CN-122017724-A - Direction-of-arrival greedy estimation method for amplitude-phase error self-calibration

CN122017724ACN 122017724 ACN122017724 ACN 122017724ACN-122017724-A

Abstract

The invention discloses a greedy estimation method of the direction of arrival of amplitude and phase error self calibration, which comprises the steps of S1, establishing an array receiving signal model containing the amplitude and phase error, S2, calculating the effective grid size and iteration stop threshold of a search grid, S3, conducting rough search on the search grid, detecting the initial direction of arrival and initial signals corresponding to the maximum projection energy, jumping to S6 if the maximum projection energy is lower than the iteration stop threshold, S4, conducting Newton iteration refinement on the initial direction of arrival, obtaining more accurate continuous angle estimation and signal estimation, S5, alternately updating a signal waveform of the source and a diagonal calibration matrix, updating residual signals according to the signal waveform, eliminating a current source, returning to S3, S6, eliminating weak sources through pruning and optimizing, and outputting final direction of arrival and diagonal calibration matrix estimation. The method avoids error transfer and remarkably improves the estimation precision of the direction of arrival under the unknown array error condition.

Inventors

QU FENGZHONG
LU YI
ZHU JIANG
ZHANG MINHAO
LI JINGXUAN
JIANG YULIN
YANG SHAOJIAN
WANG FANGYONG

Assignees

浙江大学

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The amplitude-phase error self-calibrated direction-of-arrival greedy estimation method is characterized by comprising the following steps of: s1, establishing an array receiving signal model containing an amplitude-phase error, wherein the amplitude-phase error is established through a diagonal calibration matrix; s2, calculating the effective grid size of the search grid based on the current diagonal calibration matrix estimation, and determining an adaptive iteration stop threshold according to the effective grid size; S3, performing rough search on the search grid, detecting an initial direction of arrival and an initial signal corresponding to the maximum projection energy, judging whether the maximum projection energy on all grid points is lower than an iteration stop threshold, if so, jumping to execute S6, and if not, continuing to execute S4; S4, newton iterative refinement is carried out on the initial direction of arrival, and more accurate continuous angle estimation and signal estimation are obtained; S5, alternately updating the waveform of the information source signal and the diagonal calibration matrix, updating the residual signal according to the waveform of the information source signal and the diagonal calibration matrix, removing the information source signal subjected to the iterative processing of the round from the received signal, and returning to the step 3; and S6, removing false sources through pruning and optimizing, and outputting final arrival direction and diagonal calibration matrix estimation.
2. The amplitude-phase error self-calibrated greedy estimation method of direction of arrival according to claim 1, wherein in the step S1, the array received signal model is specifically that the received signal matrix is equal to a diagonal calibration matrix multiplied by an ideal steering matrix, multiplied by a source signal matrix and finally added with an additive white gaussian noise matrix, each diagonal element of the diagonal calibration matrix represents a complex gain of a corresponding array element and is used for representing an amplitude to be estimated and a phase error of the array element, and each column of the ideal steering matrix is a steering vector determined by a corresponding direction of arrival.
3. The amplitude-phase error self-calibrated direction-of-arrival greedy estimation method according to claim 1, wherein in S2, the effective mesh size is calculated by: Correcting the ideal guide vector according to the estimated value of the current diagonal calibration matrix to obtain a group of corrected guide vectors related to angles; Calculating cross correlation coefficients between correction guide vectors corresponding to each point on the search grid, and constructing a coherent matrix; And calculating the ratio of the square of all eigenvalues of the coherence matrix to the sum of the squares of all eigenvalues, wherein the ratio is the effective grid size and is used for replacing the nominal grid point total number in the iteration stop threshold calculation.
4. The amplitude-phase error self-calibration method for estimating the direction of arrival greedy according to claim 1, wherein the rough search in the step S3 is specifically that a preset search grid is traversed, for each grid angle, the sum of the energy projected by the current residual signal to the corrected guide vector corresponding to the angle on all snapshots is calculated, the grid angle corresponding to the maximum value is taken as an initial direction-of-arrival estimated value detected by the iteration of the current round, and an initial signal estimated value is obtained according to the initial direction-of-arrival estimated value.
5. The amplitude-phase error self-calibration direction-of-arrival greedy estimation method according to claim 1, wherein the Newton iterative refinement in S4 is characterized by defining an objective function taking an angle as a variable, wherein the objective function is the projection energy of a current residual signal on a corresponding correction guide vector, a new angle estimation value is equal to the current angle estimation value minus a correction term, the correction term is the ratio of the sum of the first derivative of the objective function to the angle to the sum of the second derivative of the objective function to the angle, signal estimation is updated according to the new angle estimation value, and a parameter pair consisting of the new angle estimation value and all snap-shot new signal estimation values is added into the current direction-of-arrival set.
6. The method for greedy estimation of direction of arrival with self-calibration of amplitude and phase error according to claim 1, wherein in S6, source signal waveforms and diagonal calibration matrices are updated alternately, and residual signals are updated accordingly, comprising the following steps: (6.1) fixing the currently estimated diagonal calibration matrix and the direction of arrival, and updating the signal waveforms of all detected sources by solving the least square problem; (6.2) fixing the current estimated direction of arrival and the updated signal waveform of the detected information source, solving the least square problem by array elements, and updating the estimated value of the diagonal element in the diagonal calibration matrix; (6.3) calculating a residual signal of a new iteration by using the updated source signal and the diagonal calibration matrix.
7. The amplitude-phase error self-calibration method for greedy estimation of direction of arrival according to claim 1, wherein the pruning operation in S6 is specifically that all candidate sources detected by iteration are stored in a candidate source set, the total increase of the system residual energy after each candidate source is temporarily removed is calculated, the contribution degree of each detected source to the fitting residual is calculated, namely the proportion of the energy increase corresponding to each candidate source to the total increase is calculated, and the candidate sources with the contribution degree lower than a preset contribution degree threshold are judged as false sources and eliminated.
8. The method for greedy estimation of direction of arrival with self-calibration of amplitude and phase errors according to claim 1, wherein the optimization in S6 is specifically that for the source reserved after pruning, the source signal waveform and the diagonal calibration matrix are updated alternately in S4 and S5 multiple times again, so that all parameter estimates are further converged.
9. The amplitude-phase error self-calibrated DOA greedy estimation device is characterized by comprising a memory and one or more processors, wherein executable codes are stored in the memory, and the one or more processors are used for realizing the amplitude-phase error self-calibrated DOA greedy estimation method according to any one of claims 1-8 when executing the executable codes.
10. A computer-readable storage medium, having stored thereon a program which, when executed by a processor, implements the amplitude-phase error self-calibrated direction-of-arrival greedy estimation method according to any one of claims 1 to 8.

Description

Direction-of-arrival greedy estimation method for amplitude-phase error self-calibration Technical Field The invention relates to the field of array signal processing and parameter estimation, in particular to a greedy estimation method of the direction of arrival of amplitude-phase error self-calibration. Background Direction of arrival (Direction of Arrival, DOA) estimation is a core fundamental technology in many fields such as radar, sonar, wireless communication and the like. The classical subspace-like algorithm has excellent performance under the assumption of an ideal array and noise, but in practical engineering, amplitude and phase errors related to array elements are inevitably introduced into each channel of the array due to hardware manufacturing tolerance, device aging, temperature variation, mutual coupling effect and the like. These errors can cause the mismatch of the actual steering vector of the array and the ideal model, thereby seriously damaging the orthogonality premise of the subspace method, causing the estimation accuracy to be drastically reduced, and the false spectral peak to appear or even completely fail. In order to solve the array error problem, the prior art is mainly divided into two types, namely off-line calibration and on-line self-calibration. Offline calibration requires the use of auxiliary sources of known locations in a controlled environment, is complex to operate and is difficult to cope with the time-varying nature of the operating environment. Although the online self-calibration method can jointly estimate error parameters and DOA by using observed data, most algorithms are limited to single-source scenes, or have high calculation complexity, or require the number of known sources in advance, and have insufficient robustness in complex multi-source and low signal-to-noise ratio actual scenes. In recent years, a method based on sparse reconstruction and compressed sensing shows high resolution potential, but explicit modeling and compensation of array amplitude-phase errors are not carried out, and a large number of false estimates are easy to generate under mismatch conditions. Therefore, there is an urgent need for a practical method that can directly and highly accurately and robustly jointly estimate the amplitude and phase errors of a plurality of sources DOA and arrays from noisy observed data without performing independent calibration in advance and without requiring the number of known sources. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a greedy estimation method of the direction of arrival of amplitude-phase error self calibration, which is characterized in that a greedy iteration framework from thick to thin is constructed, and DOA estimation, signal waveform recovery and array amplitude-phase error calibration are integrated and jointly optimized, so that the greedy estimation method is suitable for the DOA high-precision estimation method under the condition that unknown amplitude and phase errors exist in a sensor array. The specific technical scheme is as follows: A greedy estimation method of the direction of arrival of the self calibration of amplitude and phase errors comprises the following steps: s1, establishing an array receiving signal model containing an amplitude-phase error, wherein the amplitude-phase error is established through a diagonal calibration matrix; s2, calculating the effective grid size of the search grid based on the current diagonal calibration matrix estimation, and determining an adaptive iteration stop threshold according to the effective grid size; S3, performing rough search on the search grid, detecting an initial direction of arrival and an initial signal corresponding to the maximum projection energy, judging whether the maximum projection energy on all grid points is lower than an iteration stop threshold, if so, jumping to execute S6, and if not, continuing to execute S4; S4, newton iterative refinement is carried out on the initial direction of arrival, and more accurate continuous angle estimation and signal estimation are obtained; S5, alternately updating the waveform of the information source signal and the diagonal calibration matrix, updating the residual signal according to the waveform of the information source signal and the diagonal calibration matrix, removing the information source signal subjected to the iterative processing of the round from the received signal, and returning to the step 3; and S6, removing false sources through pruning and optimizing, and outputting final arrival direction and diagonal calibration matrix estimation. Further, in the step S1, the array receiving signal model specifically includes that a receiving signal matrix is equal to a diagonal calibration matrix multiplied by an ideal steering matrix, multiplied by a source signal matrix and finally added with an additive Gaussian white noise matrix, each diagonal element of the di