Beam search method based on high frequency band information assistance

Abstract

The application belongs to the field of wireless communication, and particularly relates to a high-efficiency beam training scheme based on high-frequency auxiliary high-frequency. The application adopts a high-frequency auxiliary high-frequency mode to deploy two arrays, wherein a full-digital array is used for the auxiliary array, and a hybrid array is used for the data transmission array. Since the auxiliary array and the data transmission array are both high-frequency bands, the angular domain information of the auxiliary array and the angular domain information of the data transmission array are reciprocal, that is, the data transmission array can use the angular domain information estimated by the auxiliary array to transmit data. Considering the resource cost, the full-digital array used for assistance adopts a small number of arrays, but the angular information estimated by the small number of arrays cannot be directly used for a large array, because the reciprocity of the angular domain information will be weakened due to the difference in the size of the two arrays, so the large array needs to perform a low-complexity beam training, so that the final angular domain information is more accurate, that is, the beam search is performed based on the angular information estimated by the auxiliary array.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of wireless communication, and relates to millimeter-wave communication (mmWave), hybrid beamforming, channel estimation, beam training, out-of-band spatial information, multiple input multiple output technology (MIMO), and in particular to a beam search method based on high-frequency band information assistance. BACKGROUND

[0002] The essence of beam training is to design a codebook in advance for the sending end and the receiving end. The beams in such a codebook need to cover the entire space. Taking exhaustive search as an example, when beam training is performed, a beam is selected from the codebook of the sending end and the receiving end to send and receive signals, respectively, until all beam pairs are selected, and then a beam pair is selected from among the beam pairs to maximize the system efficiency. Although the exhaustive search method can maximize the performance of the system, the training complexity is quite high. Therefore, many studies aim to reduce the complexity of beam training and approximate the performance of the exhaustive search method as much as possible. With the development of wireless communication technology, the frequency of the antenna array is getting higher and higher, and the number of antennas required is also increasing, which leads to extremely strong directivity of the formed beams, and more beams are needed to cover the entire space, so the size of the beam codebook becomes larger, and therefore the time spent on such beam training is also increasing. Therefore, a more effective method is needed to reduce the time complexity of beam training and ensure its performance as much as possible. SUMMARY

[0003] The application is based on the channel reciprocity of the same high frequency band to propose a high-efficiency beam training scheme based on high-frequency assisted high-frequency. In the application, the high-frequency assisted high-frequency mode is used to use the high-frequency auxiliary array to assist the data transmission array of the same high frequency band, wherein the auxiliary array adopts a full-digital array and the data transmission array adopts a hybrid array. Since the auxiliary array and the data transmission array are both high-frequency bands, the angular domain information of the auxiliary array and the angular domain information of the data transmission array are reciprocal, that is, the data transmission array can use the angular domain information estimated by the auxiliary array to transmit data. Considering the resource cost, the full-digital array used for assistance adopts a small number of arrays, but the AOA / AOD estimated by the small number of arrays cannot be directly used for a large array because the reciprocity of the angular domain information will be weakened due to the difference in the size of the two arrays, so the large array needs to perform a low-complexity beam training to make the final angular domain information more accurate, that is, to perform beam search based on the AOA / AOD estimated by the auxiliary array. Further, under the condition of ensuring that the auxiliary array and the data transmission array do not interfere with each other, the auxiliary array and the data transmission array can work at the same time, that is, when the data transmission array transmits the data of the current user, the auxiliary array estimates the AOA / AOD of the next user. In this way, the system efficiency can be greatly improved. In addition, in the case that the auxiliary array is idle, the auxiliary array can also be used for transmitting data or signaling transmission, so as to ensure the maximum utilization of resources.

[0004] Therefore, the work of the application mainly includes two parts: the AOA / AOD estimation stage of the auxiliary array and the beam search stage of the data transmission array.

[0005] The specific method of the application is as follows:

[0006] S1, establishing that the data transmission array and the auxiliary array are used for signal sending and receiving at the same time, and the number of antennas of the data transmission array is more than that of the auxiliary array, the data transmission array is a hybrid array, the auxiliary array is a full-digital array, the data transmission array and the auxiliary array are in the same high frequency band, and the angular domain information of the auxiliary array and the angular domain information of the data transmission array are reciprocal;

[0007] S2, performing channel estimation of the auxiliary array, and sending the estimated angle of departure and angle of arrival to the data transmission array;

[0008] S3, the data transmission array performs beam search according to the received angle of departure and angle of arrival.

[0009] Further, it is defined that the data transmission array and the auxiliary array are both rectangular planar arrays, and the channel matrix h s of the auxiliary array is:

[0010]

[0011] Wherein L represents the path number, a l is the path complex gain, and R,l , respectively are the vertical angle and azimuth angle of the angle of arrival of the ith path, T,l 、 respectively are the vertical angle and azimuth angle of the angle of departure of the ith path, and respectively represent the array response functions of the uniform rectangular planar array of the receiving end and the transmitting end:

[0012]

[0013]

[0014] where m1, n1 are the length and width of the rectangular auxiliary array of the receiving end respectively, m2, n2 are the length and width of the rectangular auxiliary array of the transmitting end respectively, denotes the Kronecker product, d is the antenna spacing, and λ is the antenna wavelength;

[0015] The channel matrix h of the data transmission array d is:

[0016]

[0017] where and respectively represent the array response functions of the rectangular planar array of the receiving end and the transmitting end:

[0018]

[0019]

[0020] M1, N1 in the above formula are the length and width of the rectangular data array of the receiving end respectively, and M2, N2 are the length and width of the rectangular data array of the transmitting end respectively.

[0021] Further, the auxiliary array uses the AOA / AOD algorithm based on SVD decomposition to perform AOA / AOD estimation.

[0022] Further, the specific method of the data transmission array for beam searching according to the received angle of departure and angle of arrival is:

[0023] S31, processing the AOA / AOD estimated by the auxiliary array:

[0024] The AOA / AOD of the lth path estimated by the auxiliary array is defined as l∈{1,2,…,N L} is converted to the coordinates in the DFT codebook grid, (x R,l ,y R,l ) is the AOA estimated by the ith path converted to the coordinates in the DFT codebook grid,T,l ,y T,l ) is estimated AOD of the first path, which is converted into coordinates in the DFT codebook;

[0025] S32, a candidate beam set is generated, and surrounding coordinates are put into set Omega with AOA / AOD coordinates as the center and search range r R ,Ω T :

[0026] The DFT codebooks of the receiving end and the sending end are respectively centered on coordinates (x R,l ,y R,l ), (x T,l ,y T,l ), and surrounding coordinates are put into sets Omega R ,Ω T :

[0027]

[0028]

[0029] According to the relationship between the DFT codebook coordinates and the angle domain, the sets Omega R ,Ω T are converted to the angle domain to obtain sets Theta R ,Theta T , that is, the candidate beam sets of the receiving end and the sending end:

[0030]

[0031]

[0032] S33, find a beam pair in the candidate beam set that maximizes the system performance:

[0033] For the purpose of maximizing the system spectral efficiency, find a beam pair in the candidate beam set of the receiving end and the candidate beam set of the sending end that maximizes the system spectral efficiency, that is, Where i is the subscript of the candidate beam of the receiving end in the candidate beam set, j is the subscript of the candidate beam of the sending end in the candidate beam set, And are the array responses of the receiving end and the sending end antenna array respectively.

[0034] The method of the present application can effectively reduce the algorithm complexity of beam training while ensuring that the system performance approaches the exhaustive search algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 System structure of data transmission array and auxiliary array

[0036] Figure 2 Schematic diagram of beam search

[0037] Figure 3 Simulation curves of spectral efficiency for high-frequency assisted beam training Detailed Implementation

[0038] The present invention will now be described in detail with reference to the accompanying drawings.

[0039] S1. Establish the channel model

[0040] First, a system model is established based on the Saleh-Valenzuela model. For example... Figure 1 As shown, the system architecture includes a data transmission array and an auxiliary array. The data transmission array is a hybrid array with a large number of antennas, where multiple antennas share a single RF chain. The auxiliary array is a fully digital array with fewer antennas, where each antenna is connected to an RF chain. This design aims to reduce the cost of the auxiliary array. Both the data transmission array and the auxiliary array adopt a rectangular planar array structure. Since the data transmission array and the auxiliary array operate in the same high-frequency band and are spatially close, the channel reciprocity between the two arrays is very high.

[0041] The following describes the establishment of the channel matrices for the two arrays: Assuming there are L scattering paths in the signal propagation environment, the channel matrix h of the auxiliary array is... s It can be represented as:

[0042]

[0043] Where L represents the number of paths, α l For path complex gain, θ R,l , These are the perpendicular angle and azimuth angle of the arrival angle of the l-th path, θ. T,l , These are the vertical angle and azimuth angle of the departure angle of the l-th path, respectively. and Let the array response functions of the uniform rectangular planar arrays at the receiving and transmitting ends be respectively:

[0044]

[0045]

[0046] In the above formula, m1 and n1 are the length and width of the rectangular auxiliary array at the receiving end, respectively, and m2 and n2 are the length and width of the rectangular auxiliary array at the transmitting end, respectively. Let d represent the Kronecker product, d be the antenna spacing, and λ be the antenna wavelength.

[0047] Similarly, the channel model of the data transmission array is established, the channel matrix of the data transmission array is different from that of the auxiliary array in the array response function, which is caused by the different sizes of the two arrays, so the channel matrix of the data transmission array h d can be expressed as:

[0048]

[0049] where L represents the number of paths, α l is the path complex gain, θ R,l , and θ T,l , are the vertical angle and azimuth angle of the arrival angle of the lth path, respectively. and respectively represent the array response functions of the uniform rectangular planar array at the receiving end and the transmitting end:

[0050]

[0051]

[0052] M1 and N1 in the above formula are the length and width of the rectangular data array at the receiving end, respectively, and M2 and N2 are the length and width of the rectangular data array at the transmitting end, respectively, denotes the Kronecker product, d is the antenna spacing, and λ is the antenna wavelength.

[0053] S2, AOA / AOD estimation of the auxiliary array

[0054] After the channel model is established, the channel estimation of the auxiliary array is performed, the auxiliary array estimates the departure angle and the arrival angle based on the channel of the auxiliary array using the SVD channel estimation method, and sends them to the data transmission array, and the data transmission array searches for more accurate departure angle and arrival angle based on the channel of the data transmission array using the beam search method.

[0055] In order to facilitate the derivation of the algorithm, the channel matrix h s of the auxiliary array is expressed in vector form H s :

[0056]

[0057] where

[0058] The transmitting end auxiliary array transmits q different pilot sequences x1, x2, …, x q , and the receiving end auxiliary array receives the pilot sequences using the combining matrix W, and the received signal can be expressed in vector form as:

[0059] Y = W H H s X + N

[0060] where W = [w1, w2, …, w q ] is the receive-end precoding matrix sequence, H s is the vector form of channel H m , X = [x1, x2, …, x q ] is the pilot sequence, Y = [y1, y2, …, y q ] is the received signal sequence, and N = [n1, n2, …, n q ] is the noise sequence.

[0061] The channel estimation problem is to estimate the above channel H s In a high-frequency channel, the channel can be regarded as sparse, so the estimation of the above sparse channel matrix H s can be converted into the estimation of the channel gain, angle of arrival and angle of departure of a limited number of paths, so the estimation problem of the sparse channel can be converted into an optimization problem:

[0062]

[0063] where a is the vector form of the channel complex gain a = [a1, a2, … a L ] T ,‖a‖0 is the number of non-zero elements in a, is the estimated channel matrix, and e is the error coefficient.

[0064] The received signal matrix is decomposed by SVD: Y = TΣV H ,Σ = diag(σ1, σ2, …, σ r ), r is the rank of Y, and U and V are both unitary matrices. If the noise is ignored, the L largest singular values and their corresponding singular vectors can be approximately represented as:

[0065]

[0066]

[0067]

[0068] where σ l is the lth singular value of Y, u l , and v l are the lth column vectors of U and V, respectively.

[0069] Due to the sparsity of the channel, only the angles of departure and arrival corresponding to the first few largest singular values need to be estimated, and from Y = UΣV H , we have: UH YV =∑, bring in Y = W H H s X+N available:

[0070] U H (W H H s X+N)V =∑

[0071] That is

[0072] For example, the first path, the above formula can be converted to:

[0073]

[0074] In order to select several paths with maximum gain, and obtain the corresponding AOA and AOD, the above formula can be converted into an optimization problem:

[0075]

[0076]

[0077] So far, the rough estimation of AOA / AOD of the auxiliary array is completed, and the specific algorithm flow is shown in Table 1.

[0078] Table 1 Channel estimation algorithm flow

[0079]

[0080] S3, beam search phase

[0081] After the AOA / AOD is estimated by the auxiliary array, it is sent to the data transmission array, and a more accurate beam search is performed in a small number of time slots during the data transmission stage. If the channel reciprocity of the auxiliary array and the data transmission array is strong enough, the beam search in this stage is unnecessary, and the AOA / AOD estimated by the auxiliary array can be directly used for data transmission during data transmission. However, considering the complex channel environment, the channel reciprocity of the two antenna arrays may not be strong enough in some cases, and the data transmission array needs to further search the beam based on the AOA / AOD estimated by the auxiliary array; and considering the cost of the auxiliary array, a small number of antenna arrays are used for AOA / AOD estimation in the previous stage, which has the disadvantage that the channel reciprocity of the small auxiliary array and the large data transmission array will be weakened due to the different sizes of the two arrays. Therefore, the second stage beam search is necessary, and the beam search in this stage is a time complexity lower beam search process based on the AOA / AOD estimation in the previous stage. Therefore, the beam training method in this stage is based on the AOA / AOD estimation in the previous stage, and the time complexity is much lower than the exhaustive search method, and the performance can approach the exhaustive search method.

[0082] The specific algorithm is as follows: first, the number of antennas of the data transmission array is used to design the codebook, and the DFT codebook is used, the number of codebook beams is consistent with the number of antennas, and in the DFT codebook of the planar array, the horizontal coordinate is the azimuth angle and the vertical coordinate is the vertical angle. Assuming that the auxiliary array estimates the AOA / AOD of a certain path, the AOA and the AOD both include the azimuth angle and the vertical angle, i.e. Convert the AOA and the AOD to the DFT codebook coordinates of the receiving end and the transmitting end respectively, i.e. find the beam closest to the AOA / AOD in the codebook, and return its coordinates, i.e. (x R,l ,y R,l ),(x T,l ,y T,l ), and in the DFT grid codebook of the receiving end / transmitting end, a square is formed with the above (x R,l ,y R,l ) / (x T,l ,y T,l ) as the center and r as the search range. The specific search range r is determined by the number of antennas of the data transmission array, and the square formed contains all the coordinates of the candidate beam set. Let Ω R be the receiving end candidate beam set, and Ω T be the transmitting end candidate beam set. Next, a beam pair is found in the receiving end candidate beam set Ω R and the transmitting end candidate beam set Ω T such that the system performance is maximized.

[0083] More specifically, taking the departure angle of a certain path as an example, as shown inFigure 2 As shown (a DFT codebook of size 64, with the horizontal axis representing the azimuth and the vertical axis representing the vertical angle), the angle estimated by the auxiliary array is converted into coordinates (3,6) in the DFT codebook, and it needs to be further precisely searched within the range of two cells surrounding (3,6).

[0084] The algorithm for precise beam search is shown in Table 2.

[0085] Table 2. Precise Beam Search Process

[0086]

[0087]

[0088] Based on the aforementioned beam search algorithm and the previous auxiliary array AOA / AOD estimation algorithm, the performance of the proposed scheme was simulated using MATLAB, and a comparison with the exhaustive search algorithm was performed. Taking some simulation results as an example, the simulation parameters are: the auxiliary array is 4×4 in size, and the spacing... A uniform rectangular planar array; the data transmission array uses a size of 16×16 and a spacing of... A uniform rectangular planar array; the channel adopts the Saleh-Valenzuela channel model, assuming 3 paths; the DFT codebook size during the beam search phase is 16×16, consistent with the number of antennas in the data transmission array. Figure 3 As shown, this graph is plotted with the signal-to-noise ratio on the horizontal axis and the system's spectral efficiency curve on the vertical axis. Figure 3 It can be seen that the proposed solution can approximate the exhaustive search algorithm.

Claims

1.A beam search method based on high frequency band information assistance, characterized in that, The method comprises the following steps: S1, the data transmission array and the auxiliary array are used for signal transmission and reception at the same time, the number of antennas of the data transmission array is more than the number of antennas of the auxiliary array, the data transmission array is a hybrid array, the auxiliary array is a full-digital array, the data transmission array and the auxiliary array are in the same high frequency band, the angular domain information of the auxiliary array and the angular domain information of the data transmission array have reciprocity; it is defined that the data transmission array and the auxiliary array are both rectangular planar arrays, the channel matrix of the auxiliary array is: : ， where denotes the number of paths, is the path complex gain, , are the vertical and azimuth angles of departure (AoD) of the mth path, are the vertical and azimuth angles of arrival (AoA) of the mth path, , are the vertical and azimuth angles of departure (AoD) of the mth path, are the vertical and azimuth angles of arrival (AoA) of the mth path, and denote the array response functions of the uniform rectangular planar array at the receiving and transmitting ends, respectively: ， ， wherein , are the length and width of the rectangular auxiliary array at the receiving end, respectively, , are the length and width of the rectangular auxiliary array at the transmitting end, respectively, denotes the Kronecker product, is the antenna spacing, is the antenna wavelength; Channel matrix of a data transmission array is: ， where and denote the array response functions of rectangular planar arrays at the receiving and transmitting ends, respectively: ， ， L and W in the above equation are the length and width of the rectangular data array at the receiving end, respectively, , L and W in the above equation are the length and width of the rectangular data array at the receiving end, respectively, , L and W in the above equation are the length and width of the rectangular data array at the receiving end, respectively, S2, performing channel estimation of the auxiliary array, and sending the estimated departure angle and arrival angle to the data transmission array; S3, the data transmission array performs beam search according to the received departure angle and arrival angle. 2.The high-frequency band information assisted beam search method of claim 1, wherein, In step S2, the auxiliary array uses the AOA / AOD algorithm based on SVD decomposition to perform AOA / AOD estimation. 3.The high-frequency band information assisted beam search method of claim 2, wherein, The specific method of S3 is: S31, processing the AOA / AOD estimated by the auxiliary array: The estimated AOA / AOD of the mthpath, is converted to a coordinate in the DFT codebook grid, is converted to a coordinate in the DFT codebook grid, is the estimated AOA of the mthpath, is converted to a coordinate in the DFT codebook grid, is the estimated AOD of the mthpath, is converted to a coordinate in the DFT codebook grid. S32, generate a candidate beam set, centered on the AOA / AOD coordinates, with a search range Put the surrounding coordinates into the set , : The DFT codebook of the receiving end and the transmitting end are respectively centered on coordinates with a search range The surrounding coordinates are put into a set , ​ ， ， According to the relationship between the DFT codebook coordinates and the angle domain, the coordinates are converted to the angle domain to obtain a set , , , , that is, the candidate beam set of the receiving end and the sending end. ， ， S33, finding the beam pair that maximizes the system performance in the candidate beam set: To maximize the system spectral efficiency, a beam pair is found in the candidate beam set at the receiving end and the candidate beam set at the transmitting end to maximize the system spectral efficiency, that is where is the superscript of the receiving end candidate beam in the candidate beam set, is the superscript of the transmitting end candidate beam in the candidate beam set, and are the array responses of the receiving end and the transmitting end antenna array, respectively.

Images