Signal depth compression and direct positioning method, system, equipment and medium

Abstract

The invention discloses a signal depth compression and direct positioning method, system and device and a medium, and the method comprises the steps: arranging at least three observation stations in a two-dimensional space, carrying out the synchronous observation of a signal transmitted by a target, carrying out the sampling compression through a Hadamard matrix, and carrying out the quantification compression through a 1-bit quantification means, and achieving the depth compression of an observation signal; according to the method, the characteristics of a Hadamard matrix and the phase invariance of 1-bit compression are utilized, a complex signal reconstruction process is avoided, position parameters such as time delay do not need to be extracted, and a cost function is directly established from deeply compressed data by utilizing a reserved time delay relation so as to determine the position of a target; the system, the device and the medium can realize signal depth compression and direct positioning based on the method. The method can remarkably shorten the data volume of signal transmission, simplifies the processing process, and is suitable for positioning scenes with limited communication bandwidth and need of quick response; on the premise of guaranteeing the positioning precision, the signal data volume is greatly compressed, and the communication bandwidth requirement for transmitting the received signal of each station to the central station is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of electronic information technology, and in particular relates to a signal deep compression and direct positioning method, system, equipment and medium. Background Art

[0002] Compressed sensing (CS) theory is a novel signal acquisition and processing theory that breaks through the limitations of the traditional Nyquist sampling theorem. By utilizing the sparsity of the signal, it can accurately restore the signal at a much lower sampling rate than the Nyquist rate. This theory has shown great application potential in many fields such as wireless communications, radar systems, and image processing.

[0003] The latest research and patent applications have disclosed a series of direct positioning methods based on data compression. Patent application "A non-reconstruction direct positioning method based on mobile receiver data compression" (application number: CN202311041714.2) provides a non-reconstruction direct positioning method based on mobile receiver data compression. This method uses the Hadamard matrix characteristics to directly estimate the source position in the compressed measurement domain without the need for signal reconstruction and time difference and frequency difference parameter extraction. Patent application "A compressed sensing signal direct positioning method" (application number CN202311126528.9) provides a direct positioning method based on compressed sensing signals. By constructing a unique compressed sensing matrix, it is possible to directly locate the radiation source without reconstructing the compressed sensing signal. "Method, device and system for directly locating a stationary radiation source based on compressed observations", patent application (application number CN202311580347.3) provides a method, device and system for directly locating a stationary radiation source when the primary station is not compressed and the auxiliary station signal is compressed. There is no need for signal reconstruction and time difference and frequency difference parameter extraction, and direct positioning can be achieved in the mixed observation domain.

[0004] Although there are some direct positioning methods based on data compression, the data compression depth of these methods is insufficient and the compression rate is limited by the design of their compression matrix. They cannot meet the requirements of real-time data transmission and direct positioning when the communication bandwidth is limited. Their compression multiples and positioning methods need to be improved. Summary of the invention

[0005] In order to overcome the deficiencies of the above-mentioned prior art, the purpose of the present invention is to provide a signal deep compression and direct positioning method, system, equipment and medium, by arranging at least three observation stations in a two-dimensional space, synchronously observing the signal emitted by the target, using the Hadamard matrix for sampling compression and 1-bit quantization means for quantization compression, so as to realize deep compression processing of the observed signal; utilizing the characteristics of the Hadamard matrix and the phase invariance of 1-bit compression, avoiding the complex signal reconstruction process, without extracting position parameters such as time delay, directly establishing a cost function from the deeply compressed data using the retained time delay relationship to determine the position of the target; the present invention can significantly shorten the amount of signal transmission data and simplify the processing process, and is suitable for positioning scenarios with limited communication bandwidth and the need for rapid response. It can greatly compress the signal data volume while ensuring the positioning accuracy, and reduce the communication bandwidth requirements for each station to receive the signal and transmit it to the central station.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A signal deep compression and direct positioning method is provided. At least three observation stations are arranged in a two-dimensional space and initialized. The number of observation stations, the position coordinates of each observation station, the uniform sampling frequency and sampling number of all observation stations, the sampling compression ratio and the target position search space are determined. Then, the signals sent by the target are observed by multiple observation stations to form a set of received signal vectors containing information sent by the target. The Hadamard matrix is ​​used as the sampling compression matrix, and the Hadamard matrix is ​​multiplied by the received signal vector to complete the sampling compression. The sampled compressed signal is quantized by 1-bit to retain only the sign information of each sampling point to form a sampling-quantization joint deep compression signal. The observation station transmits the sampling-quantization joint deep compression signal to a central station for aggregating and processing data for processing. The Hadamard matrix is ​​used to restore the time delay relationship after sampling compression. The cost function about the target position is established through the sampling-quantization joint deep compression signal. Finally, the grid search method is used to exhaustively search the cost function in a predetermined search space to find the maximum eigenvalue of the cost function. The target position coordinates are estimated by solving the target position corresponding to the maximum eigenvalue of the cost function.

[0008] A signal deep compression and direct positioning method specifically comprises the following steps:

[0009] Step 1: Arrange at least three observation stations at random positions in the two-dimensional space and initialize them to determine the number of observation stations, the position coordinates of each observation station, the unified sampling frequency and number of samples for all observation stations, the sampling compression ratio of all observation stations, and the target position search space;

[0010] Step 2: All observation stations determined by step 1 synchronously observe the signals sent by the target at the set time interval, collect and form a set of received signal vectors containing the information sent by the target, and provide original signal data for the entire positioning process;

[0011] Step 3: Design the Hadamard matrix as the sampling compression matrix according to the sampling compression ratio determined in step 1 HM×N , multiply the Hadamard matrix by the received signal vector in step 2 to perform sampling compression; then perform 1-bit quantization compression on the sampled and compressed signal through a sign function to form a deep compressed signal of sampling-quantization joint compression;

[0012] Step 4: Each observation station transmits the deep compressed signal after sampling-quantization joint compression in step 3 to the central station for constructing the cost function about the target position.

[0013] Step 5: Use the grid search method to find start ,x end ]、[y start ,y end ] Iteratively search the cost function of step 4 to find the maximum eigenvalue of the cost function; find the target position corresponding to the maximum eigenvalue of the cost function, and then accurately estimate the coordinates of the target position.

[0014] The specific method of step 1 is: set the observation station at any position in the space, initialize and determine the number of observation stations to be L and L≤3, and the two-dimensional position coordinates of the lth observation station are p l , l=1,2,...,L; determine the unified sampling frequency F for all observation stations s And the number of sampling points N = 2 n , n is a non-negative integer; determine the compression ratio to be N / M and divisible by 4, where M = 2 m is the number of sampling points after compression, m is a non-negative integer; the target position search space is determined as the x-axis direction [x start ,x end ], y-axis direction [y start ,y end ].

[0015] The specific method of step 2 is: all observation stations synchronously observe the signal emitted by the target at a set time interval, and the target receiving signal observed by the lth observation station at time t is:

[0016] r l (t) = b l (t)s(t-τ l )+ω l (t), (1)

[0017] Among them, b l is the unknown complex path attenuation, s(t) is the target transmission signal, τ l is the delay of the signal from the transmitter to the observation station l, ω l (t) is the independent complex Gaussian white noise generated during the propagation process, satisfying And assume that for each observation station The sampling interval is T s =1 / F s The received signal vector formed after Nyquist sampling is:

[0018] r l =b l Q l s+ω l , (2)

[0019] The vectors and matrices in formula (2) are:

[0020]

[0021]

[0022] The specific method of step 3 is:

[0023] According to the compression ratio N / M determined in step 1, the first M rows of the N×N Hadamard matrix form the sampling compression matrix H M×N , with the following characteristics:

[0024]

[0025] The sampled compressed matrix H M×N The signal after sampling and compressing the original signal obtained in step 2 is:

[0026] y l =H M×N r l , (11)

[0027] Then, the sampled compressed signal is quantized and compressed by 1-bit, using the sign function of formula (12):

[0028]

[0029] The cost function constructed in step 4 is:

[0030] C(p) = N·[sign(Re{R c})+j·sign(Im{R c})], (13)

[0031] Among them, p is the search space, that is, the candidate target position in the set of coordinate points traversed when calculating the cost function.

[0032]

[0033] τ ij =τ i (p)-τ j (p),i,j∈{1,...L} (15)

[0034] It is a signal whose time delay relationship is restored from sampling compression using the relationship shown in formula (16):

[0035]

[0036] The method of step 5 is specifically as follows:

[0037] In the predetermined search space [x start ,x end ]、[y start ,y end ] to calculate and search the cost function; first set the search grid size to g, that is, the traversal step size, and according to g, the search space [x start ,x end ]、[y start ,y end ] is divided into [(x end -x start ) / g+1][(y end -y start) / g+1] grid points; take the coordinates p corresponding to the grid points respectively g Substitute into the calculated cost function C(p g ); search cost function C(p g )The coordinates corresponding to the maximum eigenvalue That is the target position coordinate estimate;

[0038] Perform iterative search according to the target position coordinate estimation process: Set the grid size of the first round of search to g1, and The cost function is calculated and searched to obtain the target position estimate by Determine a new search space for the center And the search grid size is g2; and so on, after multiple rounds of searching, the estimated result of the target coordinates is finally obtained.

[0039] A signal deep compression and direct positioning system, comprising:

[0040] The signal receiving module is used in step 2 to synchronously sample the signal from the target through the distributed observation stations to obtain the original received signal vector, which can ensure that the delay information contained in the signal of each station is consistent with the corresponding spatial propagation;

[0041] The deep compression module is used in step 3 to design the Hadamard matrix according to the compression rate determined by initialization in step 1, and to realize the sampling compression of the signal by left multiplying the received signal vector; to perform 1-bit quantization compression on the sampled compressed signal through the sign function; to obtain the sampled-quantized hybrid deep compressed signal, which significantly reduces the amount of transmitted data while ensuring that the delay relationship is not destroyed;

[0042] The cost function building module is used in step 4 to establish a cost function about the target position by correlating the deep compression signal with the target position in a time difference relationship, and obtain the cost function value corresponding to each grid point, which can reflect the probability of the target being located at the position. The grid point is used as a candidate solution for estimating the target position.

[0043] The position estimation module is used in step 5. By specifying the search space range and the layer-by-layer grid division strategy, the target position corresponding to the maximum value of the cost function output by the cost function construction module is searched, and the accurate estimate of the target position can be determined efficiently.

[0044] A signal deep compression and direct positioning device, specifically comprising:

[0045] Memory for storing computer programs;

[0046] A processor is used to implement the signal depth compression and direct positioning method described in steps 1 to 5 when executing the computer program.

[0047] A computer-readable storage medium stores a computer program, which, when executed by a processor, can perform signal depth compression and direct positioning based on the signal depth compression and direct positioning method described in steps 1 to 5.

[0048] Compared with the prior art, the present invention has the following advantages:

[0049] 1. The present invention uses Hadamard matrix and 1-bit quantization to achieve deep compression of received signals. If the original signal has N sampling points, each point is B-bit, then after being compressed into M points by Hadamard matrix and then quantized to 1-bit, the compression ratio can reach BN / M times, which greatly reduces the amount of data transmitted from each observation station to the central station, significantly reduces the pressure of data transmission, and is suitable for positioning scenarios with limited communication bandwidth such as extreme field and channel congestion.

[0050] 2. The present invention ensures that the time delay relationship is not destroyed during the compression process through the characteristics of the Hadamard matrix and 1-bit quantization. Even if the original signal is deeply compressed, the time delay information it contains can still be used to accurately estimate the target position.

[0051] 3. The present invention directly establishes a cost function about the target position through the time delay relationship contained in the deep compression signal. There is no need to restore the original signal from the compressed signal. The target position is directly estimated, which avoids the signal reconstruction and intermediate parameter estimation process, simplifies the data processing process, and improves the positioning efficiency.

[0052] In summary, the present invention arranges at least three observation stations in a two-dimensional space, synchronously observes the signals emitted by the target, uses the Hadamard matrix for sampling compression and the 1-bit quantization method for quantization compression, and realizes deep compression processing of the observed signal; utilizes the characteristics of the Hadamard matrix and the phase invariance of 1-bit compression to avoid the complex signal reconstruction process, and there is no need to extract position parameters such as time delay. A cost function is directly established from the deeply compressed data using the retained time delay relationship to determine the position of the target; it can significantly shorten the amount of signal transmission data and simplify the processing process, and is suitable for positioning scenarios with limited communication bandwidth and requiring rapid response; the signal data volume is greatly compressed while ensuring the positioning accuracy, thereby reducing the communication bandwidth requirements for each station to receive the signal and transmit it to the central station. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the target direct positioning scenario of the present invention.

[0054] Figure 2 It is a schematic diagram of an embodiment scenario of the present invention.

[0055] Figure 3 is the position spectrum of the cost function of the present invention.

[0056] Figure 4 It is a target positioning performance diagram under different compression ratios of the present invention.

[0057] Table 1 is a comparison table of positioning errors under different compression ratios and 1-bit quantization of the present invention.

[0058] Table 2 is a table of CRB values ​​under different signal-to-noise ratios of the present invention. DETAILED DESCRIPTION

[0059] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0060] like Figure 1As shown, a signal deep compression and direct positioning method is shown, at least three observation stations are arranged in a two-dimensional space and initialized: the number of observation stations, the position coordinates of each observation station, the unified sampling frequency and sampling number of all observation stations, the sampling compression ratio and the target position search space are determined; then, the signal sent by the target is observed by multiple observation stations to form a set of received signal vectors containing information sent by the target; the Hadamard matrix is ​​used as the sampling compression matrix, and the Hadamard matrix is ​​multiplied by the received signal vector to complete the sampling compression; the sampling compression signal is 1-bit quantized, and only the sign information of each sampling point is retained to form a sampling-quantization joint deep compression signal; the observation station transmits the sampling-quantization joint deep compression signal to a central station for aggregating and processing data for processing: the Hadamard matrix is ​​used to restore the time delay relationship after sampling compression; the cost function about the target position is established through the sampling-quantization joint deep compression signal; finally, the cost function is exhaustively searched in a predetermined search space by using a grid search method to find the maximum eigenvalue of the cost function; by solving the target position corresponding to the maximum eigenvalue of the cost function, the position coordinates of the target are accurately estimated.

[0061] like Figure 2 As shown, this embodiment takes three observation stations with known positions and one target that needs to be positioned in two-dimensional space as examples. The position coordinates of the three observation stations are (0,0), (2000,100) and (0,2500) (unit: meter); the real two-dimensional coordinates of the initial position of the target are (4500,3000) (unit: meter).

[0062] A signal deep compression and direct positioning method, the specific steps are as follows:

[0063] Step 1: Set the observation station to any position in the two-dimensional space, initialize the number of observation stations L = 3, and determine the position coordinates of each observation station to be (0,0), (2000,100) and (0,2500) (unit: meter); determine the sampling rate F of all observation stations s =2 23 Hz and the number of observed samples is N=512; the sampling compression ratio is determined to be 2, 4. Each original sampling point is a 64-bit double-precision floating point number. This embodiment needs to estimate the position of the target.

[0064] Step 2: The three observation stations simultaneously observe the signal emitted by the target to form the original observation signal vector r1∈á 512×1 、r2∈á 512×1 、r3∈á 512×1 There are 1536 sampling data points in total, and the total data size is 12kB.

[0065] Step 3: Construct the sample compressed Hadamard matrix H512×512 , when the compression ratio is 2 and 4, the compression matrix is ​​H 256×512 and H 128×512 The signals after sampling and compression are recorded as

[0066] y l =H 256×512 r l ∈á 256×1 ,(N / M=2),l=1,2,3

[0067] y l =H 128×512 r l ∈á 128×1 ,(N / M=4),l=1,2,3

[0068] Next, for y l After quantization compression, the compressed signal is recorded as

[0069]

[0070] It can be calculated that when the compression ratio is 2 and 4, the signal data size after sampling and quantization compression is 96Byte and 48Byte respectively, and the total compression multiples are 128 times and 256 times respectively.

[0071] Step 4: Substitute the compressed data into the cost function

[0072] C(p) = N·[sign(Re{R c})+j·sign(Im{R c})],

[0073] The cost function value at the grid point p can be calculated.

[0074] Step 5: Construct the search space [2500, 6500], [1000, 5000] to calculate and search the cost function, set the initial traversal step length to g = 1000, and reduce the step length to 1 / 5 of the previous round in each round. After multiple rounds of search, the target position coordinates are estimated. Among them, the cost function of the first round of search is as follows Figure 3 shown.

[0075] Through simulation experiments, five different compression strategies were compared, and the target position estimation performance of the non-reconstruction direct positioning method of the present invention and Hadamard compression was tested under different signal-to-noise ratios, and the target position estimation root mean square error was used as the evaluation criterion for evaluation. The comparison also included the theoretical minimum estimation error CRLB when only Hadamard matrix compression was used to better evaluate the accuracy loss caused by the fusion of 1-bit quantization. The simulation results are shown in Tables 1, 2 and Figure 4As shown. When the compression ratio is the same, the introduction of quantization compression can increase the data compression ratio by 64 times, while the performance loss is only 0.4 to 1 times. Compared with CRLB, after sampling-quantization hybrid compression, the root mean square error of the target position determined by the present invention can be controlled within 2 times of the theoretical performance boundary. It can be seen that the present invention significantly improves the data compression ratio compared with the prior art, and is 64 times higher than the Hadamard matrix sampling compression method; while greatly reducing the amount of transmitted data, it ensures a small performance loss, and the performance loss of the non-reconstruction direct positioning method of Hadamard compression does not exceed 1 times. Therefore, the present invention innovatively integrates Hadamard sampling and 1-bit quantization to achieve sampling-quantization deep compression of data, and at the same time proposes a direct positioning method based on sampling-quantization deep compressed data, which greatly improves the positioning accuracy and significantly reduces the positioning accuracy loss caused by data deep compression.

[0076] Table 1

[0077]

[0078] Table 2

[0079]

[0080] Furthermore, an embodiment of the present invention provides a signal depth compression and direct positioning system, including:

[0081] The signal receiving module is used in step 2 to synchronously sample the signal from the target through the distributed observation stations to obtain the original received signal vector, which can ensure that the delay information contained in the signal of each station is consistent with the corresponding spatial propagation;

[0082] The deep compression module is used in step 3 to design the Hadamard matrix according to the compression rate determined by initialization in step 1, and to realize the sampling compression of the signal by left multiplying the received signal vector; to perform 1-bit quantization compression on the sampled compressed signal through the sign function; to obtain the sampled-quantized hybrid deep compressed signal, which significantly reduces the amount of transmitted data while ensuring that the delay relationship is not destroyed;

[0083] The cost function building module is used in step 4 to establish a cost function about the target position by correlating the deep compression signal with the target position in a time difference relationship, and obtain the cost function value corresponding to each grid point, which can reflect the probability of the target being located at the position. The grid point is used as a candidate solution for estimating the target position.

[0084] The position estimation module is used in step 5. By specifying the search space range and the layer-by-layer grid division strategy, the target position corresponding to the maximum value of the cost function output by the cost function construction module is searched, and the accurate estimate of the target position can be determined efficiently.

[0085] Furthermore, an embodiment of the present invention provides a signal deep compression and direct positioning device, specifically comprising:

[0086] Memory for storing computer programs;

[0087] A processor is used to implement the signal depth compression and direct positioning method described in steps 1 to 5 when executing the computer program.

[0088] Furthermore, an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it can perform signal depth compression and direct positioning based on the signal depth compression and direct positioning method described in steps 1 to 5.

Claims

1. A signal deep compression and direct positioning method, characterized in that: At least three observation stations are arranged in a two-dimensional space and initialized: the number of observation stations, the location coordinates of each observation station, the unified sampling frequency and number of samples for all observation stations, the sampling compression ratio, and the target location search space are determined; then, the signals sent by the target are observed by multiple observation stations to form a set of received signal vectors containing information sent by the target; the Hadamard matrix is ​​used as the sampling compression matrix, and the Hadamard matrix is ​​multiplied by the received signal vector to complete the sampling compression; the sampled compressed signal is quantized by 1-bit, and only the sign information of each sampling point is retained to form a sampling-quantization joint deep compression signal; the observation station transmits the sampling-quantization joint deep compression signal to the central station for data aggregation and processing for processing: the Hadamard matrix is ​​used to restore the delay relationship after sampling compression; The cost function about the target position is established by sampling-quantizing the combined deep compression signal; finally, the grid search method is used to exhaustively search the cost function in the predetermined search space to find the maximum eigenvalue of the cost function; By solving the target position corresponding to the maximum eigenvalue of the cost function, the position coordinates of the target are estimated.

2. A signal depth compression and direct positioning method according to claim 1, characterized in that: The specific steps include: Step 1: Arrange at least three observation stations at random positions in the two-dimensional space and initialize them to determine the number of observation stations, the position coordinates of each observation station, the unified sampling frequency and number of samples for all observation stations, the sampling compression ratio of all observation stations, and the target position search space; Step 2: All observation stations determined by step 1 synchronously observe the signals sent by the target at the set time interval, collect and form a set of received signal vectors containing the information sent by the target, and provide original signal data for the entire positioning process; Step 3: Design the Hadamard matrix as the sampling compression matrix H according to the sampling compression ratio determined in step 1 M×N , multiply the Hadamard matrix by the received signal vector in step 2 to perform sampling compression; then perform 1-bit quantization compression on the sampled and compressed signal through a sign function to form a deep compressed signal of sampling-quantization joint compression; Step 4: Each observation station transmits the deep compressed signal after sampling-quantization joint compression in step 3 to the central station for constructing the cost function about the target position; Step 5: Use the grid search method to find start ,x end ]、[y start ,y end ] Iteratively search the cost function of step 4 to find the maximum eigenvalue of the cost function; find the target position corresponding to the maximum eigenvalue of the cost function, and then accurately estimate the coordinates of the target position.

3. A signal depth compression and direct positioning method according to claim 2, characterized in that: The specific method of step 1 is: set the observation station at any position in the space, initialize and determine the number of observation stations to be L and L≤3, and the two-dimensional position coordinates of the lth observation station are p l , l=1,2,...,L; determine the unified sampling frequency F for all observation stations s And the number of sampling points N = 2 n , n is a non-negative integer; determine the compression ratio to be N / M and divisible by 4, where M = 2 m is the number of sampling points after compression, m is a non-negative integer; the target position search space is determined as the x-axis direction [x start ,x end ], y-axis direction [y start ,y end ].

4. A signal depth compression and direct positioning method according to claim 2, characterized in that: The specific method of step 2 is: all observation stations synchronously observe the signal emitted by the target at a set time interval, and the target receiving signal observed by the lth observation station at time t is: r l (t)=b l (t)s(t-τ l )+ω l (t), (1) Among them, b l is the unknown complex path attenuation, s(t) is the target transmission signal, τ l is the delay of the signal from the transmitter to the observation station l, ω l (t) is the independent complex Gaussian white noise generated during the propagation process, satisfying And assume that for each observation station The sampling interval is T s =1 / F s The received signal vector formed after Nyquist sampling is: r l =b l Q l s+ω l , (2) The vectors and matrices in formula (2) are:

5. A signal depth compression and direct positioning method according to claim 2, characterized in that: The specific method of step 3 is: According to the compression ratio N / M determined in step 1, the first M rows of the N×N Hadamard matrix form the sampling compression matrix H M×N , with the following characteristics: The sampled compressed matrix H M×N The signal after sampling and compressing the original signal obtained in step 2 is: y l =H M×N r l , (11) Then, the sampled compressed signal is quantized and compressed by 1-bit, using the sign function of formula (12):

6. A signal depth compression and direct positioning method according to claim 1, characterized in that: The cost function constructed in step 4 is: C(p)=N·[sign(Re{R c })+j·sign(Im{R c })],(13) Among them, p is the search space, that is, the candidate target position in the set of coordinate points traversed when calculating the cost function. t ij =t i (p)-t j (p),i,j∈{1,...L} (15) It is a signal whose time delay relationship is restored from sampling compression using the relationship shown in formula (16):

7. A signal depth compression and direct positioning method according to claim 1, characterized in that: The method of step 5 is specifically as follows: In the predetermined search space [x start ,x end ]、[y start ,y end ] to calculate and search the cost function; first set the search grid size to g, that is, the traversal step size, and according to g, the search space [x start ,x end ]、[y start ,y end ] is divided into [(x end -x start ) / g+1][(y end -y start ) / g+1] grid points; take the coordinates p corresponding to the grid points respectively g Substitute into the calculated cost function C(p g ); search cost function C(p g )The coordinates corresponding to the maximum eigenvalue That is the target position coordinate estimation; Perform iterative search according to the target position coordinate estimation process: Set the grid size of the first round of search to g1, and The cost function is calculated and searched to obtain the target position estimate by Determine a new search space for the center And the search grid size is g2; and so on, after multiple rounds of searching, the estimated result of the target coordinates is finally obtained.

8. A signal depth compression and direct positioning system according to any one of claims 1 to 7, characterized in that: include: The signal receiving module is used in step 2 to synchronously sample the signal from the target through the distributed observation stations to obtain the original received signal vector, which can ensure that the delay information contained in the signal of each station is consistent with the corresponding spatial propagation; The deep compression module is used in step 3 to design the Hadamard matrix according to the compression rate determined by initialization in step 1, and to realize the sampling compression of the signal by left multiplying the received signal vector; to perform 1-bit quantization compression on the sampled compressed signal through the sign function; to obtain the sampled-quantized hybrid deep compressed signal, which significantly reduces the amount of transmitted data while ensuring that the delay relationship is not destroyed; The cost function building module is used in step 4 to establish a cost function about the target position by correlating the deep compression signal with the target position in a time difference relationship, and obtain the cost function value corresponding to each grid point, which can reflect the probability of the target being located at the position. The grid point is used as a candidate solution for estimating the target position. The position estimation module is used in step 5. By specifying the search space range and the layer-by-layer grid division strategy, the target position corresponding to the maximum value of the cost function output by the cost function construction module is searched, and the accurate estimate of the target position can be determined efficiently.

9. A signal deep compression and direct positioning device according to any one of claims 1 to 7, comprising: Memory for storing computer programs; A processor is used to implement the signal depth compression and direct positioning method described in steps 1 to 5 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it can perform signal depth compression and direct positioning based on the signal depth compression and direct positioning method described in steps 1 to 5.

Images