Near-field beam training method of super-large scale array based on energy diffusion effect

Abstract

The invention provides a near-field beam training method and device for a super-large-scale array based on an energy diffusion effect, a storage medium and electronic equipment, and relates to the technical field of near-field communication. The method can be applied to three array structures of ULA, UCA and UPA. The method comprises the following steps: firstly, constructing a downlink frequency division duplex communication system, and uniformly modeling near-field super-large-scale channels of three array structures into a function related to a wave path difference; and secondly, a new codebook design criterion is formulated, and an optimal focus point is obtained by maximizing an overlapping range of a near-field region and a beam direction and utilizing an energy diffusion effect, so that the gain of the beam in a coverage blind area is improved. And finally, based on the codebook design criterion, obtaining a two-stage codebook structure and a corresponding beam training scheme. The designed near-field codebook structure is applied to a corresponding beam training scheme, the training overhead can be balanced, and a blind area in an angle training stage can be effectively repaired, so that the beam training precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of near-field communication, and particularly relates to a near-field beam training method, device, storage medium, and electronic device for an extremely large-scale array based on the energy diffusion effect. Background Art

[0002] The extremely large aperture array (ELAA) introduces a new communication method, namely near-field communication. Users are likely to be located in the near-field region, where spherical wave propagation needs to be considered, and beam training requires searching both the angle and distance dimensions simultaneously. Therefore, it is necessary to redesign the near-field codebook and consider a new beam training method.

[0003] In the related art, the beam training process is achieved by searching for the optimal beam from a predefined codebook. Currently, the main methods for beam training in near-field communication are two-stage beam training. In the first stage of two-stage beam training, the orthogonal discrete Fourier transform (DFT) codebook is used for angle estimation and an optimal angle candidate set is constructed. In the second stage, distance search is performed on the optimal angle candidate set to determine the optimal codeword.

[0004] In addition, in the ELAA system, the performance of beam training also depends on the configuration of the antenna array. Among them, the uniform linear antenna array (ULA) is the simplest and most widely used array configuration. In fact, when the incident angle of the ULA is large, the near-field region will shrink, making it possible that the user equipment in this region may not receive the training signal. The uniform circular antenna array (UCA) has a unique rotational symmetry, which magnifies the visible near-field region at large angles. Therefore, the UCA can enable more near-field users to receive training signals. Compared with the one-dimensional ULA and UCA, the uniform planar array (UPA) can be regarded as a two-dimensional array, which can support a larger number of radiating antenna elements under the same array aperture, meaning that the UPA provides more adjustable design freedoms.

[0005] Due to near-field beam focusing, multiple codewords are required to achieve near-field coverage in any angular direction, and the training overhead for obtaining the optimal angle of the user cannot be ignored. However, in the traditional two-stage beam training scheme, using a far-field codebook for angle training will result in the phenomenon of coverage holes (blind zones). When the user is in the near-field blind zone, this phenomenon will lead to a low training success rate. Summary of the Invention

[0006] (1) Technical Problems to be Solved

[0007] In view of the deficiencies of the prior art, the present invention provides a near-field beam training method, device, storage medium, and electronic device for a very large-scale array based on the energy diffusion effect, which solves the technical problems of large training overhead and low training success rate in the traditional two-stage beam training scheme.

[0008] (2) Technical solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0010] A near-field beam training method for a very large-scale array based on the energy diffusion effect, including:

[0011] Construct a downlink frequency-division duplex communication system, and uniformly model the near-field very large-scale channels of the three array structures of ULA, UCA, and UPA as functions related to the wave path difference;

[0012] Formulate a codebook design criterion for maximizing the near-field distance coverage; including:

[0013] By adjusting the focusing distance, maximize the overlapping range of the near-field communication area and the distance dimension in any given angular direction;

[0014] Based on the energy diffusion effect of the beam, and combined with the beam pattern corresponding to the near-field very large-scale channel, determine the closed-form solution of the achievable boundary covered by the beam to determine the optimal focusing distance of the three array structures in any given angular direction;

[0015] Based on the codebook design criterion, obtain a two-stage codebook structure applicable to the three array structures and the corresponding beam training scheme.

[0016] Preferably, in the downlink frequency-division duplex communication system:

[0017] The base station BS is an ELAA with N antennas, the user UE is an antenna with 1 antenna, the base station BS has one radio frequency RF chain, and beamforming is performed in the analog domain through a phase shifter, then the transmit beamforming vector, i.e., the codeword Each element of has a constant modulus constraint; where represents a complex matrix;

[0018] Represent the channel between the base station BS and the user UE as In the process of the downlink signal being transmitted from the base station BS to the user UE, the received signal of the user UE is expressed as:

[0019] y = h H wx + z (1)

[0020] Among them, y represents the received signal; the superscript H represents the transpose matrix; x and z respectively represent the transmitted symbol and the additive Gaussian white noise with normalized power;

[0021] In the near - field region, the near - field channel is modeled based on the spherical - wave propagation model and is expressed as:

[0022]

[0023] where β l and b(r l , θ l , φ l ) represent the complex gain of path l and the steering vector corresponding to path l, respectively; L represents the number of propagation paths; the steering vector b(r l , θ l , φ l ) is expressed as:

[0024]

[0025] where j represents the imaginary unit; λ represents the wavelength; r l , θ l and φ l represent the distance, azimuth angle, and elevation angle between the user UE or scatterer on path l and the center of the base - station antenna array, respectively; Δr n (r l , θ l , φ l ) represents the wave - path difference between r l and r n (r l , θ l , φ l ); r n (r l , θ l , φ l ) is the distance between the user UE or scatterer on path l and the nth antenna;

[0026] For the channel modeling of ULA:

[0027] Assume that the coordinates of the nth antenna of the base station are [0, τ n d], where d is the antenna spacing. Then, the distance between the user UE or scatterer located at (r l , θ l ) and the nth antenna on path l is:

[0028]

[0029] where sin and cos represent the sine and cosine functions, respectively;

[0030] Approximating (a) by the second - order Taylor - series expansion, the wave - path difference Δr n (rl , θ l ) is expressed as:

[0031]

[0032] For the modeling of the UCA channel:

[0033] Assume that the center of the UCA is the coordinate origin, and let R be the radius of the UCA, such that the coordinates of the nth antenna are where where n = 0, 1, … N - 1;

[0034] According to the geometric relationship, r in the UCA n (r l , θ l ) is expressed as:

[0035]

[0036] The corresponding path difference Δr n (r l , θ l ) is expressed as:

[0037]

[0038] For the modeling of the UPA channel:

[0039] Consider a base station configured with a two-dimensional UPA. The total number of array antennas N = N I N J , where N I is the number of antennas in the horizontal direction, and N J is the number of antennas in the vertical direction; with the center of the array as the coordinate origin, the coordinates of the (i, j)th array element are [0, τ i d, τ j d] T , where

[0040] By using the geometric properties of a triangle, r in the UPA n (r l , θ l , φ l ) is expressed as:

[0041]

[0042] The corresponding path difference Δr n (r l , θ l ) is expressed as:

[0043]

[0044] Unify the near - field very - large - scale channel of the ULA, UCA, and UPA array structures into a function related to the wave - path difference, expressed as:

[0045]

[0046] Substitute formula (10) into formula (3) to obtain the steering vector expression, and obtain the explicit expression of formula (2).

[0047] Preferably, before formulating the codebook design criterion for maximizing the near - field distance coverage, it includes:

[0048] By applying the codeword w, rewrite the received signal (1) of the user UE located at (r, θ, φ) as:

[0049]

[0050] where β represents the complex fading parameter of the channel; w(n) represents the n - th element in the codeword w, and Δr n (r, θ, φ) represents the wave - path difference between the user UE located at (r, θ, φ) and the n - th antenna at the base station BS;

[0051] Define w p as the codeword at point p (r p , θ p , φ p ), and its n - th element is written as

[0052]

[0053] Therefore, the beam pattern of the received signal (1) of the user UE located at (r, θ, φ) is calculated as:

[0054]

[0055] where |·| 2 represents the square of the absolute - value function;

[0056] Plot the beam patterns of the near - field codeword w p , and the far - field codeword respectively to clarify that the far - field codebook cannot capture the user UE in the blind area; where the blind area refers to the near - field region between the Fresnel distance R f and the minimum beam coverage distance R bf .

[0057] Preferably, when maximizing the overlapping range of the near - field communication area and the distance dimension in any given angular direction by adjusting the focusing distance, it means:

[0058] When the angles are aligned, i.e., θ = θp , φ = φ p , by adjusting the focusing distance r p , to maximize the overlapping area of the beam in the near - field region in the distance dimension, which is expressed as:

[0059]

[0060] where, is the optimal focusing distance; argmax is the variable value that maximizes the objective function; Ω0 and Ω p are the near - field communication region and the beam direction range of the codeword w p respectively, γ represents the overlapping range of the two, and the symbol ∩ represents the intersection;

[0061] Since determining the beam direction range Ω p is a non - convex problem, a 3dB main lobe is introduced, that is, for the codeword w centered at point p p to make the received power of the user UE greater than half of the coverage range:

[0062]

[0063] where, the symbol is defined as.

[0064] Preferably, the process of determining the optimal focusing distance of the three array structures in any given angular direction includes:

[0065] Define the achievable boundary including the minimum coverage distance R bf and the maximum coverage distance R br ;

[0066] For the three array structures, for the codeword w focused on any point p p , when the angular directions are aligned, the inequality (15) is further simplified by formula (13) to:

[0067]

[0068] Then, the inequality about the beam coverage distance r, R bf ≤ r ≤ R br is solved by formula (16), where the minimum and maximum beam coverage distances R bf and R br before the received power decays to half are as follows:

[0069]

[0070] where,

[0071]

[0072] For far - field codewords, they are obtained by setting the sample distance r p →∞, where the minimum covering distance R can be obtained according to the lemma bf ; Define the function as the difference between the minimum beam covering distance R bf and the Fresnel distance R at any given position (r, θ, φ) to represent the blind area along the distance dimension, specifically: bf Since the far - field codebook cannot capture the user UEs within the blind area, define the beam depth as

[0073]

[0074] Substitute it into formula (17) to get: Substitute it into formula (17) to get:

[0075]

[0076] Based on formulas (17) and (20), and aiming to maximize the overlapping range of the near - field communication area and the distance dimension in any given angular direction, when the maximum beam covering distance R br coincides with the Rayleigh distance, determine the optimal focusing distance of the three array structures in any given angular direction:

[0077]

[0078] where D represents the array aperture.

[0079] Preferably, the codebook structure and the corresponding beam training scheme in the first stage include:

[0080] Define the angular - domain codebook as Each codeword focuses on different position points, where for n points, the focusing distance in the focusing direction (θ n , φ n ) is set as Then the angular - domain codeword corresponding to the n points is written as:

[0081]

[0082] For the three array structures, the setting of the focusing direction should ensure that the overall N directions should cover all directions in the near - field region, and based on formula (20), we get:

[0083] For the ULA array structure:

[0084]

[0085] For the UCA array structure:

[0086]

[0087] Among them, is the beam width of the UCA array;

[0088] For the UPA array structure:

[0089]

[0090] Among them, arcsin is the arcsine function. For the one-dimensional arrays ULA and UCA, θ n represents the azimuth angle of the nth point, n = 1, 2, … N; for the two-dimensional array UPA, is the azimuth angle of the nth i point, represents the elevation angle of the nth j point, and n i = 1, 2, … N I , n j = 1, 2, … N J , N = N I N J ;

[0091] At this stage, the user UE selects the codeword corresponding to the maximum received signal power and takes multiple angles near the estimated angle index to form an angle candidate set.

[0092] Preferably, the codebook structure and the corresponding beam training scheme in the second stage include:

[0093] Based on the angle candidate set, adopting a non-uniform distance sampling strategy, at each given angle (θ q , φ q ), design the distance domain codebook as W q = {w q,0 , w q,1 , … w q,S-1}, S is the number of distance samplings at each angle, where each distance domain codeword is denoted as:

[0094]

[0095] Among them, Δ is the sampling threshold and is obtained based on formula (10):

[0096] For the ULA array structure:

[0097]

[0098] For the UCA array structure:

[0099]

[0100] For the UPA array structure:

[0101]

[0102] By searching all the distance domain codewords in the angle candidate set, the optimal distance domain codeword is obtained and the estimated position of the user UE

[0103] A near-field beam training device for a very large-scale array based on the energy diffusion effect, comprising:

[0104] A channel modeling module, configured to construct a downlink frequency division duplex communication system, and uniformly model the near-field very large-scale channels of three array structures, namely ULA, UCA, and UPA, as functions related to the wave path difference;

[0105] A criterion formulation module, configured to formulate a codebook design criterion for maximizing the near-field distance coverage; including:

[0106] An adjustment unit, by adjusting the focusing distance, maximizes the overlapping range of the near-field communication area and the distance dimension in any given angular direction;

[0107] A determination unit, based on the energy diffusion effect of the beam and in combination with the beam pattern corresponding to the near-field very large-scale channel, determines a closed-form solution of the achievable boundary of the beam coverage to determine the optimal focusing distance of the three array structures in any given angular direction;

[0108] A scheme acquisition module, configured to obtain a two-stage codebook structure applicable to the three array structures and a corresponding beam training scheme based on the codebook design criterion.

[0109] A storage medium stores a computer program for near-field beam training of a very large-scale array based on the energy diffusion effect, wherein the computer program enables a computer to control the near-field beam training method as described above.

[0110] An electronic device, comprising:

[0111] One or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs include those for controlling the near-field beam training method as described above

[0112] (III) Beneficial effects

[0113] The present invention provides a near-field beam training method, device, storage medium, and electronic device for a very large-scale array based on the energy diffusion effect. Compared with the prior art, the following beneficial effects are achieved:

[0114] The present invention can be applied to three array structures: ULA, UCA, and UPA. First, a downlink frequency division duplex communication system is constructed, and the near-field very large-scale channels of the three array structures are uniformly modeled as a function related to the wave path difference. Second, a new codebook design criterion is formulated. By maximizing the overlapping range between the near-field region and the beam direction and utilizing the energy diffusion effect, the best focus point is obtained, thereby improving the gain blind area of the beam in the coverage blind area. Finally, based on this codebook design criterion, a two-stage codebook structure and the corresponding beam training scheme are obtained. Applying the designed near-field codebook structure to the corresponding beam training scheme can balance the training overhead and effectively repair the blind area in the angle training stage, thereby improving the accuracy of beam training. Description of the Drawings

[0115] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0116] Figure 1 It is a block diagram of a near-field beam training method for a very large-scale array based on the energy diffusion effect provided by an embodiment of the present invention;

[0117] Figure 2 It is a multi-panel array geometry model of three antenna array structures provided by an embodiment of the present invention;

[0118] Figure 3 It is a schematic diagram (beam pattern) of the coverage of a single far-field codeword in the near-field region provided by an embodiment of the present invention;

[0119] Figure 4 It is a schematic diagram (beam pattern) of the coverage of multiple near-field codewords in the near-field region provided by an embodiment of the present invention;

[0120] Figure 5 It is a schematic diagram of the relationship between the array aperture and the near-field blind area provided by an embodiment of the present invention;

[0121] Figure 6 It is a schematic diagram (beam pattern) of the coverage of a designed focusing distance in the near-field region provided by an embodiment of the present invention;

[0122] Figure 7 It is a schematic diagram of the relationship between the success rate and the transmission distance provided by an embodiment of the present invention. Detailed Embodiments

[0123] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0124] By providing a near-field beam training method, apparatus, storage medium, and electronic device for a very large-scale array based on the energy diffusion effect, the embodiments of the present application solve the technical problems of large training overhead and low training success rate in traditional two-stage beam training schemes.

[0125] The general idea of the technical solutions in the embodiments of the present application to solve the above technical problems is as follows:

[0126] 1. The embodiments of the present invention establish a unified model to characterize the near-field very large-scale channels of three array structures: ULA, UCA, and UPA. By introducing the concept of wave path difference, the channel is modeled as a function related to the wave path difference, where the wave path difference involves the common angle of departure (AoD) and distance between the base station BS and the user UE, as well as the array radius and elevation angle of UCA and UPA.

[0127] 2. The embodiments of the present invention reveal the near-field blind zone phenomenon of the far-field codebook and the coverage complexity of the near-field codebook in the near-field region. In view of this problem, by integrating the low overhead of the far-field codebook and the reinforcement advantage of the near-field codebook blind zone, the idea of maximizing the near-field distance coverage is proposed, and by finding the optimal focusing distance, an optimization problem of maximizing the overlapping range between the near-field communication area and the beam direction is formulated.

[0128] 3. By analyzing the energy diffusion effect of the beam, the embodiments of the present invention determine the achievable boundary of the beam coverage in closed form and obtain the optimal focusing distance in any angular direction for the three array configurations. This novel design guideline can patch the blind zone while reducing the coverage complexity.

[0129] 4. According to the obtained codebook design criteria, the embodiments of the present invention respectively develop a two-stage codebook structure and the corresponding beam training scheme to reduce the training overhead and improve the success rate. The proposed beam training scheme is applicable to the above three types of array structures.

[0130] It should be noted that the term "energy diffusion effect" in the embodiments of the present invention refers to the energy of the beam diffusing in a part of the area rather than focusing on a specific position point.

[0131] To better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the accompanying drawings of the specification and specific embodiments.

[0132] Embodiment 1:

[0133] As Figure 1 shown, an embodiment of the present invention provides a near - field beam training method for a very large - scale array based on the energy diffusion effect, including:

[0134] S1. Construct a downlink frequency - division duplex communication system, and uniformly model the near - field very large - scale channels of three array structures, namely ULA, UCA, and UPA, as functions related to the wave - path difference;

[0135] S2. Develop a codebook design criterion for maximizing the near - field distance coverage; including:

[0136] S21. By adjusting the focusing distance, maximize the overlapping range of the near - field communication area and the distance dimension in any given angular direction;

[0137] S22. Based on the energy diffusion effect of the beam and in combination with the beam pattern corresponding to the near - field very large - scale channel, determine a closed - form solution for the achievable boundary of the beam coverage to determine the optimal focusing distance of the three array structures in any given angular direction;

[0138] S3. Based on the codebook design criterion, obtain a two - stage codebook structure applicable to the three array structures and the corresponding beam training scheme.

[0139] The embodiment of the present invention applies the designed near - field codebook structure to the corresponding beam training scheme, which can balance the training overhead and effectively repair the blind area in the angle training stage, thereby improving the accuracy of beam training.

[0140] Next, each step of the above - mentioned scheme will be introduced in detail:

[0141] In step S1, construct a downlink frequency - division duplex communication system, and uniformly model the near - field very large - scale channels of three array structures, namely ULA, UCA, and UPA, as functions related to the wave - path difference.

[0142] Specifically, in the downlink frequency - division duplex communication system (Frequency Division Duplexing, FDD):

[0143] The base station BS is an ELAA with N antennas, the user UE has 1 antenna, the base station BS has one radio frequency (RF) chain, and beamforming is performed in the analog domain through a phase shifter, then each element of the transmit beamforming vector, i.e., the codeword has a constant - modulus constraint; where represents a complex matrix;

[0144] Represent the channel between the base station BS and the user UE as During the process of downlink signal transmission from the base station BS to the user UE, the received signal of the user UE is expressed as:

[0145] y = h H wx + z (1)

[0146] where y represents the received signal; the superscript H represents the transpose matrix; x and z represent the transmitted (pilot or training) symbol and the additive Gaussian white noise with normalized power, respectively.

[0147] The boundary between the far - field and near - field regions is defined as the Rayleigh distance (the Rayleigh distance is calculated based on the phase difference between the plane wavefront and the spherical wavefront reaching ). In the near - field region, the near - field channel is modeled based on the spherical wave propagation model and is expressed as:

[0148]

[0149] where β l and b(r l , θ l , φ l ) represent the complex gain of path l and the steering vector corresponding to path l, respectively; L represents the number of propagation paths; the steering vector b(r l , θ l , φ l ) is expressed as:

[0150]

[0151] where j represents the imaginary unit; λ represents the wavelength; r l , θ l and φ l represent the distance between the user UE or the scatterer and the center of the base - station antenna array, the azimuth angle, and the elevation angle on path l, respectively; Δr n (r l , θ l , φ l ) represents the wave - path difference between r l and r n (r l , θ l , φ l ); r n (r l , θ l , φ l ) is the distance between the user UE or the scatterer and the n - th antenna on path l.

[0152] Furthermore, as Figure 2 shown, Figure 2 are the schematic diagrams of the geometric models of multi - panel arrays with three different array structures: ULA, UCA, and UPA, respectively.

[0153] As shown in Figure 2 (a) of

[0154] For the channel modeling of the ULA, assume that the coordinates of the nth antenna of the base station are [0, τ n d], where d is the antenna spacing. Then, the distance between the user UE or scatterer located at (r l , θ l ) and the nth antenna on the path l is:

[0155]

[0156] where sin and cos represent the sine and cosine functions respectively;

[0157] Approximating (a) as the second-order Taylor formula expansion, the path difference Δr n (r l , θ l ) is expressed as:

[0158]

[0159] As shown in Figure 2 (b) of

[0160] For the channel modeling of the UCA, assume that the center of the UCA is the coordinate origin, and R is represented as the radius of the UCA, such that the coordinates of the nth antenna are where where n = 0, 1, … N-1;

[0161] According to the geometric relationship, r n (r l , θ l ) in the UCA is expressed as:

[0162]

[0163] The corresponding path difference Δr n (r l , θ l ) is expressed as:

[0164]

[0165] As shown in Figure 2 (c) of

[0166] Considering that the base station is configured with a two-dimensional UPA, the total number of array antennas N = N I N J , where N I is the number of antennas in the horizontal direction, and N Jis the number of vertical antennas; taking the center of the array as the coordinate origin, the coordinates of the (i,j)-th array element are [0,τ i d,τ j d] T , where

[0167] By using the geometric properties of triangles, r n (r l ,θ l ,φ l ) is expressed as:

[0168]

[0169] The corresponding path difference Δr n (r l ,θ l ) is expressed as:

[0170]

[0171] It should be noted that the differences in the channel models of the three array structures mainly come from the differences in the path differences in the steering vectors. Therefore, the near-field very large-scale channels of the three array structures, namely ULA, UCA, and UPA, are unifiedly modeled as a function related to the path difference, expressed as:

[0172]

[0173] Finally, substituting formula (10) into formula (3) to obtain the steering vector expression, and obtaining the explicit expression of formula (2).

[0174] In step S2, a codebook design criterion for maximizing the near-field distance coverage is formulated.

[0175] Generally speaking, beam training is to search for the optimal codeword w from a predefined codebook W to maximize the beam gain. Since each near-field beam only focuses the radiation energy within a limited area rather than along a specific direction, the design of the near-field beam codebook is affected by both the beam coverage range and the focusing position.

[0176] Therefore, the key point of the embodiment of the present invention is to guide a reasonable and effective codebook W, whose radiation concentration can fully cover the area where the user is located, that is, all possible positions of the user in the near-field environment.

[0177] Specifically, the embodiment of the present invention proposes a method for designing a near-field codebook by analyzing the beam coverage range and the focusing position. In practical situations, the Line of Sight (LoS) path gain is much greater than the Non-Line of Sight (NLoS) gain, especially in the millimeter-wave or terahertz frequency bands. Therefore, it is considered to find the beam with the strongest LoS path.

[0178] Correspondingly, before formulating the codebook design criterion for maximizing the near-field distance coverage, this step further includes:

[0179] By applying the codeword w, the received signal (1) of the user UE located at (r, θ, φ) is rewritten as:

[0180]

[0181] where β represents the complex fading parameter of the channel; w(n) represents the nth element in the codeword w, and Δr n (r, θ, φ) represents the path difference between the user UE located at (r, θ, φ) and the nth antenna at the base station BS;

[0182] To ensure the beam training quality, it is important to reasonably design the codeword structure w to maximize the beam gain power. First, w p is defined as the codeword at point p (r p , θ p , φ p ), and its nth element is written as

[0183]

[0184] Therefore, the beam pattern of the received signal (1) of the user UE located at (r, θ, φ) is calculated as:

[0185]

[0186] where |·| 2 represents the square of the absolute value function;

[0187] Next, the beam patterns of the near-field codeword w p , and the far-field codeword are respectively plotted to clarify that the far-field codebook cannot capture the user UE in the blind area; where the blind area refers to the near-field area between the Fresnel distance R f and the minimum beam coverage distance R bf .

[0188] Exemplarily, the DFT codeword in the near-field area and the far-field codeword are adopted, as Figures 3 - 4 shown. Figures 3 - 4Represents the normalized beam pattern of the far-field and near-field codebooks. R f ,R r ,R bf are the Fresnel distance, Rayleigh distance, and minimum beam coverage distance, respectively.

[0189] By observing Figures 3 - 4 , two physical phenomena can be found:

[0190] (1) Incomplete coverage: Since the wave path difference is inversely proportional to the BS-UE distance, the distance factor in the wave path difference dominates the beam gain in the near-field region. Therefore, when the UE is located between the Fresnel distance and the minimum beam coverage distance, the beam gain of the far-field codeword decreases significantly, as shown in Figure 3 . Therefore, this region is defined as the near-field blind area here, and the training signal is submerged in noise.

[0191] (2) Coverage complexity: Due to the beam focusing characteristics of the near-field codewords in the near-field region, the BS must use multiple near-field codewords to achieve full coverage of the entire near-field region in any angular direction, as shown in Figure 4 .

[0192] Based on the above physical phenomena, it is concluded that the far-field codebook may not be able to capture the user UE in the blind area. When using the exhaustive search scheme, the training overhead of the near-field codebook is larger than that of the far-field codebook. Fortunately, the energy diffusion effect of the far-field codebook can reduce the overhead of angular training, and the near-field codebook has the ability to repair the blind area. By combining the advantages of the far-field codebook and the near-field codebook, the embodiment of the present invention finally proposes a codebook design guideline for maximizing near-field coverage to balance the training overhead of the near-field codebook and the coverage area of the far-field codebook.

[0193] On this basis, further clarify the codebook design criteria for maximizing the near-field distance coverage in this step, including:

[0194] S21. By adjusting the focusing distance, maximize the overlapping range of the near-field communication area and the distance dimension in any given angular direction, which means:

[0195] When the angles are aligned, that is, θ = θ p , φ = φ p , by adjusting the focusing distance r p , to maximize the overlapping area of the beam in the near-field region in the distance dimension, which is expressed as:

[0196]

[0197] where is the optimal focusing distance; argmax is the value of the variable that maximizes the objective function; Ω0 and Ω pThe near - field communication area and the codeword w respectively p are the beam direction ranges, γ represents the overlapping range of the two, and the symbol ∩ represents the intersection;

[0198] Since determining the beam direction range Ω p is a non - convex problem, a 3dB main lobe is introduced, that is, the codeword w centered on point p p such that the received power by the user UE is greater than half of the coverage range:

[0199]

[0200] Among them, the symbol is defined as.

[0201] S22. Based on the beam energy diffusion effect and combined with the beam direction pattern corresponding to the near - field very large - scale channel, determine the closed - form solution of the achievable boundary of the beam coverage to determine the optimal focusing distance of the three array structures in any given angular direction, including:

[0202] Define that the achievable boundary includes the minimum coverage distance R bf and the maximum coverage distance R br ;

[0203] For the three array structures, in the codeword w focused on an arbitrary point p p , when the angular directions are aligned, the inequality (15) is further simplified by formula (13) to:

[0204]

[0205] Then, by solving the inequality about the beam coverage distance r through formula (16), R bf ≤r≤R br , where the minimum and maximum beam coverage distances R bf and R br before the received power decays to half are as follows:

[0206]

[0207] Among them,

[0208]

[0209] It is not difficult to see that R bf increases as r p increases. This means that as the beam focus p moves away from the base station BS, the minimum beam coverage distance will gradually increase.

[0210] For the far - field codeword, by setting the sample distance r p→∞, where the minimum covering distance R can be obtained according to the lemma bf ; The function Defined as the minimum beam coverage distance R bf The Fresnel distance R from any given position (r,θ,φ) bf The difference between and is used to represent the blind spot along the distance dimension, specifically:

[0211]

[0212] For example, Figure 5 As shown, Figure 5 The working frequency is 28Ghz,θ=0, When r→∞, the relationship between the near-field blind area and the antenna array aperture under different arrays. Figure 5 It can be observed that when using the far-field codebook, the blind area grows exponentially with the increase of the array aperture. Therefore, for ELAA, the blind area will become non-negligible.

[0213] In order to maximize the coverage of the beam codeword and the overlap range of the near-field area, combined with the far-field codeword explained above, the user UE in the blind area cannot be captured. First, the beam depth is defined as Substituting into formula (17) we get:

[0214]

[0215] To obtain the optimal beam focusing distance The following points can be used for analysis:

[0216] 1). For (1.20), the beam depth ΔR increases with the focusing distance r p increases with the increase of .

[0217] 2). For (1.17), the minimum / maximum beam coverage distance varies with the focusing distance r p increases with the increase of .

[0218] 3) The overlapping range of the beam patterns in the near field should be as large as possible.

[0219] In order to obtain a greater coverage depth within the communication area, r p Should be as large as possible; when the maximum beam coverage distance R br When the near field region (Rayleigh distance) is exceeded, the focusing distance r is increased. p This will result in the minimum beam coverage distance R bf Increase, then the overlap range γ between the beam pattern and the near field area will decrease. Therefore, when the maximum beam coverage distance R br When it coincides with the Rayleigh distance, the optimal solution to the problem can be obtained.

[0220] Generally speaking, based on formulas (17) and (20), and aiming to maximize the overlapping range of the near-field communication area and the distance dimension in any given angular direction, when the maximum beam coverage distance R br coincides with the Rayleigh distance, the optimal focusing distances of the three array structures in any given angular direction are determined:

[0221]

[0222] where D represents the array aperture.

[0223] In step S3, based on the codebook design criterion, a two-stage codebook structure applicable to the three array structures and the corresponding beam training scheme are obtained.

[0224] Based on the formulated codebook design criterion, the embodiments of the present invention also develop a two-stage codebook structure applicable to the foregoing three array structures of ULA, UCA, and UPA, and the corresponding beam training scheme, that is, the two-stage beam training scheme (Two-Stage Beam Training, abbreviated as TSBT). Among them: in the first stage, beam scanning is performed using the codebook structure based on the design criterion to obtain a candidate angle set; in the second stage, a search for distance-domain codewords is only performed on the candidate angles to obtain the optimal codeword.

[0225] Specifically:

[0226] In the first stage: the codebook structure and the corresponding beam training scheme in the first stage include:

[0227] Define the angle-domain codebook as Each codeword focuses on a different position point, where for the nth point, the focusing distance in the focusing direction (θ n , φ n ) is set to Then the angle-domain codeword corresponding to the nth point is written as:

[0228]

[0229] For the three array structures, the setting of the focusing direction should ensure that the overall N directions should cover all directions in the near-field region, and based on formula (20), it is obtained that:

[0230] For the ULA array structure:

[0231]

[0232] For the UCA array structure:

[0233]

[0234] Among them, is the beam width of the UCA array, which is only related to the number of antennas;

[0235] For the UPA array structure:

[0236]

[0237] Among them, θ n represents the azimuth angle of the nth point; arcsin is the arcsine function; φ n represents the elevation angle of the nth point.

[0238] Exemplarily, the normalized beam direction diagram of the designed codeword is drawn here, as Figure 6 shown. Compared with Figures 3 - 4 , the designed codeword balances the traditional far-field codebook and the near-field codebook, greatly reducing the near-field blind area generated by the far-field codebook and significantly reducing the coverage complexity of the near-field codebook.

[0239] At this stage, the user UE selects the codeword corresponding to the maximum received signal power and takes multiple angles near the estimated angle index to form an angle candidate set.

[0240] In the second stage: The codebook structure and the corresponding beam training scheme in the second stage include:

[0241] Based on the angle candidate set, adopting a non-uniform distance sampling strategy, at each given angle (θ q , φ q ), the distance-domain codebook is designed as W q ={w q,0 , w q,1 ,… w q,S-1}, S is the number of distance samplings at each angle, and each distance-domain codeword is denoted as:

[0242]

[0243] Among them, Δ is a freely adjustable sampling threshold and is obtained based on formula (10):

[0244] For the ULA array structure:

[0245]

[0246] For the UCA array structure:

[0247]

[0248] For the UPA array structure:

[0249]

[0250] By searching all the distance domain codewords within the set of angle candidates, the optimal distance domain codeword is obtained and the estimated location of the user UE.

[0251] So far, the embodiment of the present invention has completed all the processes of the near-field beam training method for a very large-scale array based on the energy diffusion effect.

[0252] To further illustrate the superiority of the embodiment of the present invention, by way of example, a uniform linear array (ULA) is taken as an example, and the number of ULAs is set to 256. In addition, the user (UE) is located at (r, θ), where the transmission distances r and θ are uniformly generated within [8m, 150m] and respectively.

[0253] As Figure 7 shown, Figure 7 shows the relationship between the success rate and the transmission distance, where the signal-to-noise ratio (SNR) is set to 20 dB. In the near-field region, the rate is not affected by the distance. In addition, as the distance between the base station and the user equipment (BS-UE) increases by about 20 m and 30 m respectively, the success rates of the proposed and traditional TSBT schemes can reach the level of the exhaustive beam training scheme. This also means that when the user UE is close to the base station, the proposed two-stage beam training (TSBT) scheme is more advantageous. This is because the energy diffusion depth of the proposed beam codeword is longer than that of the traditional far-field codewords (channel steering vectors or DFTs), thus overcoming the blind zone effect.

[0254] Embodiment 2:

[0255] The embodiment of the present invention provides a near-field beam training device for a very large-scale array based on the energy diffusion effect, including:

[0256] A channel modeling module, configured to construct a downlink frequency division duplex communication system and uniformly model the near-field very large-scale channels of three array structures, namely ULA, UCA, and UPA, as functions related to the wave path difference;

[0257] A criterion formulation module, configured to formulate a codebook design criterion for maximizing the near-field distance coverage; including:

[0258] An adjustment unit, which maximizes the overlapping range of the near-field communication area and the distance dimension in any given angular direction by adjusting the focusing distance;

[0259] A determination unit, based on the energy diffusion effect of the beam and in combination with the beam pattern corresponding to the near-field very large-scale channel, determines a closed-form solution for the achievable boundary of the beam coverage to determine the optimal focusing distance of the three array structures in any given angular direction;

[0260] A scheme acquisition module, configured to obtain a two-stage codebook structure applicable to the three array structures and a corresponding beam training scheme based on the codebook design criterion.

[0261] Embodiment 3:

[0262] An embodiment of the present invention provides a storage medium storing a computer program for near-field beam training of a very large-scale array based on the energy diffusion effect, wherein the computer program causes a computer to control the near-field beam training method as described in Embodiment 1.

[0263] Embodiment 4:

[0264] An embodiment of the present invention provides an electronic device, including:

[0265] One or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the programs include those for controlling the near-field beam training method as described in Embodiment 1.

[0266] It can be understood that the near-field beam training device, storage medium, and electronic device of a very large-scale array based on the energy diffusion effect provided by the embodiments of the present invention correspond to a near-field beam training method of a very large-scale array based on the energy diffusion effect provided by the embodiments of the present invention. For the explanations, examples, beneficial effects, etc. of the relevant content, reference can be made to the corresponding parts in the near-field beam training method, which will not be elaborated here.

[0267] In summary, compared with the prior art, the following beneficial effects are achieved:

[0268] 1. The embodiments of the present invention reveal the near-field blind zone phenomenon generated when a far-field codebook is applied to the near-field region, and give an explicit expression of the beam coverage distance.

[0269] 2. Compared with the exhaustive search scheme, the two-stage beam training scheme based on the codebook structure of the energy diffusion effect proposed by the present invention greatly reduces the training overhead.

[0270] 3. The embodiments of the present invention can be applicable to three different array structures, namely ULA, UCA, and UPA, and have a certain universality. Compared with the traditional two-stage beam training, the proposed near-field codebook structure based on the beam energy diffusion effect can be applied to the beam training scheme of the ELAA with three array structures, can balance the training overhead, and effectively repair the blind zone in the angle training stage, thereby improving the accuracy of beam training.

[0271] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0272] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A near-field beam training method for a very large-scale array based on energy diffusion effect, characterized in that: include: A downlink frequency division duplex communication system is constructed, and the near-field ultra-large-scale channels of three array structures, ULA, UCA, and UPA, are uniformly modeled as functions related to the wave path difference; Develop codebook design criteria to maximize near-field range coverage; including: By adjusting the focus distance, the overlap range of the near field communication area and the distance dimension in any given angular direction is maximized; Based on the energy diffusion effect of the beam and in combination with the beam pattern corresponding to the near-field ultra-large-scale channel, a closed-form solution of the achievable boundary of the beam coverage is determined to determine the optimal focusing distance of the three array structures in any given angular direction; Based on the codebook design criterion, a two-stage codebook structure applicable to the three array structures and a corresponding beam training scheme are obtained.

2. The near-field beam training method according to claim 1, characterized in that: In the downlink frequency division duplex communication system: The base station BS is an ELAA with N antennas, the user UE is a single antenna, and the base station BS has an RF chain. The beamforming is performed in the analog domain through a phase shifter. The transmitted beamforming vector is the codeword Each element of has a constant modulus constraint; represents a complex matrix; The channel between the base station BS and the user UE is represented as In the process of transmitting the downlink signal from the base station BS to the user UE, the received signal of the user UE is expressed as: y=h H wx+z (1) Where y represents the received signal; the superscript H represents the transposed matrix; x and z represent the additive Gaussian white noise of the transmitted symbol and normalized power, respectively; In the near field region, the near field channel is modeled based on the spherical wave propagation model and is expressed as: Among them, β l and b(r l ,θ l ,φ l ) represent the complex gain of path l and the steering vector corresponding to path l respectively; L represents the number of propagation paths; the steering vector b(r l ,θ l ,φ l ) is expressed as: Where j represents the imaginary unit; λ represents the wavelength; r l ,θ l and φ l Respectively represent the distance, azimuth and elevation between the user UE or scatterer and the center of the base station antenna array on path l; Δr n (r l ,θ l ,φ l ) represents r l and r n (r l ,θ l ,φ l ), r n (r l ,θ l ,φ l ) is the distance between the user UE or scatterer and the nth antenna on path l; For ULA channel modeling: Assume that the coordinates of the nth antenna of the base station are [0,τ n d], where d is the antenna spacing, then (r l ,θ l The distance between the user UE or scatterer at ) and the nth antenna on path l is: Among them, sin and cos represent sine and cosine functions respectively; Approximately (a) is a second-order Taylor formula expansion, then the path difference Δr n (r l ,θ l ) is expressed as: For UCA channel modeling: Assume that the center of UCA is the origin of coordinates and R is the radius of UCA, so that the coordinates of the nth antenna are in Where n = 0, 1, ... N-1; According to the geometric relationship, the r in UCA n (r l ,θ l ) is expressed as: The corresponding path difference Δr n (r l ,θ l ) is expressed as: For UPA channel modeling: Consider a base station with a two-dimensional UPA, with a total number of array antennas N = N I N J , where N I is the number of antennas in the horizontal direction, N J is the number of antennas in the vertical direction; taking the center of the array as the origin of the coordinates, the coordinates of the (i, j)th array element are [0,τ i d,τ j d] T ,in By using the geometric properties of triangles, the r in UPA n (r l ,θ l ,φ l ) is expressed as: The corresponding path difference Δr n (r l ,θ l ) is expressed as: The near-field ultra-large-scale channels of the three array structures of ULA, UCA and UPA are uniformly modeled as a function related to the wave path difference, which is expressed as: Substituting formula (10) into formula (3) yields the steering vector expression, and obtaining the explicit expression of formula (2).

3. The near-field beam training method according to claim 2, characterized in that: Before formulating the codebook design criteria for maximizing the near-field distance coverage, it includes: By applying the codeword w, the received signal (1) of the user UE located at (r, θ, φ) is rewritten as: Where β represents the complex fading parameter of the channel; w(n) represents the nth element in the codeword w, Δr n (r,θ,φ) represents the path difference between the user UE located at (r,θ,φ) and the nth antenna at the base station BS; w p Defined as point p (r p ,θ p ,φ p ), whose nth element is written as Therefore, the beam pattern of the received signal (1) of the user UE located at (r, θ, φ) is calculated as: Among them, |·| 2 represents the square of the absolute value function; Draw the near-field codeword w separately p , and the beam pattern of the far-field codeword to make it clear that the far-field codebook cannot capture the user UE in the blind area; wherein the blind area refers to the user UE located at the Fresnel distance R f and minimum beam coverage distance R bf The near field area between.

4. The near-field beam training method according to claim 3, characterized in that: The maximizing the overlapping range between the near field communication area and the distance dimension in any given angular direction by adjusting the focusing distance means: When the angles are aligned, θ = θ p ,φ=φ p , by adjusting the focusing distance r p , in order to maximize the overlap area of ​​the beams in the near field in the distance dimension, expressed as: in, is the optimal focusing distance; argmax is the variable value that maximizes the objective function; Ω0 and Ω p are the near field region and the codeword w respectively. p The beam direction range of , γ represents the overlapping range of the two, and the symbol ∩ represents the intersection; Due to the determination of the beam direction range Ω p As a non-convex problem, we introduce a 3dB main lobe, i.e., the codeword w centered at point p p To make the user UE receive power greater than half of the coverage: Among them, the symbol Means defined as.

5. The near-field beam training method according to claim 4, characterized in that: The process of determining the best focusing distance of the three array structures in any given angular direction includes: The achievable boundary is defined as the minimum coverage distance R of the beam. bf and the maximum coverage distance R br ; For the three array structures, when focusing on the codeword w at any point p, p , when the angle directions are aligned, inequality (15) is further simplified by formula (13) as follows: Then, the inequality R about the beam coverage distance r can be obtained by solving formula (16): bf ≤r≤R br , where the minimum and maximum beam coverage distances R before the received power decays to half bf and R br As shown below: in, For far-field codewords, by setting the sample distance r p →∞, where the minimum covering distance R can be obtained according to the lemma bf ; The function Defined as the minimum beam coverage distance R bf The Fresnel distance R from any given position (r,θ,φ) bf The difference between and is used to represent the blind spot along the distance dimension, specifically: Since the far-field codebook cannot capture the user UE in the blind area, the beam depth is defined as Substituting into formula (17) we get: Based on formulas (17) and (20), and with the goal of maximizing the overlap range between the near field communication area and the distance dimension in any given angular direction, when the maximum beam coverage distance R br When coincident with the Rayleigh distance, the optimal focusing distances of the three array structures in any given angular direction are determined as: Where D represents the array aperture.

6. The near-field beam training method according to claim 5, characterized in that: The codebook structure and corresponding beam training scheme of the first stage include: The angle domain codebook is defined as Each codeword is focused at a different position, where n points are in the focusing direction (θ n ,φ n ) The focus distance is set to Then the angle domain code corresponding to point n is written as: For the three array structures, the setting of the focusing direction should ensure that the total N directions should cover all directions in the near field area, and based on formula (20), it is obtained: For ULA array structure: For UCA array structure: in, is the beamwidth of the UCA array; For UPA array structure: Where arcsin is the inverse sine function. For the one-dimensional arrays ULA and UCA, θ n represents the azimuth of n points, n = 1, 2, ... N; for the two-dimensional array UPA, n i The azimuth of the point, Indicates n j The elevation angle of the point, and n i =1,2,…N I ,n j =1,2,…N J ,N=N I N J ; At this stage, the user UE selects the codeword corresponding to the maximum received signal power and takes multiple angles near the estimated angle index to form an angle candidate set.

7. The near-field beam training method according to claim 6, characterized in that: The codebook structure and corresponding beam training scheme of the second stage include: Based on the angle candidate set, a non-uniform distance sampling strategy is adopted to p ,φ q ) on the basis of the distance domain codebook, the design is W p,q ={w p,q,0 ,w p,q,1 ,…w p,q,S-1 }, S is the number of distance samples at each angle, where each distance domain codeword is recorded as: Where Δ is the sampling threshold, which is different for different arrays and is obtained based on formula (10): For ULA array structure: For UCA array structure: For UPA array structure: By searching all distance domain codewords in the angle candidate set, the optimal distance domain codeword is obtained. and the estimated location of the user UE 8. A near-field beam training device for a very large array based on energy diffusion effect, characterized in that: include: The channel modeling module is used to build a downlink frequency division duplex communication system, and uniformly model the near-field ultra-large-scale channels of three array structures, namely, ULA, UCA, and UPA, as a function related to the wave path difference; A criterion formulation module is used to formulate codebook design criteria for maximizing near-field distance coverage; including: An adjustment unit, which maximizes the overlap range between the near field communication area and the distance dimension in any given angular direction by adjusting the focusing distance; A determination unit, based on the energy diffusion effect of the beam and in combination with the beam pattern corresponding to the near-field ultra-large-scale channel, determines a closed-form solution of the achievable boundary of the beam coverage, so as to determine the optimal focusing distance of the three array structures in any given angular direction; The scheme acquisition module is used to acquire the two-stage codebook structure applicable to the three array structures and the corresponding beam training scheme based on the codebook design criterion.

9. A storage medium, characterized in that: The computer program for near-field beam training of a very large-scale array based on energy diffusion effect is stored therein, wherein the computer program enables a computer to control the near-field beam training method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: one or more processors; Memory; and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, the programs including a program for controlling the near-field beam training method according to any one of claims 1 to 7.

Images