CN-114966531-B - Subspace wave arrival direction estimation method based on distributed algorithm

CN114966531BCN 114966531 BCN114966531 BCN 114966531BCN-114966531-B

Abstract

The invention belongs to the field of signal processing, in particular to a subspace wave arrival direction estimation method based on a distributed algorithm, which particularly relates to a wave arrival direction estimation method of electromagnetic signals and sonar signals, and can be used for passive positioning and target detection, and comprises the following steps of S1, establishing a distributed array model at a receiving end, S2, calculating a signal subspace of a received signal S3, calculating the weight w (theta), S4, calculating the spatial spectrum and searching the spatial domain to find the DOA value, wherein the method provided by the invention realizes the distributed solution of the algorithm, thereby essentially avoiding the defects of a centralized algorithm, simultaneously, the method can keep good estimation performance, is similar to the estimation performance of the centralized algorithm, and is superior to the distributed MUSIC method when the signal-to-noise ratio is low.

Inventors

CHEN ZHILEI
WU XIAOHUAN
JIA XIAOYUAN

Assignees

南京邮电大学
南京邮电大学

Dates

Publication Date: 20260421
Application Date: 20220525
Priority Date: 20220525

Claims (4)

1. The subspace wave arrival direction estimation method based on the distributed algorithm is characterized by comprising the following steps of: s1, establishing a distributed array model at a receiving end; s2, calculating a signal subspace of the received signal ; S3, calculating weight ; S4, calculating a spatial spectrum and searching a spatial domain to find DOA values; in the step S1, M array elements are used to form a uniform linear array with a distance half of the wavelength of the incident narrowband signal, which is denoted as an array x, and K far-field narrowband signals are set Upon receiving L snapshots, the signal received by array x is represented as: ; Wherein Y is an array element receiving data matrix, X is an incident signal data matrix, N is a noise data matrix, noise and signals are mutually independent, Is an array flow pattern matrix, is an array guiding vector Is a set of (a) and (b), ; Dividing the M arrays into P subarrays, i.e. the number of subarray array elements is Each subarray has a subarray processor and can only store data acquired by processing the subarray, and the adjacent node of each node is specified and expressed as the adjacent node of the ith node The definition of the variables is increased as follows, Represents the vector component belonging to the j-th subarray, The incident signal X is expressed as , wherein, Incident signals received for the 1 st subarray; in the step S2, a signal subspace of the received data is calculated The method specifically comprises the following steps: S21, firstly, calculating a first eigenvector of a covariance matrix R, and randomly initializing the vector Multiplying it with covariance matrix, iterating Secondary, i.e , Wherein, the In order to take the conjugate of the matrix vector, Is the incident signal vector component belonging to the jth subarray of the t-th snapshot, To at the first The eigenvector components belonging to the j-th sub-array in the iteration, Is an average consensus protocol operation, meaning that the data in all sub-brackets is averaged, i.e ; The definition of the laplace matrix is such that when node i can communicate with node j, If not, 0, the degree of freedom of the node itself is The value of (i), i.e ; Let Laplace matrix B have R different eigenvalues And (2) and Then the initial weight is ; The weight is thereafter ; Requirements for Mean of (i) is The iterative formula of the protocol is that ; S22, calculating the modular length of the first feature vector: ; s23, each node normalizes the obtained data: ; S24 and S21 pass through enough iteration times Calculating the rest Feature vectors belonging to large feature values: s25, normalizing the obtained S23 and S24; s26, solving signal subspace Is formed by large eigenvalues of covariance matrix, i.e ; Wherein, the The component belonging to the jth subarray is the ith eigenvector.
2. The method for estimating the direction of arrival of subspace class based on distributed algorithm according to claim 1, wherein in step S3, the weights are solved In the specific form of 。
3. The method for estimating the direction of arrival of subspace class based on distributed algorithm according to claim 2, wherein the step S3 is performed by using a distributed conjugate gradient method The specific steps of each iteration are as follows: s31, calculating the search step length : ; Wherein, the As a residual for the i-th iteration, For the search direction of the i-th iteration, ; S32, each node calculates an update target : ; S33, updating residual errors by each node: ; S34, updating the Schmidt component for searching the direction : ; Wherein, the Distributed solution of (a) in the form S31 The method is carried out by solving the problems, S35, updating the search direction by each node: ; When the iteration number is greater than the set maximum iteration number or the target result updated by the previous iteration and the next iteration has little change, namely The iteration is stopped and, S36, obtaining a repeating unit After that, the processing unit is configured to, The calculation formula of (2) is 。
4. The method for estimating the direction of arrival of subspace class based on distributed algorithm according to claim 3, wherein said step S4 calculates a spatial spectrum : Order the Then ; Due to Acquired by each processor so that the sub-processors can process the data belonging to their own node Data, record Then ; And scanning the space domain to obtain the DOA estimated value.

Description

Subspace wave arrival direction estimation method based on distributed algorithm Technical Field The invention belongs to the field of signal processing, and particularly relates to a subspace wave arrival direction estimation method based on a distributed algorithm, which can be used for passive positioning and target detection. Background DOA (Direction OfArrival ) estimation technology is an important research direction in array signal processing, and is characterized in that signals are received by utilizing sensor arrays at different geographic positions, observed discrete data are obtained, useful information in information source signals is obtained through a series of mathematical algorithm processing, interference information is inhibited, and the DOA (Direction OfArrival ) estimation technology has wide application in the civil and military fields such as automatic driving, wireless communication and radar. The conventional DOA estimation technique is a centralized algorithm, i.e. all array elements in the array are required to transmit the received signal information to a central processor and then to process the correlation algorithm. However, this algorithm idea may be limited in many scenarios. For example, in a large array of geographically dispersed sub-arrays, it is necessary to transfer large amounts of observation data between sub-arrays to a central processor, however the capacity of the communication link may limit the data transfer between sub-arrays. In addition, in many scenarios, centralized processing may fail entirely, such as in a sonar environment, because the signals are received by widely spaced sub-arrays, losing continuity, and thus failing to perform centralized coherent processing. Therefore, in order to improve the adaptability of the algorithm, it is necessary to design a decentered direction-of-arrival estimation algorithm. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a subspace-like direction-of-arrival estimation method based on a distributed algorithm, which is used for carrying out decentralization on the solving process of the subspace-like method, thereby avoiding the influence caused by the traditional centralized algorithm and obtaining the estimation performance similar to that of the centralized algorithm. The invention adopts the following specific technical scheme: A subspace-like direction-of-arrival estimation method based on a distributed algorithm, the method comprising the steps of: s1, establishing a distributed array model at a receiving end; s2, calculating a signal subspace of the received signal S3, calculating a weight w (theta); s4, calculating a spatial spectrum and searching a spatial domain to find DOA values. In the further improvement of the invention, in the step S1, M array elements are utilized to form a uniform linear array with a distance which is half of the wavelength of an incident narrow-band signal, and the uniform linear array is recorded as an array x, K far-field narrow-band signals theta k are assumed, k=1, the number of the far-field narrow-band signals is equal to the number of the far-field narrow-band signals, and after L snapshots are received, the signals received by the array x are expressed as follows: Y=A(θ)X+N Wherein Y is an array element receiving data matrix, X is an incident signal data matrix, N is a noise data matrix, noise and signals are mutually independent, A (theta) = [ a (theta 1),…,a(θK) ] is an array flow pattern matrix, and a (theta k)＝[1,eiπcosθ,…,eiπ(M-1)cosθ]T) is an array guide vector a (theta k) set. The M arrays are divided into P subarrays (nodes), namely, the number of subarrays is C=M/P, each subarray is provided with one sub-processor, and only the data which can be acquired by processing the subarrays can be stored. In addition, the neighbor node of each node is specified, and the neighbor node of the ith node is expressed asThus, the original signal is also divided into components belonging to the respective subarrays. The definition of the original model is increased as follows, (■) j denotes a vector (or matrix) component belonging to the j-th subarray, j=1. For example, the incident signal X may be represented as x= [ X 1,x2,…,xP ], where X 1 is the incident signal received by the 1 st subarray. In a further development of the invention, in step S2, a signal subspace of the received data is calculatedThe method comprises the following steps: S21, firstly, calculating a first eigenvector of the covariance matrix R, randomly initializing the vector u (0), multiplying the vector u by the covariance matrix, and iterating I pw times, namely Where (-) H is the matrix (vector conjugate), x j (t) is the vector component of the incident signal belonging to the jth subarray for the t-th snapshot, and u 1,j(Ipw) is the eigenvector component belonging to the jth subarray in the I pw -th iteration. AC (-) is an average consensus protocol operation, meaning tha