A reconfigurable intelligent surface configuration method based on statistical channel state information

Abstract

The application belongs to the technical field of wireless communication, and particularly relates to a reconfigurable intelligent surface configuration method based on statistical channel state information. The method is used in a reconfigurable intelligent surface assisted wireless communication scene, according to the reciprocity of time division duplex uplink and downlink channels, and statistical channel state information estimation values are obtained by using an improved variational posterior matching method and a parameter estimation method in the uplink process, and reconfigurable intelligent surface configurations are obtained by using the statistical channel state estimation values and an optimization algorithm in the downlink process, so that the downlink receiving rate of users is improved. The method can effectively improve the average receiving rate of the lowest user.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless communication, and particularly relates to a reconfigurable intelligent surface configuration method based on statistical channel state information. TECHNICAL BACKGROUND

[0002] With the development of material technology, a reconfigurable intelligent surface which can actively change the electromagnetic propagation environment at a low cost emerges as the times require. The reconfigurable intelligent surface is arranged between the transmitting end and the receiving end of wireless signals, so that the dynamic adjustment of the wireless channel can be realized, and the performance of the wireless communication system can be improved.

[0003] The reconfigurable intelligent surface is composed of a large number of units with the same structure and similar functions, and each unit contains circuit devices with variable electrical characteristics, such as diodes, variable capacitors, etc. [1-5] Through the controller, these units can be independently controlled to change the reflection coefficient of the reconfigurable intelligent surface to the incident electromagnetic wave, so as to generate a direction-controllable reflected electromagnetic wave. The direction of the electromagnetic wave is also reflected in the phase of the electromagnetic wave. When the reflected electromagnetic wave at the target position is in phase, the received power at the target position increases; when the reflected electromagnetic wave at the target position is out of phase, the received power at the target position decreases. In order to configure the reconfigurable intelligent surface, it is necessary to estimate the channel state information of the wireless channel. Since the reconfigurable intelligent surface does not have the ability of channel estimation, the time overhead required by the base station or the user to estimate the wireless channel related to the reconfigurable intelligent surface is proportional to the number of reconfigurable intelligent surface units [6-8] . Frequent estimation of the instantaneous channel state information of these wireless channels will occupy a large amount of system resources and reduce the performance of the wireless communication system [9,10] . Compared with the instantaneous channel state information, the statistical channel state information has a slower change frequency and is suitable for configuring the reconfigurable intelligent surface. SUMMARY

[0004] The application provides a reconfigurable intelligent surface configuration method for assisted downlink communication which can effectively improve the downlink receiving rate of users.

[0005] The reconfigurable intelligent surface configuration method for assisted downlink communication provided by the application is based on statistical channel state information, and covers the estimation of statistical channel state information, user scheduling, reconfigurable intelligent surface configuration and the like, which can effectively improve the lowest average downlink transmission rate of the scheduled users.

[0006] The application is applicable to a system model as shown in Figure 1 . The base station has N B antennas and serves K single-antenna users. The base station uses a time division duplex mode to distinguish the uplink and downlink of signals, and directly calculates the channel state information of the downlink channel by using the channel state information of the uplink channel by using the reciprocity of the uplink and downlink channels. The reconfigurable intelligent surface is located between the base station and the users, and has NR There are N reconfigurable smart surface elements. The reconfigurable smart surface is a rectangular plane, which can be divided into a μ-axis and a v-axis. The number of elements along the μ-axis of the reconfigurable smart surface is N. R,μ The number of elements along the v-axis of the reconfigurable smart surface is N. R,v Satisfying N R =N R,μ N R,v .

[0007] This invention primarily focuses on the direct path of a wireless channel. In the downlink channel, the direct path from the base station to the reconfigurable smart surface is... in a B,R The modulus is N B a R,B The modulus is N R The locations of the base station and the reconfigurable smart surface are fixed, and the base station can know the direct path a in advance. B,R and a R,B The direct path from the reconfigurable smart surface to the k-th user is... Where a R,k The modulus is N R The reconfigurable smart surface is divided into μ-axis and v-axis, reaching the radial direction. in a R,k,μ The modulus is N R,μ a R,k,v The modulus is N R,v , It is the Kronecker product of the matrix.

[0008] This invention uses codewords from a hierarchical codebook based on a binary tree to approximate the direct path a. R,k,μ and a R,k,v The base station has a depth of L. μ The hierarchical codebook, the lth... μ Layer, nth μ The code is Among them l μ ∈{0,1,…,L μ}, The base station also has a depth of L v The hierarchical codebook, the lth... v Layer, nth v The code is Among them l v ∈{0,1,…,L v},

[0009] The application provides a configuration method of a reconfigurable intelligent surface assisting downlink communication, in a reconfigurable intelligent surface assisted wireless communication scene, by means of time division duplex reciprocity of uplink and downlink channels, an improved variational posterior matching method and parameter estimation method are used to obtain an estimated value of statistical channel state information in the uplink process, and the estimated value of the statistical channel state and an optimization algorithm are used to obtain the reconfigurable intelligent surface configuration in the downlink process, so that the downlink receiving rate of users is improved; and specific research cases show that the method can effectively improve the average receiving rate of the lowest user.

[0010] The application provides a configuration method of a reconfigurable intelligent surface assisting downlink communication, and the specific steps are as follows:

[0011] Step 1: The base station numbers the users, and the number k is in {1, 2,..., K}, wherein K is the number of users. The base station schedules the users for uplink communication in turn in a time division manner. The initial user number is k←1.

[0012] Step 2: The user k continuously transmits an uplink pilot. The base station receives the uplink pilot through the reconfigurable intelligent surface, changes the configuration of the reconfigurable intelligent surface according to the uplink pilot, and estimates the statistical channel state information χ k ,κ k and wherein is a complex number, N R is the number of reconfigurable intelligent surface units, and the modulus of a R,k is N R .

[0013] Step 3: The user number k is incremented. If k < K, steps 2 and 3 are repeated.

[0014] Step 4: The base station groups the users. The number of groups is J, and the groups are j∈1,2,…,J, the number of users in the group is K j , and the user number in the group is k, k∈{1,2,…,K j}. For any two users p and q in the group , the following condition is met wherein H is a conjugate symmetric operator.

[0015] Step 5: The base station calculates the reconfigurable intelligent surface configuration for the user group wherein j∈{1,2,…,J}.

[0016] Step 6: The base station communicates with the users in different groups in turn in a time division manner. The initial user group number is j←1.

[0017] Step 7: The base station configures the reconfigurable intelligent surface as φ j , and communicates with users in user group downlink.

[0018] Step 8: The user group number j is incremented. If the statistical channel state information χ k , κ k and a R,k change, repeat steps 1 to 8; if j≥J, repeat steps 6 to 8; otherwise, repeat steps 7 and 8.

[0019] Further:

[0020] The step of estimating the statistical channel state information χ k , κ k and a R,k of user k in step 2 specifically includes:

[0021] Step 2.1: The base station uses two hierarchical codebooks with depths L μ and L v , corresponding to a R,k,μ and a R,k,v respectively.

[0022] Step 2.2: The base station obtains the position prior of user k. Taking the μ-axis as an example, the base station selects (l μ , n μ ) from the hierarchical codebook that satisfies with a prior probability greater than 0.99, and records it as Similarly, the base station selects (l v , n v ) from the hierarchical codebook that satisfies with a prior probability greater than 0.99, and records it as

[0023] Step 2.3: The base station sets the observation time t←0, and sets the intermediate variable τ←1 and the intermediate variable s←1.

[0024] Step 2.4: The base station increments the observation time t.

[0025] Step 2.5: The base station calculates the configuration of the reconfigurable intelligent surface according to and , that is, where ⊙ is the bitwise multiplication operator, * is the conjugate operator, is the downlink channel vector when the reconfigurable intelligent surface receives the signal from the base station.

[0026] Step 2.6: User k transmits the uplink pilot where Tul For pilot length, the modulo of uplink pilot x is equal to p ul T ul , p ul is the power of a single uplink pilot symbol.

[0027] Step 2.7: The base station receives signals as where N B is the number of base station antennas. Using x H right-multiply Y, using left-multiply Y, we get where H is the conjugate transpose operator, is the uplink channel vector when the base station receives signals from the reconfigurable intelligent surface.

[0028] Step 2.8: The base station uses observations {y τ , y τ+1 , …, y t} and reconfigurable intelligent surface configurations {φ τ , φ τ+1 , …, φ t} to calculate

[11] and and

[0029] Step 2.9: If , the last 3 observation times and remain unchanged, go to Step 2.10; otherwise, repeat Steps 2.4 to 2.9.

[0030] Step 2.10: Repeat Steps 2.4 to 2.7 for 5 times.

[0031] Step 2.11: The base station estimates statistical channel state information and

[0032] Step 2.12: The base station calculates intermediate variables and according to observations {y1, y2, …, y t} and reconfigurable intelligent surface configurations {φ1, φ2, …, φ t}, where

[0033]

[0034]

[0035] and re-estimate statistical channel state information χ k and κ k , respectively, as​

[0036]

[0037]

[0038] where N B is the number of base station antennas, N R is the number of reconfigurable intelligent surface units, is the uplink base station received noise power, p dl,k is the downlink base station transmit power, is the downlink user k received noise power.

[0039] The steps of the improved variational hierarchical posterior matching method adopted by the base station in step 2.8 specifically include:

[0040] Step 2.8.1: If s = 1, jump to step 2.8.2; if s = 2, jump to step 2.8.5.

[0041] Step 2.8.2: Using observations {y τ ,y τ+1 ,…,y t} and reconfigurable intelligent surface configurations {φ τ ,φ τ+1 ,…,φ t}, the posterior probability that satisfies is calculated by using the variational inference method. The posterior probability of each code word in the hierarchical codebook is calculated according to the posterior probability where

[0042] Step 2.8.3:

[0043] Step 2.8.4: If the consecutive 3 observation times remain unchanged, s <— 2, τ <— t + 1, and jump out of step 2.8.

[0044] Step 2.8.5: Using observations {y τ ,y τ+1 ,…,y t} and reconfigurable intelligent surface configurations {φ τ ,φ τ+1 ,…,φ t}, the posterior probability that satisfies is calculated by using the variational inference method. The posterior probability of each code word in the hierarchical codebook is calculated according to the posterior probability

[0045] Step 2.8.6:

[0046] In the step 5, the step of calculating the reconfigurable intelligent surface configuration φ for the user group j specifically includes:

[0047] Step 5.1: initialize iteration number i = 0, intermediate variables intermediate variable intermediate variable γ (0) = 0, where is a real number, K j is the number of users of the user group . The correction parameter where N B is the number of base station antennas, and ψ is the digamma function. Since N B -K j + 1 ∈ {1, 2, …, N B}, the value of is limited and can be obtained in advance.

[0048] Step 5.2: the iteration number i is incremented.

[0049] Step 5.3: calculate the intermediate variable

[0050] Step 5.4: introduce an auxiliary variable γ. The intermediate variable α (i) is the solution of the following convex optimization problem:

[0051]

[0052]

[0053]

[0054]

[0055] The optimization problem can be converted into a second-order cone problem, and the intermediate variable α (i) can be obtained by using the interior point method.

[0056] Step 5.5: calculate the intermediate variable

[0057] Step 5.6: if |γ (i) - γ (i-1) | ≥ ∈, repeat steps 5.2 to 5.6, where ∈ is the iteration error.

[0058] Step 5.7: the reconfigurable intelligent surface configuration φ j is where ⊙ is a bit-wise multiplication operator, * is a conjugate operator, a R,B is a downlink channel vector of the reconfigurable intelligent surface receiving base station signals,

[0059]

[0060] round is a rounding operator, the 1st to the N1th elements of a R,1 the 1st to the N1th elements of a the N1+1th to the N1+N2th elements of a R，2 the N1+1th to the N1+N2th elements of a the 1st to the N th elements of a R the 1st to the N th elements of a the 1st to the N R th elements of a BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 : An example of a reconfigurable intelligent surface assisted wireless communication system.

[0062] Figure 2 : Normalized mean square error of statistical channel state information when the user coordinates are (x, 2, 0) m.

[0063] Figure 3 : Normalized mean square error of statistical channel state information when the user coordinates are (23, y, 0) m.

[0064] Figure 4 : Impact of reconfigurable intelligent surface configuration on system performance. DETAILED DESCRIPTION

[0065] The present application will be further described below through specific implementation cases.

[0066] In the embodiment, the base station, the reconfigurable intelligent surface and the users are located in a three-dimensional space. The base station antenna center coordinates are (0, 0, 6) m, the base station antenna array is a uniform linear array and is parallel to the y-axis, the number of base station antennas N B = 16. The number of users K = 3, and the user antenna coordinates are (21, 1, 0) m, (22, 2, 0) m and (23, 3, 0) m, respectively, and the number of user antennas is 1. The reconfigurable intelligent surface center coordinates are (20, 5, 3) m, the unit array is a uniform planar array and is parallel to the xOz plane, and the number of units N R = 400, where N R，μ = 20, N R，v = 20. In addition, the downlink parameters uplink parameters Layered codebook for AMCF-ZCI

[12] , Layered codebook depth L μ = 6, L v = 6, user location prior follows truncated Gaussian distribution with variance 0.01 radian in radian domain, maximum number of base station observation set {y1, y2, …, y t} is 55.

[0067] In embodiments, appropriate assumptions are made for the channels between the base station, reconfigurable intelligent surface and users. The carrier wavelength of the wireless signal is λ = 10.7 mm, and the base station antenna spacing and reconfigurable intelligent surface element spacing are both half of the wavelength. The downlink channel from the base station to the reconfigurable intelligent surface is H, which follows a Rician distribution, i.e.,

[0068]

[0069] where the Rician factor is K H = 30, and the elements of the multipath component follow independent zero-mean unit-variance complex Gaussian distribution, and the path gain is The path loss factor is a B,R = 2.1, and the distance between the base station and the reconfigurable intelligent surface is d B,R , and the base station’s elevation angle to the reconfigurable intelligent surface is The downlink channel from the reconfigurable intelligent surface to the kth user is The reconfigurable intelligent surface is deployed around the user, so the multipath component of h R，k can be ignored, i.e., where the path gain is The path loss factor is a R,k = 2.0, and the distance between the reconfigurable intelligent surface and the kth user is d R,k , and the kth user’s elevation angle to the reconfigurable intelligent surface is The downlink channel from the base station to the kth user is h B,k , which follows a Rayleigh distribution, i.e., where the elements of the multipath component follow independent zero-mean unit-variance complex Gaussian distribution, and the path gain is The path loss factor is a B,k = 4.0, and the distance between the base station and the kth user is d B,k .

[0070] The simulation software of the specific embodiments includes Python 3.8.12, Numpy 1.21.2, and Scipy 1.7.1.

[0071] Figure 2and Figure 3 The normalized mean square error (NMSE) of statistical channel state information under different user coordinates is given, where Figure 2 The user's coordinates are (x,2,0)m, x∈[20,25]. Figure 3 The user's coordinates are (23, y, 0)m, y∈[-2.5, 2.5]. The normalized mean square error is defined as:

[0072]

[0073]

[0074]

[0075] The true value of the statistical channel state information is... and The estimated value of statistical channel state information is χ. k κ k and a R,k .Depend on Figure 2 and Figure 3 It can be seen that the normalized mean square error of the statistical channel state information is all below 0.1, indicating that the accuracy of the estimated values ​​is relatively high.

[0076] Figure 4 The impact of different reconfigurable smart surface configurations on system performance is presented, where the system performance index is defined as follows:

[0077]

[0078] Where j = J = 1, meaning the three users are grouped together. This performance metric represents the lowest downlink average receive rate among user groups. When the estimated value χ is based on statistical channel state information... k k k and a R,k Determine the reconfigurable smart surface configuration φ j At that time, φ j The specific value of φ is affected by the randomness of the wireless channel. j If r is a random variable, then min It is also a random variable. Figure 4 The black line describes the system performance index r when configuring reconfigurable smart surfaces based on statistical channel state information estimates. min The cumulative probability density function is given by the dashed line, which describes the system performance index r when configuring reconfigurable smart surfaces based on the true values ​​of statistical channel state information. min The cumulative probability density function, the dotted line describes the function using the default configuration φ. j When = 1, the system performance index r minThe cumulative probability density function of the estimated value can be found that the system performance of the configuration based on the estimated value is lower than that of the configuration based on the real value, but is still better than the default configuration, which shows the effectiveness and superiority of the technical solutions of the present application.

[0079] REFERENCES

[0080] [1] CUI T J, QI M Q, WAN X, et al. Coding metamaterials, digital metamaterials and programmable metamaterials [J]. Light: Science & Applications, 2014, 3 (10): e218-e218.

[0081] [2] TANG W, CHEN M Z, CHEN X, et al. Wireless Communications with Reconfigurable Intelligent Surface: Path Loss Modeling and Experimental Measurement [J]. IEEE Transactions on Wireless Communications, 2021, 20 (1): 421-439.

[0082] [3] DAI L, WANG B, WANG M, et al. Reconfigurable Intelligent Surface-Based Wireless Communications: Antenna Design, Prototyping, and Experimental Results [J]. IEEE Access, 2020, 8: 45913-45923.

[0083] [4] LIASKOS C, NIE S, TSIOLIARIDOU A, et al. A New Wireless Communication Paradigm through Software-Controlled Metasurfaces [J]. IEEE Communications Magazine, 2018, 56 (9): 162-169.

[0084] [5] DUNNA M, ZHANG C, SIEVENPIPER D, et al. Scatter MIMO: enabling virtual MIMO with smart surfaces; proceedings of the Proceedings of the 26th Annual International Conference on Mobile Computing and Networking, London, UK, F, 2020 [C]. Association for Computing Machinery.

[0085] [6] MISHRA D, JOHANSSON H. Channel Estimation and Low-complexity Beamforming Design for Passive Intelligent Surface Assisted MISO Wireless Energy Transfer; proceedings of the ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Brighton, UK, F 12-17 May 2019, 2019 [C]. IEEE.

[0086] [7] JENSEN T L, CARVALHO E D. An Optimal Channel Estimation Scheme for Intelligent Reflecting Surfaces Based on a Minimum Variance Unbiased Estimator; proceedings of the ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Barcelona, Spain, F 4-8 May 2020, 2020 [C]. IEEE.

[0087] [8] HE Z, YUAN X. Cascaded Channel Estimation for Large Intelligent Metasurface Assisted Massive MIMO [J]. IEEE Wireless Communications Letters, 2020, 9(2): 210-214.

[0088] [9] YUAN X, ZHANG Y J A, SHI Y, et al. Reconfigurable-Intelligent-Surface Empowered Wireless Communications: Challenges and Opportunities [J]. IEEE Wireless Communications, 2021, 28(2): 136-43.

[0089]

[10] KUNDU N K, MCKAY M R. Large Intelligent Surfaces with Channel Estimation Overhead: Achievable Rate and Optimal Configuration [J]. IEEE Wireless Communications Letters, 2021, 10(5): 986-990.

[0090]

[11] AKDIM N, MANCHóN C N, BENJILLALI M, et al. Variational Hierarchical Posterior Matching for mmWave Wireless Channels Online Learning; proceedings of the 2020 IEEE 21st International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), Atlanta, Georgia, USA, F 26-29 May 2020, 2020 [C]. IEEE.

[0091]

[12] QI C, CHEN K, DOBRE O A, et al. Hierarchical codebook-based multiuser beam training for millimeter wave massive MIMO[J]. IEEE Transactions on Wireless Communications, 2020, 19(12): 8142-8152.

Claims

1. A reconfigurable smart surface configuration method based on statistical channel state information, applicable to the following system model, wherein, The base station has N B There are N antennas serving K single-antenna users; the base station uses time-division duplex mode to distinguish between uplink and downlink signals, and utilizes the reciprocity of uplink and downlink channels to directly calculate the downlink channel state information using the uplink channel state information; a reconfigurable smart surface is located between the base station and the users, and has N... R There are N reconfigurable smart surface units; the reconfigurable smart surface is a rectangular plane divided into a μ-axis and a v-axis, and the number of units along the μ-axis of the reconfigurable smart surface is N. R,μ The number of elements along the v-axis of the reconfigurable smart surface is N. R,v Satisfying N R =N R,μ N R,v ; Focus on the direct path of the wireless channel; in the downlink channel, the direct path from the base station to the reconfigurable smart surface is... in a B,R The modulus is N B a R,B The modulus is N R The locations of the base station and the reconfigurable smart surface are fixed, and the base station can know the direct path a in advance. B,R and a R,B The direct path from the reconfigurable smart surface to the k-th user is: Where a R,k The modulus is N R The reconfigurable smart surface is divided into μ-axis and v-axis, reaching the radial direction. in a R,k,μ The modulus is N R,μ a R,k,v The modulus is N R,v , The Kronecker product of matrices; The direct path a is approximated using codewords from a binary tree-based hierarchical codebook. R,k,μ and a R,k,v The base station has a depth of L μ The hierarchical codebook, the lth... μ Layer, nth μ The code is Among them l μ ∈{0,1,…,L μ }, The base station also has a depth of L v The hierarchical codebook, the lth... v Layer, nth v The code is Among them l v ∈{0,1,…,L v }, In the scenario of reconfigurable intelligent surface-assisted wireless communication, by virtue of the reciprocity of the uplink and downlink channels in time division duplex, an improved variational hierarchical posterior matching method and parameter estimation method are used to obtain the estimated value of the statistical channel state information during the uplink process, and the reconfigurable intelligent surface configuration is obtained by using the estimated value of the statistical channel state and an optimization algorithm during the downlink process to improve the downlink reception rate of users; The specific steps are as follows: Step 1: The base station numbers the users, and the number k ∈ {1, 2,..., K}, where K is the number of users; the base station schedules the users for uplink communication in turn in a time division manner; the initial user number is k ← 1; Step 2: User k continuously transmits uplink pilot signals; the base station receives the uplink pilot signals passing through the reconfigurable smart surface, changes the configuration of the reconfigurable smart surface according to the uplink pilot signals, and estimates the statistical channel state information X. k k k and in N is a complex number. R For the number of reconfigurable smart surface units, a R,k The modulus is N R ; Step 3: The user number k is incremented; if k < K, repeat Step 2 and Step 3; Step 4: The base station groups users; the number of groups is J, and the groups are... Grouping The number of users in the middle is K j Grouping The user ID is k, k∈{1,2,…,K} j }; For grouping For any two users p and q, satisfying Where H is the conjugate symmetric operator; Step 5: Base station calculates user packets Reconfigurable smart surface configuration Where j∈{1,2,…,J}; Step 6: The base station uses a time-division multiplexing method to sequentially connect with different groups. Users in the group conduct downlink communication; the initial user group number is j←1; Step 7: The base station configures the reconfigurable smart surface as φ j and grouped with users Users in the middle conduct downlink communication; Step 8: Increment the user group number j; if the statistical channel state information χ k κ k and a R,k If the situation changes, repeat steps 1 to 8; if j ≥ J, repeat steps 6 to 8; otherwise, repeat steps 7 and 8. Step 2 involves estimating the statistical channel state information X of user k. k κ k and a R,k The specific steps are as follows: Step 2.1: Base station usage depth is L μ and L v The two hierarchical codebooks correspond to a respectively. R,k,μ and a R,k,v ; Step 2.2: The base station obtains the location prior of user k; for the μ axis, the base station selects from the hierarchical codebook that satisfy the location prior. The prior probability is greater than 0.99 (l μ ,n μ ), denoted as Similarly, the base station selects from the hierarchical codebook that meet the location prior based on the location. The prior probability is greater than 0.99 (l v ,n v ), denoted as Step 2.3: The base station sets the observation time t ← 0, sets the intermediate variable τ ← 1, and the intermediate variable s ← 1; Step 2.4: The base station increments the observation time t; Step 2.5: The base station according to and The configuration of a computationally reconfigurable smart surface is... Where ⊙ is the bitwise multiplication operator and * is the conjugate operator. This is the downlink channel vector when a reconfigurable smart surface receives signals from a base station. Step 2.6: User k transmits uplink pilot. Where T ul Given the pilot length, the magnitude of the uplink pilot x is equal to p. ul T ul p ul The power of a single uplink pilot symbol; Step 2.7: The base station receives the signal as follows: Where N B This represents the number of base station antennas; use x. H Right-multiply by Y, use Multiply by Y on the left to get Where H is the conjugate transpose operator. The uplink channel vector when the base station receives signals from the reconfigurable smart surface; Step 2.8: The base station uses the observation {y} τ ,y τ+1 ,…,y t } and reconfigurable smart surface configuration {φ τ ,φ τ+1 ,…,φ t Using an improved variational hierarchical posterior matching method, the following calculations are performed. and Step 2.9: If Three consecutive observation times and If the condition remains unchanged, proceed to step 2.10; otherwise, repeat steps 2.4 through 2.

9. Step 2.10: Repeat Steps 2.4 to 2.7 five times; Step 2.11: The base station according to and Estimating statistical channel state information Step 2.12: The base station, based on the observations {y1, y2, ..., y...} t } and reconfigurable smart surface configuration {φ1,φ2,…,φ t }, calculate intermediate variables and in: Re-estimate the statistical channel state information X k and k k They are respectively: Where, N B N is the number of base station antennas. R It is the number of reconfigurable smart surface units. It is the uplink base station received noise power, p dl,k It is the downlink base station transmit power. It is the downlink user k received noise power; For the improved variational hierarchical posterior matching method adopted by the base station in Step 2.8, the specific steps are as follows: Step 2.8.1: If s = 1, jump to Step 2.8.2; if s = 2, jump to Step 2.8.5; Step 2.8.2: Use the observation {y} τ ,y τ+1 ,…,y t } and reconfigurable smart surface configuration {φ τ ,φ τ+1 ,…,φ t Using variational inference, calculate the result that satisfies... posterior probability Based on posterior probability Calculate the posterior probability of each codeword in the hierarchical codebook. in Step 2.8.3: Step 2.8.4: If there are 3 consecutive observation times Keeping the same, s←2, τ←t+1, exit step 2.8; Step 2.8.5: Use the observation {y} τ ,y τ+1 ,…,y t } and reconfigurable smart surface configuration {φ τ ,φ τ+1 ,…,φ t Using variational inference, calculate the result that satisfies... posterior probability Based on posterior probability Calculate the posterior probability of each codeword in the hierarchical codebook. in 2. The reconfigurable smart surface configuration method according to claim 1, characterized in that, The calculation described in step 5 is for user groups. Reconfigurable smart surface configuration φ j The specific steps are as follows: Step 5.1: Initialize iteration number i = 0, intermediate variables intermediate variables intermediate variable γ (0) =0, where K is a real number. j User groups Number of users; adjust parameters Where N B Where N is the number of base station antennas, and ψ is the digamma function; since N B -K j +1∈{1,2,…,N B }, The possible values ​​are finite and can be obtained in advance; Step 5.2: The iteration number i is incremented; Step 5.3: Calculate intermediate variables Step 5.4: Introduce auxiliary variable γ and intermediate variable α (i) For the solution to the following convex optimization problem: This optimization problem can be transformed into a second-order cone problem, and the intermediate variable α can be obtained using the interior-point method. (i) ; Step 5.5: Calculate intermediate variables Step 5.6: If |γ (i) -γ (i-1) If |≥∈, repeat steps 5.2 to 5.6, where ∈ represents the iteration error; Step 5.7: Reconfigurable smart surface configuration φ j for Where ⊙ is the bitwise multiplication operator, * is the conjugate operator, and a R,B This is the downlink channel vector for receiving base station signals on a reconfigurable smart surface. round is the rounding operator. The first to N1th elements are a R,1 The first to N1th elements, The N1+1 to N1+N2 elements are a R,2 The N1+1 to N1+N2 elements... The Up to the Nth R Element is The Up to the Nth R element.

Images