holographic codebook, while the BER of the holographic layered codebook is better than that of the holographic codebook with redundancy of 2, and is closer to the BER performance of the perfect beam alignment case with the prior assistance.

[0102] Figure 7 The SE curves under different precoding schemes are compared, and the performance ranking of the four precoding schemes is Figure 6 The situation in is exactly the same. From the simulation figure, it can be seen that the performance of the four precoding schemes is the same: the prior-aided steering vector as the sensing matrix is ​​better than the layered holographic codebook as the sensing matrix, which is better than the holographic codebook with redundancy of 2, and the non-redundant holographic codebook has the worst performance, which is also consistent with the expectations when the codebook is designed.

[0103] The simulation results confirm that under the holographic precoding architecture, the higher quantization accuracy provided by the proposed holographic layered codebook provides better performance for the system's BER and SE.

[0104] The following mainly analyzes the computational complexity of the holographic analog precoder and compares the computational complexity of the proposed layered codebook generation method with that of the redundant codebook. Assuming that the number of codebook columns as the perception matrix is ​​W, the computational complexity can be expressed as It can be seen that the computational complexity is linearly related to the codebook dimension. The dimension of the non-redundant holographic codebook is N r ×N rK, the dimension of the holographic codebook with redundancy ρ is N r ×ρN r K, the dimension of each layer of hierarchical holographic codebook is N r ×N r K. It can be seen that the complexity of the SOMP algorithm using a non-redundant codebook as the sensing matrix is ​​the lowest, but it corresponds to a larger quantization error, which is reflected in the BER and SE curve simulations. When the number of transmitting antennas is large, the complexity is mainly due to ((2N t +1)WKN c ), the computational complexity of using a holographic codebook with redundancy ρ = 2 is the same as the complexity of the hierarchical holographic codebook scheme using the SOMP algorithm twice, but the quantization accuracy of the holographic hierarchical codebook is equivalent to that of the redundancy Quantization accuracy of the codebook. The holographic principle is applied to optimize the precoding process. By generating a holographic layered codebook for the simulated precoding matrix to select codewords for transmitter precoding, the iterative steps are simple and easy to converge, reducing the computational complexity at the same quantization accuracy.

[0105] The specific description above further illustrates the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A holographic hybrid precoding method based on a reconfigurable intelligent metasurface, characterized in that: It includes the following steps: Step 1: Generate a holographic hierarchical codebook based on the holographic principle to be applicable to a hybrid precoder in a scenario where the base station uses the RHS as an antenna, and adopt a hierarchical codebook structure to reduce computational complexity; Step 2, obtaining the holographic analog precoding matrix at the transmitting end by approximating the all-digital optimal precoding matrix; Step 3: Apply the minimum mean square error criterion to the product of the frequency domain channel and the precoding matrix to obtain the fully digital optimal combiner corresponding to the lth user, and use the SOMP algorithm to find the analog combiner matrix W that is closest to the combiner. l With the digital combiner matrix P l ; Step 4: Use the zero-forcing method to obtain the transmitting end digital precoding matrix V by calculating the pseudo-inverse of the equivalent channel matrix from the receiving end baseband to the transmitting end RF chain.

2. The holographic hybrid precoding method based on a reconfigurable intelligent metasurface according to claim 1, characterized in that: The method also includes step 5, deploying the transmitting RHS and receiving array antenna of the communication transmission system according to the precoding matrix obtained in step 4, realizing dynamic beamforming, preferentially matching the holographic pattern to the feed source with the highest gain, improving the bit error rate and spectrum efficiency performance of the MIMO-OFDM communication system assisted by the reconfigurable intelligent metasurface, and reducing the computational complexity.

3. A holographic hybrid precoding method based on a reconfigurable intelligent metasurface as claimed in claim 2, characterized in that: The implementation method of step 1 is: Step 1.1: Calculate the azimuth sampling value set of the first layer codebook And the pitch angle sampling value set Where M and N are the number of antennas in the uniform planar array in the corresponding dimensions; Step 1.2: Generate reference wave and pointing based on holographic principle The object wave in the direction of , where θ is the azimuth angle, is the pitch angle; the interference pattern of the two waves is recorded as the column vector m k , the column vector represents the holographic pattern corresponding to the kth feed source, where m k The elements of correspond to the amplitude of the electromagnetic waves radiated by each array element of RHS; Step 1.3: For Any θ, The first layer of holographic layered codebook is obtained from step 1.2 Step 1.4: Obtain the optimal all-digital precoding matrix F from the result of the frequency domain channel matrix SVD opt , matched by the SOMP algorithm The closest to the full digital precoding matrix F opt The holographic pattern and obtain the index i of the holographic pattern in the codebook; Step 1.5, calculating the indexes t1 and p1 of the azimuth and elevation angles corresponding to the matched holographic pattern from the index i; Step 1.6: Calculate the azimuth sampling value set of the second layer codebook And the pitch angle sampling value set Then, the second-layer holographic hierarchical codebook is obtained by step 1.2 4. The holographic hybrid precoding method based on a reconfigurable intelligent metasurface according to claim 3, characterized in that: The implementation method of step 2 is: Step 2.1, perform singular value decomposition on the frequency domain channel matrix to obtain the fully digital optimal precoding matrix as the approximation target of the transmitter and the encoder; Step 2.2, using a multi-feed joint SOMP algorithm to select a holographic pattern from the holographic layered codebook to approximate the all-digital optimal precoding matrix; wherein the multi-feed joint SOMP algorithm combines the characteristics of the holographic codebook on the basis of the original SOMP algorithm, combines multiple feeds to jointly match the holographic pattern, and obtains the transmitter analog precoding matrix; The second-layer holographic layered codebook corresponding to the k-th feed source is obtained by the layered codebook scheme proposed in step 1. Where k = {1, 2, ..., K}, K is the number of feed sources; the second layer holographic layered codebooks calculated for different k values ​​are The codebook D2 is formed as the perception matrix input to the SOMP algorithm. The holographic pattern index i corresponding to the maximum inner product value and the feed source k0 corresponding to the holographic pattern are found. After each iteration, all codeword sets corresponding to the feed source index k0 of D2(:,i) are deleted from D2. And let the holographic simulation precoding matrix M corresponding to the lth user be l Medium l (:, k0) = D2(:, i), iterate K times to get the complete M l ; Then the M corresponding to different users l The matrices are concatenated into matrix M.

Images