Channel estimation and user positioning method based on cylindrical array intelligent metasurface

By adopting intelligent metasurface and multi-carrier multi-beam training methods of cylindrical arrays in 6G wireless systems, combined with tensor completion algorithm, the problem of insufficient signal reception capability of RIS reflective unit is solved, and high-precision channel estimation and user Positioning has expanded the application scenarios of RIS.

CN118984172APending Publication Date: 2024-11-19SOUTHEAST UNIV
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411110153.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-14
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

In the 6G wireless system, the intelligent metasurface (RIS) reflection unit lacks signal reception or processing capabilities, resulting in a large training overhead for channel estimation methods and is unable to effectively support user positioning and environment perception. In addition, traditional two-dimensional plane arrays are used to three-dimensional space. The environment capture capability is insufficient, and channel blur is introduced.

Method used

Design an intelligent metasurface based on cylindrical array, adopt multi-carrier multi-beam training method and tensor completion algorithm, and realize high-precision acquisition and environmental perception of channel state information through the topology and channel modeling of cylindrical arrays to avoid common The shape shading effect supports the precise recovery of three-dimensional spatial signal parameters and user positioning.

Benefits of technology

It realizes channel state information acquisition, transmission data detection and environment perception with low hardware cost and training overhead, operates in parallel, supports high-precision user positioning, breaks through the fuzzy problem of channel estimation in planar arrays, and expands the application of RIS in 6G wireless systems. Scene.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118984172A_ABST
    Figure CN118984172A_ABST
Patent Text Reader

Abstract

The invention relates to a channel estimation and user positioning method based on a cylindrical array intelligent metasurface. The method specifically comprises the following steps: designing a physical structure and electromagnetic response of a cylindrical array intelligent metasurface (RIS), and establishing a communication sensing integrated system framework and a channel model; designing a channel estimation beam training method, and establishing a tensor sampling signal model of array element domain-beam domain manifold transformation; designing a standard multivariate tensor completion method to recover a beam domain signal, and obtaining an array element domain signal by adopting manifold inverse transformation; designing a visual field detection method based on a window sliding strategy, and determining effective data in the field to extract channel equivalent coupling parameters; and designing an equivalent parameter decoupling method combining array manifold and spatial propagation characteristics, and obtaining accurate estimation of actual parameters of a channel so as to realize user perception positioning. According to the method, the unambiguous channel recovery which cannot be realized by the traditional planar array RIS can be realized with greatly reduced hardware cost and training overhead, so that high-precision communication perception integrated application is supported.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of wireless communication and sensing technology, and in particular to a channel estimation and user positioning method based on a cylindrical array intelligent metasurface. Background Art

[0002] The connotation of the next generation mobile communication system (6G) will go far beyond the scope of communication, integrating communication, perception, and computing capabilities. On the basis of the three major application scenarios of 5G, 6G will add two new application scenarios: artificial intelligence and perception. Communication and perception integration is one of the research hotspots of 6G's potential key technologies. Its design concept is to enable wireless communication and wireless perception to be realized in the same system and to be mutually beneficial: the communication system reuses spectrum, hardware or signal processing modules to complete multiple types of perception services; the perception results are used to assist communication access or management to improve service quality and communication efficiency.

[0003] Facing the massive growth of information interaction density and emerging business needs of 6G, the technical evolution route of "trading scale for gain" faces development bottlenecks in cost, energy consumption and complexity. In addition to traditional enhanced wireless air interface technology, the industry is also actively exploring new physical dimensions to achieve revolutionary breakthroughs in information transmission methods. Intelligent metasurface technology (RIS) has become one of the pillars of enabling key capabilities for integrated communication and perception with its low overhead, programmability and easy deployment advantages. RIS uses new sub-wavelength metamaterials to actively control electromagnetic waves through digital coding to form an electromagnetic field with controllable amplitude, phase, polarization and frequency, to achieve propagation direction control, signal enhancement or interference suppression, thereby improving multiplexing gain, combating multipath fading and enhancing system coverage, and building a new paradigm of intelligent programmable wireless environment.

[0004] The challenges and difficulties faced by intelligent metasurface technology mainly include physical models and design of metasurface materials, channel modeling, channel state information acquisition, beamforming design, passive information transmission and AI-enabled design. The RIS reflection unit does not have the ability to receive or process signals, and the training overhead of traditional channel estimation methods in large-scale ISAC systems is too large. Conventional two-dimensional planar arrays are insufficient in capturing three-dimensional spatial environmental information, introduce inherent channel estimation ambiguity, and cannot support services such as user positioning and environmental perception. In addition, in order to adapt to various practical application scenarios, RIS needs to be deployed on the surface of a conformal carrier to form a curvature array manifold, which has a conformal shielding effect, limiting the application of signal processing and data analysis methods. Summary of the invention

[0005] Technical problem: The purpose of this invention is to provide a channel estimation and user positioning method based on cylindrical array intelligent metasurface, which integrates RIS-assisted communication and perception in a systematic way, greatly expanding the application scenarios of RIS in 6G wireless systems. It can realize the parallel operation of the three functions of channel state information acquisition, transmission data detection and environmental perception with low hardware cost and training overhead, avoid estimation ambiguity to support high-precision user positioning applications.

[0006] Technical solution:

[0007] S1. Cylindrical array design and system channel modeling

[0008] The cylindrical array RIS is composed of a total of N reflection units, which are distributed as N along the cylindrical topology. V uniform circular arrays of radius R, each containing N H reflection units, satisfying N H N V = N, the subscripts H and V represent the horizontal and vertical directions respectively. All the circular arrays are arranged parallel to the z-axis of the Cartesian coordinate system. V The nth circle H The elevation of the unit is The azimuth is where d V is the unit spacing of the vertical uniform linear array. Assume that the incident azimuth angle of the far-field plane electromagnetic wave signal is φ, the zenith angle is θ, and its direction vector is [cosφsinθ, sinφsinθ, cosθ] T ,in(·) T Indicates transposition, corresponding to the nth V The nth circle H The phase shift at each unit is split into

[0009]

[0010] in, and Represents the phase shift components in the horizontal and vertical directions, k0=2π / λ c The basic wavelength is λ c The electromagnetic wave number. Horizontal phase shift Equivalent to a two-dimensional phase shift and The amplitude at this reflector unit is determined by the horizontal RIS power radiation pattern.

[0011]

[0012] Where ε≥0 represents the power radiation factor. Formula (2) shows that the power radiation level is inversely proportional to the angle between the signal direction and the normal direction of the RIS unit. max(0,x) means that when the angle between a certain signal direction and the normal direction of the RIS unit exceeds π / 2, that is, when it is facing away from the reflective unit, the unit is blocked by the curved surface of the carrier and cannot receive or reflect the signal. This shielding effect is an inherent feature of the cylindrical array RIS deployed on a conformal carrier platform that is different from the traditional planar array RIS. The complete RIS cylindrical array manifold is expressed as

[0013]

[0014] in, and are the horizontal circular array and vertical linear array manifolds, represents the complex domain space of dimension n, e x represents an exponential function with the natural constant e as the base, represents the imaginary unit, represents the Kronecker product.

[0015] The RIS-assisted communication perception integrated system includes U users with single antenna configurations, and the broadband system is divided into K subcarriers with a basic frequency of f c , the interval frequency is f s The user-RIS channel is dominated by the line-of-sight path, so the channel coefficient of the u-th user on the k-th subcarrier is modeled as

[0016]

[0017] Among them, f k =f c +kf s / K represents the subcarrier k frequency, α u and τ u denote the fading and delay of the line-of-sight path, φ u and θ u are the azimuth and zenith angles of the line-of-sight path on the RIS side. According to the free space loss characteristics, the path fading is modeled as α u =(4πf c τ u ) -1 .

[0018] S2. Beam training method and sampling tensor modeling

[0019] Channel estimation uses a multi-carrier multi-beam training method, using T time frames and K subcarriers, and configuring the RIS reflection coefficient vector in the tth training frame Each user sends a pilot symbol {x u,k}, then the beam domain signal received at the u-th user is

[0020]

[0021] Where I = U-1 represents the number of equivalent paths, and index i is uniquely mapped from the binary index (u, u′). eq,i,k =h u,k #h u′,k represents the cascade channel link, # represents the Hadamard product, α eq,i =α u α u′ and τ eq,i =τ u +τ u′ denote the equivalent attenuation and equivalent delay respectively, φ eq,i ={φ u ,φ u′} and θ eq,i ={θ u ,θ u′} represent the binary azimuth and zenith angle respectively. The equivalent pilot symbol χ i,k According to the bijective mapping In the set {x u′,k}. a(φ eq,i ,θ eq,i )=a(φ u ,θ u )#a(φ u′ ,θ u′ ), w u,k,t Represents noise. All users have the same status in the network. Without loss of generality, the user index subscript u is omitted below.

[0022] The RIS reflection coefficient vector structure is designed as in and Represent the beamforming subvectors of the horizontal circular array and the vertical linear array respectively. Construct the array element domain-beam domain transformation matrix in and Contains beamforming vector groups and Then the complete signal flow y in the beam domain is k =[y k,1 ,...,y k,N ] T Written as

[0023]

[0024] Among them, a H (φ eq,i ,θ eq,i )=aH (φ u ,θ u )#a H (φ u′ ,θ u′ ), a V (θ eq,i )=a V (θ u )#a V (θ u′ ), and Contains vector groups {a H (φ eq,i ,θ eq,i )} and {a V (θ eq,i )}. Denotes the equivalent gain, g k =[g 1,k ,...,g I,k ] T .w k =[w k,1 ,...,w k,T ] T represents the noise vector. The beam domain signal flow of K subcarriers is combined have to

[0025]

[0026] in, Formula (7) is equivalent to the third-order tensor The mode-3 matrix expansion, where the position Store received signal Y fits the standard multivariate model

[0027]

[0028] Among them, g i =[g i,1 ,...,g i,K ] T , Represents the unit core tensor, where [I 3,I ] i,i,i =1, the rest are 0, Represents the combined noise tensor. The symbol represents the vector outer product, × n Represents a tensor-matrix product of mode-n. Assuming binary indexing Where N H ={1,...,N H},N V ={1,...,N V}, whose cardinality is |N H×V |=T. The sampling position of the tth time frame is (n H (t H ),n V (t V ), the sampling signals corresponding to equations (7) and (8) are and Where n(t H ,t V )=(n H (t H )-1)N V +n V (t V ), then when When is a single shot, the training signal of T time frames is equivalent to the mode-3 fiber sampling of the beam domain tensor Y at T indices.

[0029] S3. Tensor completion method and array element signal recovery

[0030] Based on the conformal shielding effect of the cylindrical array RIS, each signal component is H (φ eq,i ,θ eq,i ) will produce a continuous visual area V eq,i , is the reflection direction {φ u ,θ u} and the incident direction {φ u′ ,θ u′}Corresponding to the intersection of the visible domain

[0031]

[0032] Among them, V u Indicates the visual range corresponding to the u-th user's viewing distance direction, The unknown visual domain is a subset of the entire unit set, that is, There are inherent and unpredictable deviations between beam-domain training signals and ideal values, and traditional subspace estimation methods are not feasible. It directly acts on the array element domain manifold and is hidden in the beam domain array manifold after RIS beamforming. A channel estimation method based on the tensor completion algorithm is designed to restore the actual array response signal data under the action of the visual domain.

[0033] Setting the mask tensor in The rest are zero, then the compressed sensing sampling in T time frames of equation (8) is equivalent to the standard multivariate tensor completion problem of fiber sampling

[0034]

[0035] Among them, ||·|| F represents the Frobenius norm.

[0036] Define a binary selection vector in lexicographic order The number of combinations is

[0037]

[0038] Where 1≤n H,1 <n H,2 ≤N H , 1≤n V,1 <n V,2 ≤N V . Define a bipartite graph P (i) , the vertex set is N H and N V , the edge set is Then the fiber sampling tensor completion problem of formula (10) has the following uniqueness condition:

[0039] Tensor The factor matrix is According to the index (n H (t H ),n V (t V )) Sample a portion of the mode-3 fiber vectors and design the selection vector d sel and the bipartite graph P (i) ,like

[0040]

[0041] Then the rank of Y is equal to I, and its standard multivariate decomposition is unique, limited to scaling and permutation blurs. rank(·) represents the rank of the matrix, diag(·) represents the diagonal matrix, Indicates that it contains A second-order composite matrix of the determinant of the 2×2 submatrices in lexicographic order.

[0042] When the above conditions are met, the beam domain tensor is completed using iterative least squares or algebraic subspace algorithms. And obtain the factor matrix estimate Only scale and displacement blurs are included. Perform inverse transformation from beam domain to element domain to restore the RIS array manifold vector group There is no conformal shielding effect on the linear array manifold in the vertical direction. According to equation (1), Directly estimate the equivalent directional parameters

[0043]

[0044] Among them, arg(·) represents the phase, (·) H represents the conjugate transpose.

[0045] S4. Visible area detection and equivalent parameter extraction

[0046] The sliding window method is designed to detect the horizontal circular array manifold visual area to avoid the influence of conformal masking effect. The basic principle is: based on the rule that the response power of the array element in the visual area is usually greater than the response power of the array element outside the visual area, this method tries to find an area to cover the smallest possible range. The main power contribution of is used as the visual area estimate. With global visual area V eq,i The relationship is The element distribution is divided into two cases: covering the middle of the array manifold, that is, falling within the range of [0,2π], forming a visual domain; or covering the head and tail of the array manifold, that is, spanning the intersection of 0 and 2π, forming two visual domains. The window sliding strategies designed for these two horizontal visual domains are as follows:

[0047] For a single viewshed, the range The window represents the area, adjust and n H Search the complete array manifold. Maximize the accumulated power in the region Equivalent to minimizing the accumulated power outside the area Minimizing the size of the region is to minimize Then the objective function is written as

[0048]

[0049] Among them, γ represents the control parameter, which is used to balance the power outside the visible area and the size of the visible area. Assuming that the azimuth difference between any two user sight paths does not exceed Δφ, the visible area size of the corresponding equivalent signal component, that is, the set V eq,i The cardinality limit is

[0050]

[0051] in, and denotes rounding down and rounding up respectively. The window search process is expressed as the following optimization problem

[0052]

[0053] For the dual viewfield case, the range The window represents the area. In contrast to the single viewport case, the accumulated power in the area is maximized. Equivalent to minimizing the accumulated power outside the area Minimizing the size of the region is to minimize Then the objective function is written as

[0054]

[0055] The window search process is expressed as the following optimization problem

[0056]

[0057] Solve the optimization problems of (16) and (18) to obtain two sets of visual domain boundary estimates: and Comparison of objective functions and Finally, the actual visual area is determined to meet the distribution characteristics corresponding to the smaller value, and the visual area set is restored according to the corresponding boundary value.

[0058] The RIS array manifold in the visible domain contains effective signal propagation information, and its phase reflects the signal response characteristics. satisfy Calculate the coordinate difference of array elements Sum signal phase difference Assume that The group index pairs meet the above conditions, and the corresponding coordinate differences and phase differences are merged into and Equivalent Direction Parameter and The least squares method is used to estimate

[0059]

[0060] in, represents the Moore–Penrose generalized inverse.

[0061] S5. Channel parameter decoupling and user-aware positioning

[0062] The equivalent signal vector Decomposed into two unit modulus vectors in any plane, the unique solution of which is two deviations angle , where ||·||2 represents the 2-norm. Since there are countless vectors through The decoupling result of a single equivalent signal direction is not unique. Construct the height for the axis The bottom radius is For a cone, the countless pairs of generatrixes with a length of 1 that are symmetrical about the axis are all decoupled feasible solutions, and the coordinates conform to the circle equation

[0063]

[0064] in, Representation and Any two unit norm vectors are orthogonal. For any value of β, the vector set are a set of solutions for the incident and reflected vectors. Given two equivalent signal vectors The circle equation (20) will generally intersect at two points. The cone generatrix pointing to these two intersection points corresponds to two feasible reflection vector solutions, and the generatrix symmetric about the cone axis corresponds to a feasible incident vector solution. Any pair of The existence of a dual solution for {β1,β2}

[0065]

[0066] in, Representation and Given three or more equivalent signal vectors, the new circle equation must intersect the first two circle equations at one of the two dual points to ensure the existence of a solution. When I ≥ 3 or U ≥ 4, the estimated equivalent signal vectors are uniquely decoupled. Based on the incident and reflected unit vectors obtained by decoupling, the estimated values ​​of the azimuth and zenith angles are directly solved.

[0067] According to the estimated results of the angle parameters of the incident and reflected paths, the signal response vector of the cylindrical array RIS is reconstructed, that is, and with dimension 1,2 factor matrix Divide the corresponding column elements to determine the scaling fuzzy coefficients for tensor completion and decomposition Then the dimension 3 factor matrix The scaling blur coefficient is According to the correction factor matrix The amplitude and phase information of the recovered equivalent delay and gain parameters are estimated as

[0068]

[0069] When the receiving user does not store the pilot sequence information of the transmitting user, equation (22) cannot be directly applied, and it is necessary to simultaneously recover the transmission symbol data and estimate the equivalent delay and gain parameters. Assuming that the transmission symbol is generated from the constant envelope constellation diagram of phase shift keying modulation, the following optimization problem is constructed to describe the detection and estimation process:

[0070]

[0071] Among them, j ∈Υ represents the modulation symbol phase, and Υ represents the phase shift keying phase set. Represents the transmission symbol χ i,k The phase estimate of . The equivalent gain on all training subcarriers is combined to solve equation (23) to reduce the estimation error and jointly recover and The constraint of equation (23) ensures that the delay phase shift will not move any symbol exactly to other constellation points. When the user sends a non-constant envelope modulation symbol, the detection energy level to recover data information.

[0072] Based on the free space loss characteristics, an algebraic method is used to decouple the delay and gain information of the incident and reflected signals.

[0073]

[0074] Since the cascade channel transmits and receives the user line-of-sight path with symmetry, there is ambiguity in the parameter pairing of the incident and reflected signals. By analyzing and clustering the output results of formula (24), the binary value repeated I times is determined as The remaining I different binary values ​​are judged as

[0075] According to the angle and delay parameter information of each user's line-of-sight path to RIS, the user's position is estimated by the azimuth-range positioning method.

[0076]

[0077] in, is the prior physical position of the cylindrical array RIS, v c is the propagation speed of electromagnetic waves.

[0078] The channel estimation and user positioning method based on cylindrical array RIS of the present invention is summarized as Algorithm 1. This algorithm mainly adopts algebraic operations and does not involve operations such as random initialization and loop iteration. It can obtain high-precision channel estimation, symbol detection and user positioning performance with low computational complexity and operation stability.

[0079]

[0080]

[0081] Beneficial effects: Compared with the prior art, the advantages of the present invention are:

[0082] 1) The present invention designs a topological paradigm of cylindrical array RIS, which supports the reflection and propagation of electromagnetic wave signals in the three-dimensional full space domain, and greatly expands the half-space domain supported by the planar array RIS.

[0083] 2) The present invention designs a channel estimation method based on cylindrical array RIS, adopts field of view detection to avoid the influence of shielding effect, supports accurate recovery and uniqueness decoupling of three-dimensional spatial signal parameters, and achieves a qualitative breakthrough compared with planar array RIS with estimation ambiguity.

[0084] 3) The present invention designs a beam domain reconstruction method based on standard multivariate tensor completion, which effectively reduces the channel estimation pilot overhead of large-scale RIS-assisted wireless systems while ensuring good channel estimation accuracy.

[0085] 4) The present invention designs a parallel method of channel estimation, symbol detection and user positioning based on RIS, realizes the integration of communication perception based on RIS, and effectively expands the application scenarios of RIS in wireless systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 The present invention is a flowchart of a method for implementing a channel estimation and user positioning based on a cylindrical array RIS according to an embodiment of the present invention.

[0087] Figure 2 It is a schematic diagram of an integrated system for wireless communication perception assisted by a cylindrical array RIS according to an embodiment of the present invention.

[0088] Figure 3 This is a relationship diagram between the tensor signal completion accuracy and the signal-to-noise ratio based on the cylindrical array RIS according to an embodiment of the present invention.

[0089] Figure 4 This is a relationship diagram between the channel angle decoupling accuracy and the signal-to-noise ratio based on the cylindrical array RIS according to an embodiment of the present invention.

[0090] Figure 5 This is a relationship diagram between the transmission symbol detection accuracy and the signal-to-noise ratio based on the cylindrical array RIS according to an embodiment of the present invention.

[0091] Figure 6 This is a relationship diagram between the positioning accuracy and signal-to-noise ratio of a mobile user based on a cylindrical array RIS according to an embodiment of the present invention. DETAILED DESCRIPTION

[0092] The following will be described in detail with reference to the relevant drawings of the present invention and in combination with the embodiments, so as to more clearly and in detail illustrate the purpose, technical solutions and outstanding advantages of the present invention.

[0093] like Figure 1 As shown, the channel estimation and user positioning method based on cylindrical array RIS provided by the embodiment of the present invention includes the following steps. It should be emphasized that, under the premise of meeting the operating requirements of the method of the present invention, the specific numerical values ​​of the embodiment do not limit the scope of application of the present invention.

[0094] S1. Cylindrical array design and system channel modeling

[0095] The embodiment of the present invention designs a cylindrical array RIS composed of N = 1024 passive reflection units, which are distributed as N along the cylindrical topology architecture. V = 16 with radius R = 1.8λ c uniform circular array, each of which contains N H = 64 reflection units, satisfying N H N V =N, the subscripts H and V represent the horizontal and vertical directions respectively, where λ c =0.01m is the basic wavelength of the system. All circular arrays are arranged in parallel with the z-axis of the Cartesian coordinate system as the axis. V The nth circle H The elevation of the unit is The azimuth is where d V =λ c / 4 = 0.25 cm is the unit spacing of the vertical uniform linear array, Δδ = 2π / N H =5.625° is the unit angular spacing of the horizontal uniform circular array. Assume that the incident azimuth angle of the far-field plane electromagnetic wave signal is φ, the zenith angle is θ, and its direction vector is [cosφsinθ, sinφsinθ, cosθ] T ,in(·) T Indicates transposition, corresponding to the nth V The nth circle H The phase shift at each unit is split into

[0096]

[0097] in, and They represent the phase shift components of the signal response in the horizontal and vertical directions along the array manifold, respectively, k0 = 2π / λ c =200πrad / m with a basic wavelength of λ c The electromagnetic wave number. Horizontal phase shift Equivalent to a two-dimensional phase shift and The amplitude at this reflector unit is determined by the horizontal RIS power radiation pattern.

[0098]

[0099] Among them, the power radiation factor is taken as ε = 0.1. Formula (2) shows that the power radiation level is inversely proportional to the angle between the signal direction and the normal direction of the RIS unit. max(0,x) means that when the angle between a certain signal direction and the normal direction of the RIS unit exceeds π / 2, that is, when it is facing away from the reflective unit, the unit is blocked by the curved surface of the carrier and will not be able to receive or reflect the signal. This shielding effect is an inherent feature of the cylindrical array RIS deployed on a conformal carrier platform that is different from the traditional planar array RIS. The complete RIS cylindrical array manifold is expressed as

[0100]

[0101] in, and are the horizontal circular array and vertical linear array manifolds, represents the complex domain space of dimension n, e x represents an exponential function with the natural constant e as the base, represents the imaginary unit, represents the Kronecker product.

[0102] like Figure 2 As shown, the embodiment of the present invention designs a RIS-assisted communication perception integrated system, including U = 4 users with single antenna configuration, the broadband system is divided into K = 1024 subcarriers, and the basic frequency is f c =30GHz, the spacing frequency is f s = 0.32GHz. The user-RIS channel is dominated by the line-of-sight path, so the channel coefficient of the u-th user on the k-th subcarrier is modeled as

[0103]

[0104] Among them, f k =f c +kf s / K represents the subcarrier k frequency, α u and τ u denote the fading and delay of the line-of-sight path, φ u and θ u are the azimuth and zenith angles of the line-of-sight path on the RIS side. According to the free space loss characteristics, the path fading is modeled as α u =(4πf c τ u ) -1 According to the distribution of wireless facilities in the embodiment of the present invention, the azimuth angle and the zenith angle are randomly generated in the range of [0°, 360°] and [0°, 180°] respectively, the delay is randomly generated in the range of (0ns, 50ns] (ns: nanoseconds), and the user position is determined by the azimuth-distance positioning method.

[0105] S2. Beam training method and sampling tensor modeling

[0106] Channel estimation uses a multi-carrier multi-beam training method, using time frames and K = 1024 subcarriers, where κ = {0.15, 0.25, 0.35} represents the sampling ratio of the complete beam domain, and the corresponding pilot length is T = {154, 256, 359}. In the embodiment of the present invention, the RIS reflection coefficient vector is configured in the tth training frame Select from the discrete Fourier transform (DFT) matrices that satisfy Where δ(x) represents the Dirac delta function. Each user sends a pilot symbol {x u,k}, selected from the orthogonal phase shift keying coded symbols, while satisfying the constant envelope constraint |x u,k |=1 and the phase constellation constraint arg(x u,k )∈{45°,135°,225°,315°}, where |·| and arg(·) represent the amplitude and phase respectively. The u-th user is taken as the reference user, and the received beam domain signal is

[0107]

[0108] Wherein, I=U-1=3 represents the number of equivalent paths, and the index i is obtained by uniquely mapping the binary index (u,u′). In the embodiment of the present invention, u=1,u′∈{2,...,U},i=u′-1. eq,i,k =h u,k #h u′,k represents the cascade channel link, # represents the Hadamard product, α eq,i =α u α u′ and τ eq,i =τ u +τ u′ denote the equivalent attenuation and equivalent delay of the cascade path, respectively, φ eq,i ={φ u ,φ u′} and θ eq,i ={θ u ,θ u′} represent the binary azimuth and zenith angle respectively. The equivalent pilot symbol χ i,k According to the bijective mapping In the set {x u′,k}. a(φ eq,i ,θ eq,i )=a(φ u ,θ u )#a(φ u′ ,θ u′ ), wu,k,t represents additive white Gaussian noise, and the noise power set in the embodiment of the present invention is Satisfy system signal-to-noise ratio constraints All users have the same status in the network. Without loss of generality, the user index subscript u is omitted below.

[0109] The embodiment of the present invention designs the RIS reflection coefficient vector to have the structure in and Represent the beamforming vectors of the horizontal circular array and the vertical linear array respectively. Construct the array element domain-beam domain transformation matrix in and Contains beamforming vector groups and The embodiment of the present invention adopts N H Point and N V Point DFT matrix construction and Unitary matrix properties Complete signal flow in beam domain k =[y k,1 ,...,y k,N ] T Written as

[0110]

[0111] Among them, a H (φ eq,i ,θ eq,i )=a H (φ u ,θ u )#a H (φ u′ ,θ u′ ), a V (θ eq,i )=a V (θ u )#a V (θ u′ ), and Contains vector groups {a H (φ eq,i ,θ eq,i )} and {a V (θ eq,i )}. Denotes the equivalent gain, g k =[g 1,k ,...,g I,k ] T , w k =[w k,1,...,w k,T ] T represents the noise vector. The beam domain signal flow of K subcarriers is combined have to

[0112]

[0113] in, Formula (7) is equivalent to the third-order tensor The mode-3 matrix expansion, where the position Store received signal Y fits the standard multivariate model

[0114]

[0115] Among them, g i =[g i,1 ,...,g i,K ] T , Represents the unit core tensor, where [I 3,I ] i,i,i =1, the rest of the positions are 0, Represents the merged noise tensor, symbol represents the vector outer product, × n Represents a tensor-matrix product of mode-n. Assuming binary indexing Where N H ={1,...,N H},N V ={1,...,N V}, whose cardinality is |N H×V |=T. The sampling position of the tth time frame is (n H (t H ),n V (t V ), the sampling signals corresponding to equations (7) and (8) are and Where n(t H ,t V )=(n H (t H )-1)N V +n V (t V ), then when When it is a single shot, the training signal of T time frames is equivalent to the mode-3 fiber sampling of the beam domain tensor Y at T indexes. The specific sampling mode of the embodiment of the present invention is determined according to the tensor completion uniqueness condition analysis in step S3.

[0116] S3. Tensor completion method and array element signal recovery

[0117] Based on the conformal shielding effect of the cylindrical array RIS, each signal component is H (φ eq,i ,θ eq,i ) will produce a continuous visual area V eq,i , is the reflection direction {φ u ,θ u} and the incident direction {φ u′ ,θ u′}Corresponding to the intersection of the visible domain

[0118]

[0119] Among them, V u Indicates the visual range corresponding to the u-th user's viewing distance direction, Since the unknown visual domain is a proper subset of the entire unit set, that is, There are inherent and unpredictable deviations between beam-domain training signals and ideal values, and traditional subspace estimation methods are not feasible. It directly acts on the array element domain manifold and is hidden in the beam domain array manifold after RIS beamforming. A channel estimation method based on the tensor completion algorithm is designed to restore the actual array response signal data under the action of the visual domain.

[0120] Setting the mask tensor in The rest are zero, then the compressed sensing sampling in T time frames of equation (8) is equivalent to the standard multivariate tensor completion problem of fiber sampling

[0121]

[0122] Among them, ||·|| F represents the Frobenius norm.

[0123] Define a binary selection vector in lexicographic order The number of combinations is

[0124]

[0125] Where 1≤n H,1 <n H,2 ≤N H , 1≤n V,1 <n V,2 ≤N V . Define a bipartite graph P (i) , the vertex set is N H and N V , the edge set is Then the fiber sampling tensor completion problem of formula (10) has the following uniqueness condition:

[0126] Tensor The factor matrix is According to the index (n H (t H ),n V (t V )) Sample a portion of the mode-3 fiber vectors and design the selection vector d sel and the bipartite graph P (i) ,like

[0127]

[0128] Then the rank of Y is equal to I, and its standard multivariate decomposition is unique, limited to scaling and permutation blurs. rank(·) represents the rank of the matrix, diag(·) represents the diagonal matrix, Indicates that it contains A second-order composite matrix of the determinant of the 2×2 submatrices in lexicographic order.

[0129] In the embodiment of the present invention, all channel path parameters are generated according to the continuous probability density function. In this case, the conditions for the first two terms of equation (12) to be true are equivalent to K ≥ I and Where ||·||0 represents the zero norm. The necessary condition to ensure the third connected graph is to sample P (i) Each row and column is at least once. When the uniqueness condition is met, the beam domain tensor signal is completed using iterative least squares or algebraic subspace algorithms. Obtain estimates of the factor matrices of each dimension Only scale and displacement blurs are included. Perform inverse transformation from beam domain to element domain to restore the RIS array manifold vector group There is no conformal shielding effect on the linear array manifold in the vertical direction. According to equation (1), Directly estimate the equivalent directional parameters

[0130]

[0131] Among them, arg(·) represents the phase, (·) H represents the conjugate transpose.

[0132] S4. Visible area detection and equivalent parameter extraction

[0133] The sliding window method is designed to detect the horizontal circular array manifold visual field to avoid the influence of conformal masking effect. The basic principle is to find a small area coverage based on the rule that the response power of the array element in the visual field is usually greater than the response power of the array element outside the visual field. The main power contribution of . Horizontal field of view With global visual area V eq,i The relationship is The element distribution can be divided into two cases: covering the middle of the array manifold, that is, falling within the range of [0,2π], forming a visual domain; or covering the head and tail of the array manifold, that is, spanning the intersection of 0 and 2π, forming two visual domains. The window sliding strategies for designing these two horizontal visual domains are as follows:

[0134] For a single visual field, the range is The window represents the area, adjust and n H Search the complete array manifold. Maximize the accumulated power in the region Equivalent to minimizing the accumulated power outside the area Minimizing the size of the region is to minimize Then the objective function can be written as

[0135]

[0136] Wherein, γ represents a control parameter, which is used to balance the power outside the visual area and the size of the visual area. In the embodiment of the present invention, the average value of the response power inside and outside the visual area is used to determine γ: Sort by absolute value in ascending order: Pick And calculate and set up The parameter μ = 0.8. Assuming that the azimuth difference between any two user sight paths does not exceed Δφ, the visual area scale of the corresponding equivalent signal component, that is, the set V eq,i The cardinality limit is

[0137]

[0138] in, and In the embodiment of the present invention, Δφ=120° is set, that is, L min =10,L max =32. The window search process is expressed as the following optimization problem

[0139]

[0140] For the dual viewfield case, the range is The window represents the area. In contrast to the single viewport case, the accumulated power in the area is maximized. Equivalent to minimizing the accumulated power outside the area Minimizing the size of the region is to minimize Then the objective function is written as

[0141]

[0142] The window search process is expressed as the following optimization problem

[0143]

[0144] Solve the optimization problems of (16) and (18) to obtain two sets of visual domain boundary estimates: and Comparison of objective functions and Finally, the actual visual area is determined to meet the distribution characteristics corresponding to the smaller value, and the visual area set is restored according to the corresponding boundary value. Since the feasible domain of one-dimensional search is limited, the embodiment of the present invention adopts a point-by-point search strategy to obtain the global optimal solution with controllable complexity.

[0145] The RIS array manifold in the visible domain contains effective signal propagation information, and its phase reflects the signal response characteristics. satisfy Calculate the coordinate difference of array elements Sum signal phase difference Assume that The group index pairs meet the above conditions, and the corresponding coordinate differences and phase differences are merged into and Equivalent Direction Parameter and The least squares method is used to estimate

[0146]

[0147] in, Denotes the Moore-Penrose generalized inverse. The embodiment of the present invention adjusts the parameters to ensure that D i ≥2 to obtain a unique solution.

[0148] S5. Channel parameter decoupling and user-aware positioning

[0149] The equivalent signal vector Decomposed into two unit modulus vectors in any plane, the unique solution of which is two deviations angle , where ||·||2 represents the 2-norm. Since there are countless vectors through The decoupling result of a single equivalent signal direction is not unique. Construct the height for the axis The bottom radius is For a cone, the countless pairs of generatrixes with a length of 1 that are symmetrical about the axis are all decoupled feasible solutions, and the coordinates conform to the circle equation

[0150]

[0151] in, Representation and Any two unit norm vectors are orthogonal. For any value of β, the vector set are a set of solutions for the incident and reflected vectors. Given two equivalent signal vectors The circle equation (20) will generally intersect at two points. The cone generatrix pointing to these two intersection points corresponds to two feasible reflection vector solutions, and the generatrix symmetric about the cone axis corresponds to a feasible incident vector solution. Any pair of The existence of a dual solution for {β1,β2}

[0152]

[0153] in, Representation and Given three or more equivalent signal vectors, the new circle equation must intersect the first two circle equations at one of the two dual points to ensure the existence of a solution. When I ≥ 3 or U ≥ 4, the estimated equivalent signal vectors are uniquely decoupled. Based on the incident and reflected unit vectors obtained by decoupling, the estimated values ​​of the azimuth and zenith angles are directly solved. The embodiment of the present invention adopts the Levenberg-Marquardt algorithm based on nonlinear least squares to solve all β values, and then calculates the signal direction vector corresponding to the cone generatrix and its azimuth and zenith angle.

[0154] According to the estimated results of the angle parameters of the incident and reflected paths, the signal response vector of the cylindrical array RIS is reconstructed, that is, and with dimension 1,2 factor matrix Divide the corresponding column elements to determine the scaling fuzzy coefficients for tensor completion and decomposition Then the dimension 3 factor matrix The scaling blur coefficient is According to the correction factor matrix The equivalent delay and gain parameters of the amplitude and phase information estimation are:

[0155]

[0156] When the receiving user does not store the pilot sequence information of the transmitting user, equation (22) cannot be directly applied, and it is necessary to simultaneously recover the transmission symbol data and estimate the equivalent delay and gain parameters. Assuming that the transmission symbol is generated from the constant envelope constellation diagram of phase shift keying modulation, the following optimization problem is constructed to describe the detection and estimation process:

[0157]

[0158] Among them, j ∈Υ represents the modulation symbol phase, and Υ represents the phase shift keying phase set. Represents the transmission symbol χ i,k The phase estimate of . The equivalent gain on all training subcarriers is combined to solve equation (23) to reduce the estimation error and jointly recover and The constraint condition of formula (23) ensures that the delay phase shift will not move any symbol exactly to other constellation points. In the embodiment of the present invention, Y={45°, 135°, 225°, 315°} is set, and a nonlinear programming algorithm based on the interior point method is used to simultaneously solve When the user sends a non-constant envelope modulation symbol, the detection energy level to recover data information.

[0159] Based on the free space loss characteristics, an algebraic method is used to decouple the delay and gain information of the incident and reflected signals.

[0160]

[0161] Since the cascade channel transmits and receives the user line-of-sight path with symmetry, there is ambiguity in the parameter pairing of the incident and reflected signals. By analyzing and clustering the output results of formula (24), the binary value repeated I times is determined as The remaining I different binary values ​​are judged as

[0162] According to the angle and delay parameter information of each user's line-of-sight path to RIS, the user's position is estimated by the azimuth-range positioning method.

[0163]

[0164] in, is the prior physical position of the cylindrical array RIS, v c is the propagation speed of electromagnetic waves.

[0165] Based on the above system parameters and scheme design, the embodiment of the present invention repeatedly runs 2×10 4Monte Carlo simulations are performed to test the performance of the channel estimation and user positioning method based on cylindrical array RIS. The performance of tensor signal completion, path angle decoupling, transmission symbol detection and mobile user positioning obtained by the channel estimation method under different sampling rates κ = T / N varies with the system signal-to-noise ratio (SNR). Figure 3 , Figure 4 , Figure 5 and Figure 6 As shown, the accuracy of tensor signal completion is measured by the standard mean square error (NMSE), the accuracy of path angle decoupling and mobile user positioning is measured by the root mean square error (RMSE), and the accuracy of transmission symbol detection is measured by the empirical cumulative distribution function (ECDF). It can be seen from the simulation results that the channel estimation method provided by the embodiment of the present invention can obtain accurate parameter recovery and decoupling performance, and can especially obtain angle estimation accuracy below 0.1° under the condition of T<<N, thereby supporting 10 -4 The transmission symbol detection accuracy is on the order of 1.0 cm and the mobile user positioning accuracy is on the order of 1.0 cm.

[0166] In summary, the channel estimation and user positioning method based on cylindrical array RIS provided by the embodiment of the present invention analyzes the three-dimensional spatial channel characteristics contained in the RIS array manifold by modeling, fully explores the potential of the three-dimensional RIS topology structure to perceive the wireless propagation environment, and breaks through the estimation ambiguity problem of passive RIS channel estimation. The embodiment of the present invention constructs a standard multivariate beam domain tensor, and uses the tensor signal completion algorithm to significantly reduce the training pilot overhead. In addition, the embodiment of the present invention uses the visual field detection method to overcome the signal loss and distortion problems caused by the conformal shielding effect, enhances the deployment flexibility of the RIS device, and expands the application scenarios of the RIS technology.

Claims

1. A channel estimation and user positioning method based on cylindrical array intelligent metasurface, characterized in that: The method comprises the following steps: Step S1, designing the cylindrical array RIS unit layout and response characteristics, and constructing the RIS-assisted communication perception integrated system framework and channel model; Step S2, designing a channel estimation beam training method with low pilot overhead, and constructing a tensor sampling signal model based on array element domain-beam domain manifold transformation; Step S3, designing a standard multivariate tensor completion method for fiber sampling, restoring the complete beam domain tensor signal, and obtaining the array element domain signal response through manifold inverse transformation; Step S4, designing a horizontal visual range detection method based on a window sliding strategy, determining the effective data of the array element domain signal in the visual range, and extracting the channel equivalent coupling parameter; Step S5, designing an equivalent parameter decoupling method based on array manifold and spatial propagation characteristics to obtain an accurate estimate of the actual channel parameters and realize user positioning perception application.

2. A method for channel estimation and user positioning based on cylindrical array intelligent metasurface according to claim 1, characterized in that: The step S1 specifically includes: The cylindrical array RIS is composed of a total of N reflection units, which are distributed as N along the cylindrical topology. V uniform circular arrays of radius R, each containing N H reflection units, satisfying N H N V = N, the subscripts H and V represent the horizontal and vertical directions respectively; all the circular arrays are arranged parallel to the z-axis of the Cartesian coordinate system. V The nth circle H The elevation of the unit is The azimuth is where d V is the unit spacing of the vertical uniform linear array; let the incident azimuth angle of the far-field plane electromagnetic wave signal be φ, the zenith angle be θ, and its direction vector be [cosφsinθ, sinφsinθ, cosθ] T ,in(·) T Indicates transposition, the nth V The nth circle H The phase shift at each unit is split into in, and Represents the phase shift components in the horizontal and vertical directions, k0=2π / λ c The basic wavelength is λ c The electromagnetic wave number; horizontal phase shift Equivalent to a two-dimensional phase shift and The amplitude here is determined by the horizontal RIS power radiation pattern Among them, ε≥0 represents the power radiation factor; max(0,x) means that when the angle between a certain signal direction and the normal direction of the RIS unit exceeds π / 2, that is, when it is facing away from the reflective unit, the unit is blocked by the curved surface of the carrier and will not be able to receive or reflect the signal; the complete RIS cylindrical array manifold is expressed as in, and are the array manifolds of horizontal circular array and vertical linear array respectively, represents the complex domain space of dimension n, e x represents an exponential function with the natural constant e as the base, represents the imaginary unit, represents the Kronecker product; The RIS-assisted communication perception integrated system includes U users with single antenna configurations, and the broadband system is divided into K subcarriers with a basic frequency of f c , the interval frequency is f s ; The user-RIS channel is dominated by the line-of-sight path, so the channel coefficient of the u-th user on the k-th subcarrier is modeled as Among them, f k =f c +kf s / K represents the subcarrier k frequency, α u and τ u denote the fading and delay of the line-of-sight path, φ u and θ u are the azimuth and zenith angles of the line-of-sight path on the RIS side; according to the free space loss characteristics, the path fading is modeled as α u =(4πf c τ u ) -1 .

3. A channel estimation and user positioning method based on cylindrical array intelligent metasurface as claimed in claim 2, characterized in that: The step S2 specifically includes: Channel estimation uses a multi-carrier multi-beam training method, using T time frames and K subcarriers, and configuring the RIS reflection coefficient in the tth training frame. Each user sends a pilot symbol {x u,k }, then the signal received by the uth user is Where I = U-1 represents the number of equivalent paths, and the index i is uniquely mapped from the binary index (u, u′); h eq,i,k =h u,k #h u′,k represents the cascade channel link, # represents the Hadamard product, α eq,i =α u α u′ and τ eq,i =τ u +τ u′ denote the equivalent attenuation and equivalent delay respectively, φ eq,i ={φ u ,φ u′ } and θ eq,i ={θ u ,θ u′ } respectively represent the binary azimuth and zenith angle; the equivalent pilot symbol χ i,k According to the bijective mapping In the set {x u′,k }; a(φ eq,i ,θ eq,i )=a(φ u ,θ u )#a(φ u′ ,θ u′ );w u,k,t represents noise; all users have the same status in the network, and without loss of generality, the user index subscript u is omitted below; The RIS reflection coefficient vector structure is designed as in and Represent the beamforming subvectors of the horizontal circular array and the vertical linear array respectively; construct the array element domain-beam domain transformation matrix in and Contains beamforming vector groups and Then the complete signal flow y in the beam domain is k =[y k,1 ,...,y k,N ] T Written as Among them, a H (φ eq,i ,θ eq,i )=a H (φ u ,θ u )#a H (φ u′ ,θ u′ ), a V (θ eq,i )=a V (θ u )#a V (θ u′ ), and Contains vector groups {a H (φ eq,i ,θ eq,i )} and {a V (θ eq,i )}; Denotes the equivalent gain, g k =[g 1,k ,...,g I,k ] T ;w k =[w k,1 ,...,w k,T ] T represents the noise vector; the beam domain signal flow of K subcarriers is combined have to in, Formula (7) is equivalent to the third-order tensor The mode-3 matrix expansion, where the position Store received signal Y fits the standard multivariate model in, Represents the unit core tensor, where [I 3,I ] i,i,i =1, the rest of the positions are 0, represents the combined noise tensor; the symbol represents the vector outer product, × n Tensor-matrix product representing mode-n; assumes binary indexing Where N H ={1,...,N H },N V ={1,...,N V }, whose cardinality is |N H×V |=T; let the sampling position of the tth time frame be (n H (t H ),n V (t V ), the sampling signals corresponding to equations (7) and (8) are and Where n(t H ,t V )=(n H (t H )-1)N V +n V (t V ), then when When is a single shot, the training signal of T time frames is equivalent to the mode-3 fiber sampling of the beam domain tensor Y at T indices.

4. A method for channel estimation and user positioning based on cylindrical array intelligent metasurface as claimed in claim 3, characterized in that: The step S3 specifically includes: Based on the conformal shielding effect of the cylindrical array RIS, each signal component is H (φ eq,i ,θ eq,i ) will produce a continuous visual area V eq,i , is the reflection direction {φ u ,θ u } and the incident direction {φ u′ ,θ u′ }Corresponding to the intersection of the visible domain Among them, V u Indicates the visual range corresponding to the u-th user's viewing distance direction, Design a standard multivariate tensor completion method to restore the complete beam domain array signal response, and obtain the actual array signal response data in the visual domain through the beam domain-array element domain inverse transformation; Setting the mask tensor in The rest are zero, then the compressed sensing sampling in T time frames of equation (8) is equivalent to the standard multivariate tensor completion problem of fiber sampling Among them, ||·|| F represents the Frobenius norm; Define a binary selection vector in lexicographic order The number of combinations is Where 1≤n H,1 <n H,2 ≤N H , 1≤n V,1 <n V,2 ≤N V ; Define the bipartite graph P (i) , the vertex set is N H and N V , the edge set is Then the fiber sampling tensor completion problem of formula (10) has the following uniqueness condition: Tensor The factor matrix is According to the index (n H (t H ),n V (t V )) Sample a portion of the mode-3 fiber vectors and select vector d sel and the bipartite graph P (i) ,like Then the rank of Y is equal to I, and its standard multivariate decomposition is unique, limited to scaling and permutation fuzziness; rank(·) represents the rank of the matrix, diag(·) represents the diagonal matrix, Indicates that it contains The second-order composite matrix of the determinant of the 2×2 submatrices in lexicographic order; When the above conditions are met, the beam domain tensor Y^ is completed using iterative least squares or algebraic subspace algorithms, and the factor matrix estimate is obtained It only includes scale and displacement blur; calculation Perform inverse transformation from beam domain to element domain to restore the RIS array manifold vector group There is no conformal shielding effect on the linear array manifold in the vertical direction. According to equation (1), Directly estimate the equivalent directional parameters Among them, arg(·) represents the phase, (·) H represents the conjugate transpose.

5. A channel estimation and user positioning method based on cylindrical array intelligent metasurface as claimed in claim 4, characterized in that: The step S4 specifically includes: Design a sliding window method to detect the horizontal array manifold visual area, horizontal visual area With global visual area V eq,i The relationship is The element distribution is divided into two cases: covering the middle of the array manifold, that is, falling within the range of [0,2π], forming a visual domain; or covering the head and tail of the array manifold, that is, spanning the intersection of 0 and 2π, forming two visual domains. The window sliding strategies for the two horizontal visual domains are designed as follows: For a single viewshed, the range The window represents the area, adjust and Search the complete array manifold; maximize the accumulated power in the region Equivalent to minimizing the accumulated power outside the area Minimizing the size of the region is to minimize Then the objective function is written as Where γ represents the control parameter, which is used to balance the power outside the visible area and the size of the visible area. Assuming that the azimuth difference between any two user sight paths does not exceed Δφ, the visible area size of the corresponding equivalent signal component, that is, the set V eq,i The cardinality limit is in, and Represents rounding down and rounding up respectively; the window search process is expressed as the following optimization problem For the dual viewfield case, the range The window represents the area; in contrast to the single view area, the accumulated power in the area is maximized Equivalent to minimizing the accumulated power outside the area Minimizing the size of the region is to minimize Then the objective function is written as The window search process is expressed as the following optimization problem Solve the optimization problems of (16) and (18) to obtain two sets of visual domain boundary estimates: and Comparison of objective functions and Finally, the actual visual area is determined to meet the distribution characteristics corresponding to the smaller value, and the visual area set is restored according to the corresponding boundary value. The RIS array manifold in the visible domain contains effective signal propagation information, and its phase reflects the signal response characteristics; for any satisfy Calculate the coordinate difference of array elements Sum signal phase difference Assume that The group index pairs meet the above conditions, and the corresponding coordinate differences and phase differences are merged into and Equivalent Direction Parameter and The least squares method is used to estimate in, represents the Moore–Penrose generalized inverse.

6. A method for channel estimation and user positioning based on cylindrical array intelligent metasurface according to claim 5, characterized in that: The step S5 specifically includes: The equivalent signal vector Decomposed into two unit modulus vectors in any plane, the unique solution of which is two deviations angle symmetric vector, where ||·||2 represents the 2-norm; since there are countless The decoupling result of a single equivalent signal direction is not unique. Construct the height for the axis The bottom radius is For a cone, the countless pairs of generatrixes with a length of 1 that are symmetrical about the axis are all decoupled feasible solutions, and the coordinates conform to the circle equation in, Representation and Any two unit norm vectors are orthogonal; for any value of β, the vector set are a set of solutions for the incident and reflected vectors; given two equivalent signal vectors The circle equation (20) will generally intersect at two points. The cone generatrix pointing to these two intersection points corresponds to two feasible reflection vector solutions, and its generatrix symmetric about the cone axis corresponds to a feasible incident vector solution; any pair of The existence of a dual solution for {β1,β2} in, Representation and vectors that are orthogonal to each other; given three or more equivalent signal vectors, the new circle equation must intersect with the previous two circle equations at one of the two dual points to ensure the existence of a solution; when I ≥ 3 or U ≥ 4, the estimated equivalent signal vectors are uniquely decoupled; based on the incident and reflected unit vectors obtained by decoupling, the estimated values ​​of the azimuth and zenith angles are directly solved According to the estimated results of the angle parameters of the incident and reflected paths, the signal response vector of the cylindrical array RIS is reconstructed, that is, and with dimension 1,2 factor matrix Divide the corresponding column elements to determine the scaling fuzzy coefficients for tensor completion and decomposition Then the dimension 3 factor matrix The scaling blur coefficient is According to the correction factor matrix The equivalent delay and gain parameters of the amplitude and phase information estimation are: When the receiving user does not store the pilot sequence information of the sending user, equation (22) cannot be directly applied. It is necessary to simultaneously recover the transmission symbol data and estimate the equivalent delay and gain parameters. Assuming that the transmission symbol is generated from the constant envelope constellation diagram of phase shift keying modulation, the following optimization problem is constructed to describe the detection and estimation process. Among them, j ∈Υ represents the modulation symbol phase, Υ represents the phase shift keying phase set; Represents the transmission symbol χ i,k The phase estimate of all training subcarriers is combined to solve equation (23) to reduce the estimation error and jointly recover and The constraint of formula (23) ensures that the delay phase shift will not move any symbol exactly to other constellation points; when the user sends a non-constant envelope modulation symbol, the detection energy level to recover data information; Based on the free space loss characteristics, an algebraic method is used to decouple the delay and gain information of the incident and reflected signals. Since the cascade channel transmitting and receiving user line-of-sight paths are symmetrical, there is ambiguity in the parameter pairing of the incident and reflected signals. By analyzing and clustering the output results of formula (24), the binary value repeated I times is determined as The remaining I different binary values ​​are judged as According to the angle and delay parameter information of each user's line-of-sight path to RIS, the user's position is estimated by the azimuth-range positioning method. in, is the prior physical position of the cylindrical array RIS, v c is the propagation speed of electromagnetic waves.

Citation Information

Cited By

  • Reconfigurable metasurface beam generation method based on branch and bound method

    CN121441358A