CN-121978619-A - Blind sparsity subspace backtracking homogeneity multi-source positioning method for residual error sparsity decision

CN121978619ACN 121978619 ACN121978619 ACN 121978619ACN-121978619-A

Abstract

The embodiment of the invention discloses a blind sparsity subspace backtracking homogeneous multi-source positioning method for residual sparsity decision, which comprises the steps of rasterizing a monitored area containing unknown signal source nodes and known sensing nodes, acquiring actual received signal strength as an observation vector (also serving as a residual vector), and constructing a sparse observation model based on grid positions and the observation vector. Calculating a normalized inner product of the residual vector and each column of the sensing matrix in the model to obtain a correlation numerical sequence, determining a self-adaptive threshold value through the absolute deviation and the mean value of the median, screening candidate elements to form a candidate support set, screening the received signal strength estimated value to generate an iterative position support set, and updating the residual vector. After each iteration, comparing the theory of the residual vector with the empirical probability distribution to obtain a difference metric value, stopping iteration when the termination condition is met, and obtaining the grid position corresponding to the current position support set as a positioning result. The method only depends on the strength of the received signal, and reduces the deployment cost of the sensor hardware.

Inventors

QI PEIHAN
REN JINYANG
ZHU PANPAN
YIN KAI
MENG YONGCHAO
Niu Kairui
CHEN HAIQIN
HAO YUPENG
ZHANG ZIHE

Assignees

西安电子科技大学

Dates

Publication Date: 20260505
Application Date: 20251224

Claims (10)

1. A blind sparsity subspace backtracking homogeneity multi-source positioning method for residual divergence decision, which is characterized by comprising the following steps: Performing rasterization division on a monitored area, wherein the monitored area comprises signal source nodes with unknown positions and sensing nodes with known positions, the signal source nodes emit electromagnetic signals, the sensing nodes acquire the electromagnetic signals, acquire actual received signal intensities of the electromagnetic signals acquired by the sensing nodes, take the actual received signal intensities as observation vectors, and construct a sparse observation model based on the grid positions of the signal source nodes and the observation vectors; Taking the observation vector as a residual vector; Calculating normalized inner products of the residual vector and each column of a sensing matrix in the sparse observation model to obtain a correlation numerical sequence, determining an adaptive threshold according to the median absolute deviation and the mean value of the correlation numerical sequence, determining candidate elements according to the comparison condition of the adaptive threshold and each element in the correlation numerical sequence, and forming a candidate support set through grid positions of the candidate elements; screening the candidate support set based on the received signal strength estimated value of the grid position in the candidate support set to generate a position support set for the iteration; Updating the residual vector according to the position support set for the current iteration; after updating the residual vector in each iteration, determining the theoretical probability distribution and the empirical probability distribution of the residual vector, comparing the theoretical probability distribution with the empirical probability distribution to obtain a difference metric value, and stopping the iteration process when the difference metric value meets a preset termination condition, wherein grid positions corresponding to the position support set of the current iteration are the final positioning result of the signal source.
2. The blind sparsity subspace backtracking homogeneous multi-source positioning method for residual sparsity decision according to claim 1, wherein the constructing a sparse observation model based on the grid position of the signal source node and the observation vector specifically comprises: taking the existence condition of the signal source node at the grid position as a sparse vector to be solved; Taking the sensing nodes as rows and grid positions as columns to construct a measurement matrix; Determining Euclidean distance between the grid positions of the sensor node and the signal source node according to the space coordinates of the sensor node and the grid positions; determining path energy loss according to the Euclidean distance to obtain a path loss matrix; Determining a sensing matrix according to the product of the measuring matrix and the path loss matrix, wherein each row vector of the sensing matrix corresponds to one sensing node, and each column vector corresponds to one grid position; And constructing a sparse observation model according to the sensing matrix and the sparse vector to be solved and combining the observation vector acquired by the sensing node.
3. The blind sparsity subspace backtracking homogeneous multisource positioning method of residual sparsity decision according to claim 1, wherein the determining an adaptive threshold according to a median absolute deviation and a mean of the correlation value sequence, determining candidate elements according to a comparison of the adaptive threshold and each element in the correlation value sequence, and forming a candidate support set by grid positions of the candidate elements, specifically comprises: Performing ascending sort on the elements in the correlation sequence to obtain a first sorting sequence; Determining the median of the first sequencing sequence, calculating the absolute difference value between each element in the sequencing sequence and the median, and constructing an absolute deviation sequence; Ascending sort is carried out on the absolute deviation sequence to obtain a second sort sequence; Determining the median of the second ordered sequence, wherein the median of the second ordered sequence is the median absolute deviation; Determining a mean value of the correlation sequence; determining an adaptive threshold according to the median absolute deviation and the mean value; And determining that an element in the correlation numerical value sequence is larger than the self-adaptive threshold, and forming a candidate support set by using the grid position corresponding to the most relevant element, wherein the element is the most relevant element.
4. The blind sparsity subspace backtracking homogeneous multi-source positioning method of residual sparsity decision according to claim 1, wherein the filtering the candidate support set based on the received signal strength estimation value corresponding to the grid position in the candidate support set, and generating a position support set for the current iteration specifically comprises: extracting column vectors corresponding to grid positions in the candidate support set from a sensing matrix to form a sub-sensing matrix; determining a received signal strength estimation value corresponding to a grid position in the sub-sensing matrix based on the sub-sensing matrix and the residual vector, and forming a received signal strength estimation matrix; And arranging the received signal strength estimated values in the received signal strength estimated matrix in a descending order, and selecting grid positions corresponding to the preset number of received signal strength estimated values as a position support set of the iteration.
5. The blind sparsity subspace backtracking homogeneous multi-source positioning method for residual sparsity decision according to claim 4, wherein the determining the received signal strength estimation value corresponding to the grid position in the sub-sensing matrix based on the sub-sensing matrix and the residual vector forms a received signal strength estimation matrix of the sub-sensing matrix, and specifically comprises: Replacing a sensing matrix in the sparse observation model with the sub-sensing matrix; And determining a received signal strength estimation value of each grid position in the sub-sensing matrix based on the mapping relation between the sub-sensing matrix and the observation vector in the sparse observation model, and obtaining the received signal strength estimation matrix of the sub-sensing matrix.
6. The blind sparsity subspace backtracking homogeneous multi-source positioning method of residual sparsity decision according to claim 1, wherein said updating said residual vector according to said position support set for the current iteration comprises: extracting column vectors corresponding to grid positions in the position support set from the sensing matrix to form a support set sensing matrix; Replacing a sensing matrix in the sparse observation model with the support set sensing matrix, and determining a received signal strength estimated value of each grid position in the support set sensing matrix based on the mapping relation between the support set sensing matrix and an observation vector in the sparse observation model to obtain a received signal strength estimated matrix of the support set sensing matrix; replacing the sparse vector to be solved with a received signal strength estimation matrix of the support set sensing matrix to obtain an updated sparse observation model; and determining an updated residual vector according to the observation vector in the updated sparse observation model, the received signal strength estimation matrix of the support set sensing matrix and the support set sensing matrix.
7. The blind sparsity subspace backtracking homogeneous multi-source positioning method of claim 1, wherein after updating the residual vector in each iteration, determining a theoretical probability distribution and an empirical probability distribution of the residual vector, comparing the theoretical probability distribution with the empirical probability distribution to obtain a difference metric, stopping the iteration process when the difference metric meets a preset termination condition, wherein grid positions corresponding to a position support set of the iteration are final positioning results of signal sources, and the method specifically comprises: after updating the residual vector in each iteration, determining the mean and variance of the residual vector; Determining a theoretical probability distribution according to the mean and variance of the residual vector; determining an empirical probability distribution of the residual vector by kernel density estimation; determining a difference metric value according to the theoretical probability distribution and the empirical probability distribution; updating the iteration times when the difference metric is larger than a preset threshold value, and returning to the step of calculating the normalized inner product of the residual vector and each column of the sensing matrix in the sparse observation model; And stopping the iterative process when the difference measurement is smaller than or equal to a preset threshold value, wherein the grid position corresponding to the position support set of the iteration is the final positioning result of the signal source.
8. The blind sparsity subspace backtracking homogeneous multi-source positioning method of residual sparsity decision of claim 7, wherein said determining a difference metric from said theoretical probability distribution and an empirical probability distribution comprises: According to A difference metric value is determined, wherein, In order to measure the value of the difference, As an empirical probability distribution of the residual vector, Is the theoretical probability distribution of the reference noise.
9. A blind sparsity subspace backtracking homogeneity multi-source positioning system for residual sparsity decisions, the system comprising: The system comprises a sparse observation model construction module, a sparse observation model, a data acquisition module and a data acquisition module, wherein the sparse observation model construction module is used for carrying out gridding division on a monitored area, the monitored area comprises signal source nodes with unknown positions and sensing nodes with known positions, the signal source nodes transmit electromagnetic signals, the sensing nodes acquire the electromagnetic signals, acquire actual received signal intensities of the electromagnetic signals acquired by the sensing nodes, the actual received signal intensities are used as observation vectors, and a sparse observation model is constructed based on the grid positions of the signal source nodes and the observation vectors; The candidate support set determining module is used for taking the observation vector as a residual vector, calculating normalized inner products of the residual vector and each column of a sensing matrix in the sparse observation model to obtain a correlation numerical value sequence, determining an adaptive threshold according to the median absolute deviation and the mean value of the correlation numerical value sequence, determining candidate elements according to the comparison condition of the adaptive threshold and each element in the correlation numerical value sequence, and forming a candidate support set through grid positions of the candidate elements; The position support set determining module is used for screening the candidate support set based on the received signal strength estimated value of the grid position in the candidate support set to generate a position support set for the current iteration; The signal source position determining module is used for updating the residual vector according to the position support set used for the current iteration, determining theoretical probability distribution and experience probability distribution of the residual vector after updating the residual vector in each iteration, comparing the theoretical probability distribution with the experience probability distribution to obtain a difference metric value, and stopping the iteration process when the difference metric value meets a preset termination condition, wherein grid positions corresponding to the position support set of the current iteration are the final positioning result of the signal source.
10. A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the method of any one of claims 1 to 8.

Description

Blind sparsity subspace backtracking homogeneity multi-source positioning method for residual error sparsity decision Technical Field The invention relates to the technical field of wireless communication and signal processing, in particular to a blind sparsity subspace backtracking homogeneity multi-source positioning method for residual error divergence decision. Background With the wide application of radio equipment such as unmanned systems, electromagnetic environments are increasingly complex. In many application scenarios, it is desirable to locate non-cooperative wireless signal sources. When multiple non-cooperative signal sources transmit signals at the same time and frequency, the signals can be seriously aliased in space and frequency domains, which poses a threat to the normal operation and security of the wireless system. Traditional passive positioning technology, such as a positioning method based on parameters such as an arrival angle, an arrival time difference and the like, is mainly designed aiming at a single signal source, and is difficult to effectively separate and position each signal source when processing multi-source aliasing signals with the same frequency at the same time, so that the positioning effect is poor. In recent years, a sparse multisource positioning technology based on a compressed sensing theory provides a new idea for solving the problem. The technology performs rasterization division on a passive monitored physical area, and converts a complex positioning problem into a sparse signal recovery mathematical problem by using sparsity priori that a signal source generally occupies only a few grids in space distribution. However, existing positioning methods based on compressed sensing still have significant drawbacks. On the one hand, most of iterative solution algorithms directly use traditional greedy algorithms, such as orthogonal matching pursuit algorithm, only one position most relevant to the residual error is selected in each iteration, and misjudgment is easy to occur under the condition that multiple signal sources interfere with each other, so that positioning accuracy is insufficient. On the other hand, and also a more critical constraint, the iteration stop condition of these algorithms is heavily dependent on a priori information, and it is often necessary to know in advance the exact number of signal sources, i.e. sparsity, or to know the exact power level of the background noise. In a true non-cooperative localization scenario, the number of signal sources and noise level are unknown, dynamically varying, and this strong dependence on a priori information greatly limits the practicality and robustness of the prior art. Disclosure of Invention Based on the above, it is necessary to provide a blind sparsity subspace backtracking homogeneous multi-source positioning method for residual sparsity decision aiming at the above problems. A blind sparsity subspace backtracking homogeneity multi-source positioning method for residual divergence decision, the method comprises the following steps: And carrying out rasterization division on a monitored area, wherein the monitored area comprises signal source nodes with unknown positions and sensing nodes with known positions, the signal source nodes emit electromagnetic signals, the sensing nodes acquire the electromagnetic signals, acquire the actual received signal strength of the electromagnetic signals acquired by the sensing nodes, take the actual received signal strength as an observation vector, and construct a sparse observation model based on the grid positions of the signal source nodes and the observation vector. The observation vector is taken as a residual vector. Calculating normalized inner products of the residual vector and each column of a sensing matrix in the sparse observation model to obtain a correlation numerical sequence, determining an adaptive threshold according to the median absolute deviation and the mean value of the correlation numerical sequence, determining candidate elements according to the comparison condition of the adaptive threshold and each element in the correlation numerical sequence, and forming a candidate support set through grid positions of the candidate elements. And screening the candidate support set based on the received signal strength estimated value of the grid position in the candidate support set to generate a position support set for the iteration. And updating the residual vector according to the position support set for the current iteration. After updating the residual vector in each iteration, determining the theoretical probability distribution and the empirical probability distribution of the residual vector, comparing the theoretical probability distribution with the empirical probability distribution to obtain a difference metric value, and stopping the iteration process when the difference metric value meets a preset termination condition, wherein