CN-121984550-A - Two-dimensional DOA estimation method under analog-digital hybrid array architecture

CN121984550ACN 121984550 ACN121984550 ACN 121984550ACN-121984550-A

Abstract

The invention relates to a two-dimensional DOA estimation method under an analog-digital mixed array architecture, the two-dimensional angle estimation problem is decoupled into two one-dimensional estimates by adopting an L-shaped antenna array and a partially connected hybrid analog-digital architecture. And secondly, performing signal-to-noise ratio calculation through multi-sector scanning based on discrete Fourier transform, and realizing quick rough estimation of azimuth and elevation. And reconstructing a spatial covariance matrix in a local angle interval corresponding to the rough estimation result, and searching for a fine spectrum peak by using a MUSIC algorithm to obtain a high-precision angle estimation value. And finally, solving the problem of angle pairing among subarrays by calculating the two-dimensional MUSIC frequency spectrum value of candidate angle pairing, and outputting a final two-dimensional DOA estimation result. Through the multi-stage processing strategies of decoupling estimation, partition search and local refinement, the calculation complexity is obviously reduced, and meanwhile, the high precision and the high resolution under the condition of low signal to noise ratio are ensured.

Inventors

TIAN YE
Qing Junjie
WU JINTAO
ZHU DOUDOU
YU SHUO

Assignees

宁波大学

Dates

Publication Date: 20260505
Application Date: 20251231

Claims (10)

1. A method for estimating two-dimensional DOA in an analog-to-digital hybrid array architecture, the method comprising the steps of: S1, constructing an L-shaped antenna array and a PC-HAD architecture model, wherein the L-shaped antenna array consists of an orthogonal x-axis uniform linear array and a z-axis uniform linear array, and the number of antenna units on the x-axis and the number of antenna units on the z-axis are both The PC-HAD architecture model is deployed with respect to both an x-axis uniform linear array and a z-axis uniform linear array Each radio frequency chain is correspondingly connected with an independent analog subarray in the L-shaped antenna array, and the analog subarray comprises And an antenna element, wherein, , ; S2, constructing a hybrid beam shaper of the PC-HAD architecture model, wherein in an analog domain, an analog combiner based on a discrete Fourier transform matrix is configured for each analog subarray on an x-axis and each analog subarray on a z-axis, the analog combiner on the x-axis divides the space angle of a received signal of an antenna array on the x-axis into N sectors, and the analog combiner on the z-axis divides the space angle of the received signal of the antenna array on the z-axis into N sectors; S3, respectively aiming at the x-axis uniform linear array and the z-axis uniform linear array, carrying out beam scanning by utilizing the sectors obtained by dividing in the step S2, and calculating the signal to noise ratio of the output signal of each sector; based on each candidate sector, determining a corresponding analog combiner and a digital combiner, and searching SNR spectrum peaks respectively aiming at an x-axis uniform linear array and a z-axis uniform linear array to respectively obtain a group of elevation angle rough estimation values and a group of azimuth angle rough estimation values; S4, constructing a local angle search grid in the neighborhood of each angle coarse estimation value obtained in the step S3, calculating a cross-correlation function between output signals of all radio frequency chains according to each angle in the local angle search grid, and reconstructing a space covariance matrix corresponding to the angle according to the cross-correlation function; s5, pairing all the obtained elevation angle fine estimation values with all the azimuth angle fine estimation values to generate candidate two-dimensional angle pairs, calculating the two-dimensional MUSIC spectrum value of each candidate two-dimensional angle pair, and selecting the first R angle pairs with the largest two-dimensional MUSIC spectrum value as a final two-dimensional DOA estimation result.
2. The method for estimating two-dimensional DOA under an analog-to-digital hybrid array architecture of claim 1 wherein the received signals of the L-shaped antenna array are specifically expressed as: Wherein, the method comprises the steps of, Representing variance as Is added to the zero-mean additive white gaussian noise of (c), Representing R unknown baseband signals; , ; ; ; , indicating the antenna spacing(s) and, Indicating the wavelength of the signal, Representing the elevation angle of the light beam, ; Indicating the azimuth angle of the beam, 。
3. The method for estimating two-dimensional DOA under an analog-to-digital hybrid array architecture of claim 2 wherein in step S2, the discrete Fourier transform matrix-based analog combiner is specifically a discrete Fourier transform matrix-based analog combiner The analog combiner From N sub-combiners The constitution, i.e Wherein each sub-combiner Are all block diagonal matrices, and diagonal blocks of each block diagonal matrix are formed by Point DFT matrix Is formed by L continuous column vectors in matrix The initial index of (a) is Each sub-combiner Expressed as: similarly, a discrete Fourier transform matrix analog combiner is used The analog combiner From N sub-combiners The constitution, i.e Wherein each sub-combiner Are all block diagonal matrices, and diagonal blocks of each block diagonal matrix are formed by Point DFT matrix Is formed from L continuous column vectors in D matrix The initial index of (a) is Each sub-combiner Expressed as: Thereby ensuring coverage of the full spatial angle, wherein, = 。
4. A two-dimensional DOA estimation method in an analog-to-digital hybrid array architecture as defined in claim 3, wherein in step S2, the analog combiner in the z-axis divides the spatial angle of the received signals of the antenna array in the z-axis into N sectors, and the analog combiner in the x-axis divides the spatial angle of the received signals of the antenna array in the x-axis into N sectors, specifically each sub-combiner Corresponds to a mathematical space angle sector, the first Analog sub-combiner Corresponding space angle sector The index of the DFT column vector used by the method is determined by the specific expression: ; Sector said spatial angle Mapping to a physical angle domain, and obtaining intersection with an effective elevation interval to obtain an effective physical sector, wherein the specific expression is as follows: ; Each sub-combiner Likewise, a mathematical space angle sector is also assigned, the first Analog sub-combiner Corresponding space angle sector The index of the DFT column vector used by the method is determined by the specific expression: ; Sector said spatial angle Mapping to a physical angle domain, and obtaining intersection with an effective horizontal angle interval to obtain an effective physical sector, wherein the specific expression is as follows: 。
5. The method for estimating two-dimensional DOA under an analog-to-digital hybrid array architecture as defined in claim 4, wherein in step S2, the specific design process of the digital combiner is as follows: according to the incoming wave direction of the target signal source Constructing corresponding array steering vectors in the directions of the z axis and the x axis And Combined analog combiner 、 Array steering vectors corresponding to each other , Respectively calculating corresponding conjugate transposed projections And The conjugate transpose projects as directed to the incoming wave direction Weight vector of digital combiner of (a) And Wherein , 。
6. The method for estimating two-dimensional DOA under an analog-to-digital hybrid array architecture as defined in claim 5, wherein the specific process of step S3 is as follows: s3.1, respectively aiming at the x-axis uniform linear array and the z-axis uniform linear array, carrying out beam scanning by utilizing the N sectors obtained by dividing in the step S2, and calculating the signal to noise ratio of the analog output signal of the nth sector: , Wherein, the method comprises the steps of, , , , , And (3) with Respectively are Corresponding first A signal component and a noise component of the segment snapshot; , , , , And (3) with Respectively are Corresponding first The signal component and the noise component of the segment snapshot, Noise variance, K is snapshot number; S3.2, setting signal-to-noise ratio threshold Screening out the analog output signals on the x-axis and the z-axis to meet the requirements The selected sector set is divided into all sectors And Screening results corresponding to the x axis and the y axis respectively; s3.3 for each candidate sector , wherein, Or (b) The candidate sector elements of the x axis and the z axis are v, and the corresponding analog combiner and digital combiner are determined based on the sector set Obtaining an analog combiner matrix for coarse estimation Digital combiner matrix Based on sector set Obtaining a sector set Analog combiner matrix of (a) Digital combiner matrix ; S3.4, using the analog combiner matrix obtained in step S3.3 And digital combiner matrix Processing the received signals of the analog subarrays on the x-axis and collecting the signals in the sectors The SNR spectrum peak search is carried out in the corresponding angle range, the first R significant peaks in the SNR spectrum are detected, and the corresponding angle values form a group of elevation angle rough estimation values R is 1 to R, and the joint simulation combiner matrix obtained in the step S3.3 is utilized And joint digital combiner matrix Processing the received signals of the analog subarrays in the z-axis and collecting the signals in the sectors Searching SNR spectrum peaks in the corresponding angle range, detecting the first R significant peaks in the SNR spectrum, wherein the corresponding angle values form a group of azimuth coarse estimation values R is 1 to R.
7. The method for two-dimensional DOA estimation in an analog-to-digital hybrid array architecture of claim 6 wherein in step S4, the local angle search grid is constructed by coarse estimation of elevation angle Setting local search interval as center Mapping the angle interval to a spatial frequency domain And at To the point of Uniformly sample Q points over a range of (1) By inverse mapping Obtaining local angle search grids Similarly, coarse estimation of azimuth angle Setting local search interval as center Mapping the angle interval to a spatial frequency domain And at To the point of Uniformly sample Q points over a range of (1) Finally by inverse mapping Obtaining local angle search grids 。
8. The method for estimating two-dimensional DOA under an analog-to-digital hybrid array architecture as defined in claim 7, wherein in step S4, a cross-correlation function between all the output signals of the radio frequency chains is calculated, and a spatial covariance matrix corresponding to the angle is reconstructed according to the cross-correlation function, and the method comprises the following specific steps: searching each elevation angle in a grid for a local angle of the z-axis Beam scanning is carried out to obtain output signals of L radio frequency chains, wherein in the first step The signal combinations received on the individual RF chains are: , Calculate the first Strip RF chain and the first Cross-correlation function between the output signals of the strip RF chains: The method comprises the following steps of: , wherein, Is the first The sub-covariance matrix is obtained after vectorizing the equation: Order the Due to The matrix is the kronecker product of two related steering vectors, Orthogonalization is carried out by adopting diagonal loading, and the following steps are obtained: Wherein, the method comprises the steps of, Is the diagonal loading coefficient of the sample, Finally, all the sub-covariance matrixes are reassembled to obtain an elevation rough estimation value The corresponding overall spatial covariance matrix: similarly, the azimuth coarse estimation value is obtained Is a global spatial covariance matrix of (a): 。
9. The method for estimating two-dimensional DOA under an analog-to-digital hybrid array architecture as defined in claim 8, wherein in step S4, the spatial covariance matrix corresponding to each angle is processed by using a MUSIC algorithm to obtain a MUSIC spatial spectrum, and the method comprises the following steps of Performing feature decomposition according to The approximation of the signal noise subspace is obtained as The MUSIC spectrum is obtained, specifically expressed as: , wherein, , Is a guide vector, in The spectrum peak value screening is carried out, and the maximum spectrum peak value is selected As a precise estimate of elevation angle And similarly, spatial covariance matrix Performing feature decomposition to obtain an approximation of signal noise subspace of The MUSIC spectrum is obtained, specifically expressed as: , wherein, In the following The spectrum peak value is screened, and the largest is selected As a precise estimate of azimuth 。
10. The method for estimating two-dimensional DOA under architecture of analog-digital hybrid array according to claim 9, wherein in step S5, all the obtained fine estimation values of elevation angle are paired with all the fine estimation values of azimuth angle to generate candidate two-dimensional angle pairs, and the two-dimensional estimation result is screened out by spectral function values, which comprises the steps of correspondingly estimating R fine estimation values of elevation angle obtained by corresponding to the analog subarray of z axis R azimuth fine estimation values obtained by corresponding estimation of analog subarrays of x axis Combining two by two to generate a set containing R.times.R candidate angle pairs: R is the number of signal sources, and the two-dimensional guide vector of the kth candidate angle pair is recorded K takes 1 to R, representing the two-dimensional received signal as: wherein , Based on covariance matrix For covariance matrix Performing feature decomposition to obtain signal noise subspace According to a spectral value function And calculating a spectrum function for each candidate angle pair, and selecting the first R angle pairs with the largest spectrum function value as two-dimensional DOA estimation results.

Description

Two-dimensional DOA estimation method under analog-digital hybrid array architecture Technical Field The invention relates to the technical field of wireless communication and signal processing, in particular to a two-dimensional DOA estimation method under an analog-digital hybrid array architecture. Background In 5G and future 6G mobile communication systems, millimeter wave (mmWave) frequency band is considered as one of key technologies for realizing ultra-high speed, ultra-low time delay and high capacity communication because it has ultra-wide bandwidth resources of several GHz level. In order to compensate for high path loss of millimeter wave signals in the propagation process, a large-scale multiple input multiple output (Massive MIMO) array is widely applied to a base station and terminal equipment, and can improve link reliability and signal gain through Beamforming (Beamforming) and space diversity (SPATIAL DIVERSITY). In such systems, angle of arrival (DOA, direction of Arrival) estimation is a core link to achieve beam pointing control, user tracking and spatial channel estimation, whose performance directly affects beam forming accuracy and system throughput. However, under the conditions of high-frequency broadband and large-scale array, the conventional all-digital receiving architecture requires an independent radio frequency link (RF chain) corresponding to each antenna unit, resulting in a drastic rise in hardware cost, power consumption and signal processing complexity, which is not suitable for practical deployment. In order to reduce the complexity of hardware, researchers have proposed Hybrid Analog-Digital (HAD) architecture, in which a part of the connection (PARTIALLY CONNECTED, PC) structure is a mainstream solution due to low cost. However, the architecture performs dimension compression on the signals in an analog domain, so that the spatial freedom degree of the received signals is reduced, and the traditional super-resolution algorithms based on feature decomposition such as MUSIC and ESPRIT are difficult to directly apply. In addition, the problem of two-dimensional DOA estimation caused by the sparse characteristic of the millimeter wave channel has nonlinear and strong coupling characteristics, and if two-dimensional joint search is directly adopted, the calculation complexity of the method is increased in O (N 2), so that the method is not suitable for a real-time communication scene. In view of the above problems, prior studies have proposed various improvements such as Compressed Sensing (CS) methods, subspace projection (Subspace Projection) methods, and DOA estimation frameworks based on deep learning, but these methods either rely on large amounts of training data or are unstable in performance under low signal-to-noise and few-snapshots conditions. Therefore, how to realize two-dimensional DOA estimation with high accuracy and low computational complexity under the environment of low hardware complexity and low signal-to-noise ratio becomes one of the key technical problems in the current millimeter wave large-scale MIMO system. Disclosure of Invention The technical problem to be solved by the invention is to provide a two-dimensional DOA estimation method under an analog-digital hybrid array architecture, which can realize high precision and low computational complexity under the environment of low hardware complexity and low signal-to-noise ratio. The technical scheme adopted by the invention is that the two-dimensional DOA estimation method under the analog-digital mixed array architecture comprises the following steps: S1, constructing an L-shaped antenna array and a PC-HAD architecture model, wherein the L-shaped antenna array consists of an orthogonal x-axis uniform linear array and a z-axis uniform linear array, and the number of antenna units on the x-axis and the number of antenna units on the z-axis are both The PC-HAD architecture model is deployed with respect to both an x-axis uniform linear array and a z-axis uniform linear arrayEach radio frequency chain is correspondingly connected with an independent analog subarray in the L-shaped antenna array, and the analog subarray comprisesAnd an antenna element, wherein,,; S2, constructing a hybrid beam shaper of the PC-HAD architecture model, wherein in an analog domain, an analog combiner based on a discrete Fourier transform matrix is configured for each analog subarray on an x-axis and each analog subarray on a z-axis, the analog combiner on the x-axis divides the space angle of a received signal of an antenna array on the x-axis into N sectors, and the analog combiner on the z-axis divides the space angle of the received signal of the antenna array on the z-axis into N sectors; S3, respectively aiming at the x-axis uniform linear array and the z-axis uniform linear array, carrying out beam scanning by utilizing the sectors obtained by dividing in the step S2, and calculating the signal to noise ratio of