6g low-altitude target monitoring algorithm with integrated sensing and cooperation matching correction

Abstract

This invention discloses a 6G integrated sensing algorithm for low-altitude target monitoring with cooperative matching correction. The method first uses a priori symbol replicas of the transmitted signal to perform frequency domain point division preprocessing on the received wave, extracting the observation vector. For L-shaped arrays, a four-dimensional decoupled sensing dictionary matrix based on the space-time-frequency cascaded Kronecker product is constructed. An improved FISTA algorithm is used to calculate the full initial fingerprint set containing both cooperative and non-cooperative targets within the observation area. Motion parameters reported by multiple cooperative targets are received in real time, and the cooperative target index is locked using a many-to-many optimal matching algorithm based on the joint cost matrix. The sensing residuals of the cooperative targets are calculated to construct a spatial bias field, outputting high-precision four-dimensional detection parameters. This invention achieves closed-loop self-calibration of sensing and automatic parameter association, improving the detection accuracy and false alarm suppression capability for non-cooperative targets in complex multipath environments, and is suitable for high-reliability sensing scenarios such as 6G low-altitude surveillance.

Description

Technical Field

[0001] This invention relates to the fields of wireless communication and radar signal processing technology, specifically to a 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction. Background Technology

[0002] With the evolution of sixth-generation mobile communication technology (6G), integrated communication and sensing has become one of the core key technologies. ISAC aims to achieve high-speed communication and high-precision environmental sensing simultaneously using the same set of hardware and spectrum resources. In the high-frequency bands of 6G (such as millimeter waves and terahertz), the signals have extremely large bandwidth and high spatial directionality, providing a physical basis for achieving high-resolution target detection.

[0003] Currently, 6G sensing typically utilizes downlink OFDM communication signals as probe waveforms. Traditional sensing algorithms are mainly divided into two categories: The first type is matched filtering algorithms based on Discrete Fourier Transform (DFT / FFT). These algorithms are computationally simple, but their detection resolution is limited by signal bandwidth and observation window length, making it impossible to overcome the Rayleigh criterion. Furthermore, they are prone to severe sidelobe interference and masking effects in multi-target scenarios. The second type is subspace-based super-resolution algorithms (such as MUSIC and ESPRIT). While these algorithms can improve resolution, they are sensitive to noise, and in three-dimensional (distance-velocity-angle) joint estimation, the spatial spectrum search process involves enormous computational overhead (curse of dimensionality), making it difficult to meet the real-time requirements of 6G base station sensing.

[0004] In recent years, sparse reconstruction algorithms based on compressed sensing (CS) have been introduced into the sensing field. These algorithms discretize the probe space into an overcomplete dictionary and utilize the spatial sparsity of the signal to achieve super-resolution estimation. However, existing sparse reconstruction algorithms still face bottlenecks in 6G sensing scenarios, such as slow convergence speed, weak clutter suppression capability, and poor computational stability. Therefore, designing a super-resolution detection algorithm with fast convergence speed, strong robustness, and adaptive clutter suppression capability is a technical challenge that urgently needs to be solved in the current field of 6G integrated sensing. Summary of the Invention

[0005] The present invention proposes a 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction, which can at least solve one of the technical problems in the background art.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction and high-precision differential correction for non-cooperative targets includes the following steps: Step S1: The base station uses the transmitted 6G communication signal as a detection waveform to receive the echo signal reflected by the target, and preprocesses the echo signal to extract the observation vector containing target information. ; Step S2: Construct a perceptual sparse representation model, and convert the observed vectors Represented as a perceptual dictionary matrix With the sparse vector to be estimated and noise vector A linear combination, i.e. ; Step S3: Establish the objective function and use the improved FISTA algorithm to calculate the initial fingerprint coordinate set of all targets within the observation area. ; Step S4: The objective function is solved using an improved fast iterative shrinkage threshold algorithm. The optimal sparse solution is obtained by adaptively adjusting the iteration step size and weight matrix. ; Step S5: Real-time reception The motion state information reported by each cooperative target is spatiotemporally aligned with the base station's sensing time using a multiple time compensation method. Then, a many-to-many optimal matching algorithm is used to match the aligned cooperative target positions with the initial fingerprint coordinate set. Perform association to identify cooperative target indexes and non-cooperative target indexes; Step S6: Construct a spatial deviation field based on the perception residual of the cooperative target, use the deviation field to perform differential position correction on the initial fingerprint coordinates of the non-cooperative target, and output the final four-dimensional detection parameters.

[0007] As a preferred embodiment of the 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction described in this invention, wherein: the sensing dictionary matrix in step S2 It is an overcomplete basis matrix composed of four dimensions decoupled from range, velocity, azimuth, and pitch. Each atom corresponds to a preset physical parameter grid, which is used to achieve target range... ,speed Azimuth Pitch angle High-precision estimation of the perceptual dictionary matrix The construction based on the physical signal model includes the following steps: S21. Divide the detection area into a four-dimensional search grid: distance grid Resolution is , Signal bandwidth; speed grid Resolution is , For the speed of light, For carrier frequency, For observation time; azimuth grid Pitch angle grid Angular resolution depends on the antenna aperture; S22. In a 6G integrated sensing system, assuming the base station uses a combination of... The array elements distributed along the Z-axis and An L-shaped array consisting of elements distributed along the Y-axis emits signals containing... Subcarriers, Construct a two-dimensional angle steering vector from a signal frame of OFDM orthogonal symbols. :

[0008] in, For pitch dimension guide vector, For azimuth dimension guidance vector, For the frequency domain components, they are defined as follows:

[0009]

[0010]

[0011] in, It is the first Doppler frequency shift corresponding to each velocity grid, It is the target's relative radial velocity. It is the duration of a single OFDM symbol, including the cyclic prefix CP. It is the total number of symbols contained within a perception processing cycle; Time-domain components To account for the inter-carrier phase change caused by the time delay:

[0012] in, It is the subcarrier spacing, It is the first Round-trip delay corresponding to each distance grid cell It is the total number of subcarriers used, calculated using the following formula:

[0013] in, It is the radial distance of the detected target; S23. Combine the above three components into a single observation atom using the Kronecker product, and arrange all the parameters into a matrix, whose corresponding dictionary atom. Defined as:

[0014] Arrange all preset grid points in lexicographical order to construct an overcomplete matrix. Suppose there is Distance points, A speed point, , Angle points, matrix Its structure is as follows:

[0015] Among them, the number of rows Represents all resource sampling points in the system, number of columns This represents all possible combinations of preset target parameters, and is also the dimension of the sparse vector. To represent the field of complex numbers, since This matrix is ​​an overcomplete dictionary matrix.

[0016] As a preferred embodiment of the 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction described in this invention, step S3, in which the process of establishing the optimization objective function specifically includes the following steps: S31. Based on the observation vector extracted in step S1 The perceptual dictionary matrix constructed in step S2 Establish the residual sum of squares term based on the least squares criterion:

[0017] S32. Actively suppress clutter by utilizing static environmental characteristics and introducing an adaptive weighting factor into the sparse penalty term. Its initial value is jointly initialized by the environmental static clutter distribution and the prior probability of the reported location of the cooperative target, in order to construct a weighted average. Normative terms:

[0018] S33. The specific form of the objective function is:

[0019] in, It is the Euclidean distance between the predicted echo and the actual received echo. The smaller this value is, the more the estimated target parameters match physical observation. This term forces the solution vector to be penalized by punishing non-zero components. It has sparsity. It is the regularization parameter, when When the noise level is large, the algorithm tends to filter out noise. When the target size is small, the system tends to retain smaller targets. It corresponds to the first The adaptive weighting factor of each grid cell eliminates the traditional method by assigning different weights to different locations. The norm addresses the problem of excessive penalty for large coefficients and improves the dynamic range of parameter estimation; It is the dimension of the sparse vector. By solving this objective function, while suppressing static environmental clutter, a preliminary reconstruction of the full set of fingerprint coordinates, including both cooperative and non-cooperative targets, can be achieved. This provides a basic solution for subsequent matching and coordinate correction.

[0020] As a preferred embodiment of the 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction described in this invention, wherein: the full fingerprint coordinate set in step S3 The reconstruction and extraction process specifically includes: First, adaptive weight coefficients are performed. Dynamic updates, specifically:

[0021] in, For the first The sparse estimate of the next iteration; It is a positive smoothing constant; Secondly, the complex gain of the target is extracted, including the target's initial phase relative to the base station and channel fading characteristics:

[0022] in, The fingerprint point's specific reflection intensity and initial phase under the current electromagnetic environment were characterized. It is an index The corresponding dictionary column atoms; Ultimately, each objective is represented as a quintuple:

[0023] in, The number of cooperative targets in the scenario, The number of non-cooperative targets in the scenario, For the first Estimated distance to each target For the first The estimated relative velocity of each target For the first Estimated pitch angle of each target For the first The estimated azimuth of each target.

[0024] As a preferred embodiment of the 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction described in this invention, step S4 involves introducing a Nesterov acceleration factor and an adaptive step size mechanism to quickly solve the objective function, specifically including the following steps: S41. Utilize intermediate variables within the current iteration cycle. Perform gradient descent to reduce observation residuals and find terms that preserve data fidelity. Minimize the descent direction, calculate the vector difference between the current predicted echo and the actual received echo, and use the residual calculation formula:

[0025] Gradient descent is performed using the current iteration point to obtain auxiliary variables. :

[0026] in, Represents the temporary estimated vector after gradient descent. To determine the adaptive step size using the backtracking search method, For residuals, The step size is an intermediate variable introduced after the momentum acceleration term. It is not a fixed constant, but rather adaptively determined and continuously reduced. Until the following inequality is satisfied:

[0027] S42. After gradient descent, for the temporary vector Execution based on weighted The soft thresholding projection operation of the norm, for Each element in According to its corresponding weight Perform nonlinear contraction:

[0028] in, Used to preserve the phase information of the echo signal, It is a regularization factor that controls global sparsity. These are adaptive weights, whose values ​​are updated based on the results of the previous iteration: If the energy at a certain parameter position is lower than the threshold If the value is higher than the threshold, it is judged as noise or clutter and forced to return to zero; if it is higher than the threshold, its physical component is retained. S43, Construction Acceleration Factor Update intermediate variables This gives it an inertial effect;

[0029]

[0030] in, It is an accelerated sequence, with the initial value set to... ; Repeat the above steps until the preset convergence condition is met:

[0031] in, As a preset small constant, the final sparse solution output is This is used for subsequent target parameter mapping.

[0032] As a preferred embodiment of the 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction described in this invention, step S5 involves time alignment and many-to-many association matching of the full-scale sensing fingerprint using cooperative target reporting information, specifically including the following steps: S51, Real-time Acquisition The motion state parameters reported by the cooperative targets via the communication tributaries, including the reporting time, are as follows: 3D physical coordinates Velocity vector and acceleration vector ; Calculate the base station sensing and observation time Time difference with the reported time The predicted position of the cooperative target at the sensing moment is calculated using the following kinematic extrapolation formula. :

[0033] S52, Construct a dimension as Cost matrix Each element in the matrix Representing the The first sensing point and the first The matching cost between reported predicted locations is calculated using a multi-dimensional feature fusion method, as shown in the following formula:

[0034] in, , , Normalized weight coefficients for each feature dimension For the target total number, For the cost of spatial distance, For the price of speed, Fingerprint attribute cost; spatial distance cost Calculate sensing coordinates using Euclidean distance Predicted coordinates aligned with time Deviation between:

[0035] Speed ​​cost The base station senses the radial Doppler velocity of the target. The cooperative target reports a three-dimensional velocity vector. When constructing the cost, the reported velocity vector is projected onto the sensing radial direction:

[0036] Calculate the velocity residuals of the two:

[0037] Fingerprint attribute cost Using the complex-valued characteristic coefficients extracted in step 3 Compare with the pre-existing RCS model of the cooperation target;

[0038] in, Radar cross section pre-stored for cooperative targets, The system's overall gain and channel compensation factor; S53. Apply the bipartite graph optimal matching operator to the joint association cost matrix. Perform global optimization on the entire target fingerprint set. Establish an injective correlation between the predicted location set and the location set to determine The unique index of each cooperative target in the perception result is used to determine the remaining unmatched fingerprint points as non-cooperative targets.

[0039] As a preferred embodiment of the 6G integrated sensing low-altitude target monitoring algorithm for cooperative matching correction described in this invention, step S6 involves performing coordinate correction using the spatial correlation between the fingerprints of cooperative and non-cooperative targets after completing the cooperative target association. This specifically includes the following sub-steps: S61. System observation residual extraction: After completing many-to-many association matching, the system obtains... The two coordinates of the cooperative goal: one is the sensing coordinate calculated by the base station. One is the actual reported coordinates after time alignment. Calculate the systematic residual vector for each cooperative objective. :

[0040] This includes distance offset caused by asynchronous base station clocks. Velocity deviation caused by frequency drift And angular jitter caused by antenna array thermal noise or atmospheric refraction. , ; S62. Construction of Spatial Deviation Field: Due to the spatial coherence of 6G high-frequency channels, the environmental influences on closely spaced targets are highly correlated. The residuals of multiple cooperative targets are utilized... A dynamic error function covering the probe space is fitted:

[0041] in, Spatial location coordinates, It is an interpolation operator based on distance weights; S63. Coordinate Correction for Non-Cooperative Targets: For identified non-cooperative targets... Its initial fingerprint coordinates are Differential stripping is performed using spatial bias fields:

[0042] Based on the coordinates of the non-cooperative target, the expected deviation value of that location is retrieved from the error map, and this deviation is subtracted from the initial perception result to obtain the corrected accurate location information.

[0043] As a preferred embodiment of the 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction described in this invention, wherein: in step S4, the optimal sparse solution is used... The process of mapping to obtain the parameters of the probe target includes the following steps: Target filtering: Set amplitude threshold From the optimal sparse solution Extract the satisfying Non-zero element index ; Index decoupling: The indexes are arranged according to a preset lexicographical order. Inverse mapping to distance index Speed ​​Index and angle index :

[0044]

[0045]

[0046]

[0047] in, This refers to the azimuth point. The pitch angle is the number of points; Physical parameter conversion: Calculate the target's physical parameter values ​​based on the raster resolution corresponding to each dimension.

[0048]

[0049]

[0050]

[0051] in, For detecting distance, For detecting speed, Detect pitch angle, Detect azimuth angle; At the speed of light, For time-delay resolution, For Doppler resolution, , The angle step size.

[0052] The beneficial effects of this invention are: This invention achieves closed-loop self-calibration and automatic parameter association in sensing, improving the detection accuracy and false alarm suppression capability of non-cooperative targets in complex multipath environments, and is suitable for high-reliability sensing scenarios such as 6G low-altitude surveillance. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to the present invention. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0055] like Figure 1 As shown, this invention proposes a 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction. The method includes the following steps: Step S1: The base station uses the transmitted 6G communication signal as a detection waveform to receive the echo signal reflected by the target, and preprocesses the echo signal to extract the observation vector containing target information. ; Step S2: Construct a perceptual sparse representation model, and convert the observation vectors... Represented as a perceptual dictionary matrix With the sparse vector to be estimated and noise vector A linear combination, i.e. ; Step S3: Establish the objective function and use the improved FISTA algorithm to calculate the initial fingerprint coordinate set of all targets within the observation area. ; Step S4: The objective function is solved using an improved fast iterative threshold shrinkage algorithm. The optimal sparse solution is obtained by adaptively adjusting the iteration step size and weight matrix. ; Step S5: Real-time reception The motion state information reported by each cooperative target is used, and the motion state information is spatiotemporally aligned with the sensing time of the base station using a multiple time compensation method. Then, a many-to-many optimal matching algorithm is used to match the aligned cooperative target positions with the initial fingerprint coordinate set. Perform association to identify cooperative target indexes and non-cooperative target indexes; Step S6: Construct a spatial bias field based on the perception residual of the cooperative target, use the bias field to perform differential position correction on the initial fingerprint coordinates of the non-cooperative target, and output the final four-dimensional detection parameters.

[0056] In step S1 of the present invention, firstly, in the base station receiver antenna array element... Above, the first The OFDM symbol, the first The original echo signal received by each subcarrier It can be represented as:

[0057] in, The target complex reflection coefficient (including path loss and antenna gain). For the echo to reach the first Spatial phase characteristics of the antenna OFDM symbol period; The term refers to the inter-symbol phase shift caused by Doppler. The term refers to the inter-carrier phase shift caused by time delay. For known communication symbols transmitted by the base station, It is additive white Gaussian noise.

[0058] Then, using prior information about the base station's known transmitted data, a "frequency domain point division" is performed:

[0059] Finally, all antennas ( ), all symbols ( ), all subcarriers ( The observations are combined into a column vector:

[0060] in, .

[0061] In step S2 of the present invention, the perception dictionary matrix It is an overcomplete basis matrix composed of four dimensions decoupled from range, velocity, azimuth, and pitch, where each atom corresponds to a preset physical parameter grid. This is to achieve target range... ,speed and azimuth Pitch angle High-precision estimation of the perceptual dictionary matrix Based on the physical signal model, the detection area is first divided into a four-dimensional search grid: a distance grid. Resolution is , For signal bandwidth, speed grid Resolution is Azimuth grid Pitch angle First, the angular resolution depends on the antenna aperture; second, in a 6G integrated sensing system, assuming the base station uses a... The array elements distributed along the Z-axis and An L-shaped array consisting of elements distributed along the Y-axis emits signals containing... Subcarriers, A signal frame of OFDM orthogonal symbols; construct a two-dimensional angle steering vector. :

[0062] in, For pitch dimension guide vector, For azimuth dimension guidance vector, For the frequency domain components, they are defined as follows:

[0063]

[0064]

[0065] in, It is the first Doppler frequency shift corresponding to each velocity grid, It is the target's relative radial velocity. It is the duration of a single OFDM symbol, including the cyclic prefix CP. It is the total number of symbols contained within a perception processing cycle; Time-domain components To account for the inter-carrier phase change caused by the time delay:

[0066] in, It is the subcarrier spacing, It is the first Round-trip delay corresponding to each distance grid cell It is the total number of subcarriers used, calculated using the following formula:

[0067] in, It is the radial distance of the detected target; Finally, to transform the multidimensional parameter estimation problem into a sparse recovery problem, the three components mentioned above need to be combined into a single observation atom through the Kronecker product, and all possible parameter combinations need to be arranged into a matrix, with their corresponding dictionary atoms. Defined as:

[0068] Arrange all preset grid points in lexicographical order to construct an overcomplete matrix. Suppose there is Distance points, A speed point, , Angle points, matrix Its structure is as follows:

[0069] Among them, the number of rows Represents all resource sampling points in the system, number of columns This represents all possible combinations of preset target parameters, and is also the dimension of the sparse vector, because... This matrix is ​​an overcomplete dictionary matrix.

[0070] In step S3 of the present invention, the process of establishing a weighted regularized optimization model is as follows: based on the observation vector extracted in step 1... The perceptual dictionary matrix constructed in step S2 Establish the residual sum of squares term based on the least squares criterion:

[0071] Actively suppressing clutter by utilizing static environmental features, an adaptive weighting factor is introduced into the sparse penalty term. Its initial value is jointly determined by the environmental static clutter distribution and the prior probability of the reported location of the cooperative target, constructing a weighted average. Normative terms:

[0072] Finally, the objective function takes the following form:

[0073] in, This minimizes the Euclidean distance between the "predicted echo" and the "actual received echo." The smaller this term is, the better the estimated target parameters match physical observations. This term forces the solution vector to be penalized by punishing non-zero components. It has sparsity. It is the regularization parameter, when When the noise level is high, the algorithm tends to filter out noise; when... When the target size is small, the system tends to retain smaller targets. It corresponds to the first The adaptive weighting coefficients of each grid cell eliminate the traditional... The norm addresses the problem of excessive penalty for large coefficients, thereby improving the dynamic range of parameter estimation. It is the dimension of the sparse vector. By solving this objective function, while suppressing static environmental clutter, a preliminary reconstruction of the full set of fingerprint coordinates, including both cooperative and non-cooperative targets, can be achieved. This provides a basic solution for subsequent matching and coordinate correction.

[0074] In step S3 of the present invention, the complete fingerprint coordinate set The reconstruction and extraction process specifically includes: First, adaptive weight coefficients are performed. Dynamic updates, with the following specific rules:

[0075] in, For the first The sparse estimate in the nth iteration, at the 6th iteration In the next iteration, if a certain position The amount A large value indicates that there is a very high probability of a real target at that location. In the next iteration, the denominator increases, causing the weight to... Decreasing the weight means the algorithm weakens its suppression of that location, while for noisy regions... Close to 0, weight This becomes extremely large, thereby reducing noise interference in subsequent iterations. To prevent positive smoothing constants with a denominator of zero, to prevent numerical calculation instability caused by a denominator of 0, and to control the algorithm's sensitivity to small perturbations; Then, the complex gain of the target is extracted, which includes the target's initial phase relative to the base station and channel fading characteristics:

[0076] in, The fingerprint point's specific reflection intensity and initial phase under the current electromagnetic environment were characterized. It is an index The corresponding dictionary of atoms; ultimately, each target is represented as a quintuple:

[0077] in, The number of cooperative targets in the scenario, The number of non-cooperative targets in the scenario, For the first Estimated distance to each target For the first The estimated relative velocity of each target For the first Estimated pitch angle of each target For the first The estimated azimuth of each target.

[0078] In step S4 of the present invention, in order to solve the objective function quickly, a Nesterov acceleration factor and an adaptive step size mechanism are introduced. First, utilize the intermediate variables within the current iteration cycle. Perform gradient descent to reduce observation residuals and find terms that preserve data fidelity. Minimize the descent direction, calculate the vector difference between the current predicted echo and the actual received echo, and use the residual calculation formula:

[0079] Gradient descent is performed using the current iteration point to obtain auxiliary variables. :

[0080] in, Represents the temporary estimated vector after gradient descent. To determine the adaptive step size using the backtracking search method, To introduce the momentum acceleration term as an intermediate variable, and to ensure the stability of the algorithm and accelerate convergence, the step size is... It is not a fixed constant, but rather adaptively determined, starting with a large initial step size and continuously decreasing. Until the following inequality (Armijo condition) is satisfied:

[0081] Secondly, after gradient descent, in order to achieve signal sparsity and filter out noise interference, the temporary vector is... Execution based on weighted The soft thresholding projection operation of the norm, for Each element in According to its corresponding weight Perform nonlinear contraction:

[0082] in, Used to preserve the phase information of the echo signal, It is a regularization factor that controls global sparsity. These are adaptive weights, whose values ​​are updated based on the results of the previous iteration: If the energy at a certain parameter position is lower than the threshold If the value is higher than the threshold, it is judged as noise or clutter and forced to return to zero; if it is higher than the threshold, its physical component is retained. Finally, the acceleration factor is constructed. Update intermediate variables This gives it an inertial effect;

[0083]

[0084] in, It is an accelerated sequence, with the initial value set to... ; Repeat the above steps until the preset convergence condition is met:

[0085] in, As a preset small constant, the final sparse solution output is This is used for subsequent target parameter mapping.

[0086] In step S5 of the present invention, the time alignment and many-to-many association matching of the full-scale sensing fingerprints are performed using the cooperative target reporting information, specifically including the following steps: Multiple motion compensation alignment: real-time acquisition The motion state parameters reported by the cooperative targets via the communication tributaries, including the reporting time, are as follows: 3D physical coordinates Velocity vector and acceleration vector ; Calculate the base station sensing and observation time Time difference with the reporting time The predicted position of the cooperative target at the sensing moment is calculated using the following kinematic extrapolation formula. :

[0087] Due to the target set sensed by 6G base stations The target data contains both cooperative and non-cooperative objectives, and there are coordinate offsets due to systematic errors, making a simple one-to-one correspondence impossible. Therefore, it is necessary to construct a dimension... Cost matrix By weighted comparison of multidimensional physical features, a criterion is provided for subsequent global optimal association. Each element in the matrix... Representing the The first sensing point and the first The matching cost between reported predicted locations is calculated using a multi-dimensional feature fusion method, as shown in the following formula:

[0088] in, , , These are the normalized weight coefficients for each feature dimension. The target total number; Spatial distance cost Calculate sensing coordinates using Euclidean distance Predicted coordinates aligned with time Deviation between:

[0089] Speed ​​cost The base station senses the radial Doppler velocity of the target. The cooperative target reports a three-dimensional velocity vector. When constructing the cost, the reported velocity vector is first projected onto the sensing radial direction:

[0090] Calculate the velocity residuals of the two:

[0091] Fingerprint attribute cost Using the complex-valued characteristic coefficients extracted in step 3 Compare with the pre-existing RCS model of the cooperation target;

[0092] in, Radar cross section pre-stored for cooperative targets, The system's overall gain and channel compensation factor; Finally, the bipartite graph optimal matching operator is used to analyze the joint association cost matrix. Perform global optimization on the entire target fingerprint set. Establish an injective correlation between the predicted location set and the location set to determine The unique index of each cooperative target in the perception result is used to determine the remaining unmatched fingerprint points as non-cooperative targets.

[0093] In step S6 of the present invention, after the cooperative target association is completed, coordinate correction is performed using the spatial correlation between the fingerprints of the cooperative target and the non-cooperative target, specifically including the following sub-steps: System observation residual extraction: After completing the many-to-many association matching, the system obtains The two coordinates of the cooperative goal: one is the sensing coordinate calculated by the base station. One is the actual reported coordinates after time alignment. Calculate the systematic residual vector for each cooperative objective. :

[0094] This includes distance offset caused by asynchronous base station clocks. Velocity deviation caused by frequency drift And angular jitter caused by antenna array thermal noise or atmospheric refraction. , ; Construction of the spatial deviation field: Because 6G high-frequency band channels have spatial coherence, the environmental influences on targets that are close to each other are highly correlated. The residuals of multiple cooperative targets can be utilized... A dynamic error function covering the probe space is fitted:

[0095] in, Spatial location coordinates, It is an interpolation operator based on distance weights.

[0096] Coordinate correction for non-cooperative targets: For the identified non-cooperative targets Its initial fingerprint coordinates are Differential stripping is performed using spatial bias fields:

[0097] The algorithm retrieves the expected deviation value of the non-cooperative target from the error map based on the target's coordinates, and subtracts this deviation from the initial perception result to obtain the corrected accurate location information.

[0098] In step S4 of the present invention, based on the optimal sparse solution The process of mapping to obtain the parameters of the probe target includes the following steps: Target filtering: Set amplitude threshold From the optimal sparse solution Extract the satisfying Non-zero element index ; Index decoupling: Based on a preset lexicographical order, the index is decoupled... Inverse mapping to distance index Speed ​​Index and angle index :

[0099]

[0100]

[0101]

[0102] in, This refers to the azimuth point. The pitch angle is the number of points; Physical parameter conversion: Calculate the target's physical parameter values ​​based on the raster resolution corresponding to each dimension.

[0103]

[0104]

[0105]

[0106] in, For detecting distance, For detecting speed, Detect pitch angle, Detect azimuth angle; At the speed of light, For time-delay resolution, For Doppler resolution, , The angle step size.

[0107] This invention achieves closed-loop self-calibration and automatic parameter association in sensing, improving the detection accuracy and false alarm suppression capability of non-cooperative targets in complex multipath environments, and is suitable for high-reliability sensing scenarios such as 6G low-altitude surveillance.

[0108] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0109] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0110] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction, characterized in that, High-precision differential correction for non-cooperative targets includes the following steps: Step S1: The base station uses the transmitted 6G communication signal as a detection waveform to receive the echo signal reflected by the target, and preprocesses the echo signal to extract the observation vector containing target information. ; Step S2: Construct a perceptual sparse representation model, and convert the observed vectors Represented as a perceptual dictionary matrix With the sparse vector to be estimated and noise vector A linear combination, i.e. ; Step S3: Establish the objective function and use the improved FISTA algorithm to calculate the initial fingerprint coordinate set of all targets within the observation area. ; Step S4: The objective function is solved using an improved fast iterative shrinkage threshold algorithm. The optimal sparse solution is obtained by adaptively adjusting the iteration step size and weight matrix. ; Step S5: Real-time reception The motion state information reported by each cooperative target is spatiotemporally aligned with the base station's sensing time using a multiple time compensation method. Then, a many-to-many optimal matching algorithm is used to match the aligned cooperative target positions with the initial fingerprint coordinate set. Perform association to identify cooperative target indexes and non-cooperative target indexes; Step S6: Construct a spatial deviation field based on the perception residual of the cooperative target, use the deviation field to perform differential position correction on the initial fingerprint coordinates of the non-cooperative target, and output the final four-dimensional detection parameters.

2. The 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to claim 1, characterized in that: The perceptual dictionary matrix mentioned in step S2 It is an overcomplete basis matrix composed of four dimensions decoupled from range, velocity, azimuth, and pitch. Each atom corresponds to a preset physical parameter grid, which is used to achieve target range... ,speed Azimuth Pitch angle High-precision estimation of the perceptual dictionary matrix The construction based on the physical signal model includes the following steps: S21. Divide the detection area into a four-dimensional search grid: distance grid Resolution is , Signal bandwidth; speed grid Resolution is , For the speed of light, For carrier frequency, For observation time; azimuth grid Pitch angle grid Angular resolution depends on the antenna aperture; S22. In a 6G integrated sensing system, assuming the base station uses a combination of... The array elements distributed along the Z-axis and An L-shaped array consisting of elements distributed along the Y-axis emits signals containing... Subcarriers, Construct a two-dimensional angle steering vector from a signal frame of OFDM orthogonal symbols. : in, For pitch dimension guide vector, For azimuth dimension guidance vector, For the frequency domain components, they are defined as follows: in, It is the first Doppler frequency shift corresponding to each velocity grid, It is the target's relative radial velocity. It is the duration of a single OFDM symbol, including the cyclic prefix CP. It is the total number of symbols contained within a perception processing cycle; Time-domain components To account for the inter-carrier phase change caused by the time delay: in, It is the subcarrier spacing, It is the first Round-trip delay corresponding to each distance grid cell It is the total number of subcarriers used, calculated using the following formula: in, It is the radial distance of the detected target; S23. Combine the above three components into a single observation atom using the Kronecker product, and arrange all the parameters into a matrix, whose corresponding dictionary atom. Defined as: Arrange all preset grid points in lexicographical order to construct an overcomplete matrix. Suppose there is Distance points, A speed point, , Angle points, matrix Its structure is as follows: Among them, the number of rows Represents all resource sampling points in the system, number of columns This represents all possible combinations of preset target parameters, and is also the dimension of the sparse vector. To represent the field of complex numbers, since This matrix is ​​an overcomplete dictionary matrix.

3. The 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to claim 2, characterized in that: Step S3, establishing the objective function, specifically includes the following steps: S31. Based on the observation vector extracted in step S1 The perceptual dictionary matrix constructed in step S2 Establish the residual sum of squares term based on the least squares criterion: S32. Actively suppress clutter by utilizing static environmental characteristics and introducing an adaptive weighting factor into the sparse penalty term. Its initial value is jointly initialized by the environmental static clutter distribution and the prior probability of the reported location of the cooperative target, in order to construct a weighted average. Normative terms: S33. The specific form of the objective function is: in, It is the Euclidean distance between the predicted echo and the actual received echo; This term forces the solution vector to be penalized by punishing non-zero components. It has sparsity. It is a regularization parameter; It corresponds to the first Adaptive weighting factors for each grid cell; It is the dimension of the sparse vector.

4. The 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to claim 3, characterized in that: The complete fingerprint coordinate set in step S3 The reconstruction and extraction process specifically includes: First, adaptive weight coefficients are performed. Dynamic updates, specifically: in, For the first The sparse estimate of the next iteration; It is a positive smoothing constant; Secondly, the complex gain of the target is extracted, including the target's initial phase relative to the base station and channel fading characteristics: in, The fingerprint point's specific reflection intensity and initial phase under the current electromagnetic environment were characterized. It is an index The corresponding dictionary column atoms; Ultimately, each objective is represented as a quintuple: in, The number of cooperative targets in the scenario, The number of non-cooperative targets in the scenario, For the first Estimated distance to each target For the first The estimated relative velocity of each target For the first Estimated pitch angle of each target For the first The estimated azimuth of each target.

5. The 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to claim 1, characterized in that: In step S4, in order to solve the objective function quickly, a Nesterov acceleration factor and an adaptive step size mechanism are introduced, which specifically includes the following steps: S41. Utilize intermediate variables within the current iteration cycle. Perform gradient descent to reduce observation residuals and find terms that preserve data fidelity. Minimize the descent direction, calculate the vector difference between the current predicted echo and the actual received echo, and use the residual calculation formula: Gradient descent is performed using the current iteration point to obtain auxiliary variables. : in, Represents the temporary estimated vector after gradient descent. To determine the adaptive step size using the backtracking search method, For residuals, The step size is an intermediate variable introduced after the momentum acceleration term. It is not a fixed constant, but rather adaptively determined and continuously reduced. Until the following inequality is satisfied: S42. After gradient descent, for the temporary vector Execution based on weighted The soft thresholding projection operation of the norm, for Each element in According to its corresponding weight Perform nonlinear contraction: in, Used to preserve the phase information of the echo signal, It is a regularization factor that controls global sparsity. These are adaptive weights, whose values ​​are updated based on the results of the previous iteration: If the energy at a certain parameter position is lower than the threshold If the value is higher than the threshold, it is judged as noise or clutter and forced to return to zero; if it is higher than the threshold, its physical component is retained. S43, Construction Acceleration Factor Update intermediate variables This gives it an inertial effect; in, It is an accelerated sequence, with the initial value set to... ; Repeat the above steps until the preset convergence condition is met: in, As a preset small constant, the final sparse solution output is This is used for subsequent target parameter mapping.

6. The 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to claim 1, characterized in that: In step S5, the information reported by the cooperative target is used to perform time alignment and many-to-many association matching on the full-scale perceived fingerprints, specifically including the following steps: S51, Real-time Acquisition The motion state parameters reported by the cooperative targets via the communication tributaries, including the reporting time, are as follows: 3D physical coordinates Velocity vector and acceleration vector ; Calculate the base station sensing and observation time Time difference with the reported time The predicted position of the cooperative target at the sensing moment is calculated using the following kinematic extrapolation formula. : S52, Construct a dimension as Cost matrix Each element in the matrix Representing the The first sensing point and the first The matching cost between reported predicted locations is calculated using a multi-dimensional feature fusion method, as shown in the following formula: in, , , Normalized weight coefficients for each feature dimension For the target total number, For the cost of spatial distance, For the price of speed, Fingerprint attribute cost; spatial distance cost Calculate sensing coordinates using Euclidean distance Predicted coordinates aligned with time Deviation between: Speed ​​cost The base station senses the radial Doppler velocity of the target. The cooperative target reports a three-dimensional velocity vector. When constructing the cost, the reported velocity vector is projected onto the sensing radial direction: Calculate the velocity residuals of the two: Fingerprint attribute cost Using the complex-valued characteristic coefficients extracted in step 3 Compare with the pre-existing RCS model of the cooperation target; in, Radar cross section pre-stored for cooperative targets, The system's overall gain and channel compensation factor; S53. Apply the bipartite graph optimal matching operator to the joint association cost matrix. Perform global optimization on the entire target fingerprint set. Establish an injective correlation between the predicted location set and the location set to determine The unique index of each cooperative target in the perception result is used to determine the remaining unmatched fingerprint points as non-cooperative targets.

7. The 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to claim 1, characterized in that: In step S6, after completing the association of cooperative targets, coordinate correction is performed using the spatial correlation between the fingerprints of cooperative and non-cooperative targets. This specifically includes the following sub-steps: S61. System observation residual extraction: After completing many-to-many association matching, the system obtains... The two coordinates of the cooperative goal: one is the sensing coordinate calculated by the base station. One is the actual reported coordinates after time alignment. Calculate the systematic residual vector for each cooperative objective. : This includes distance offset caused by asynchronous base station clocks. Velocity deviation caused by frequency drift And angular jitter caused by antenna array thermal noise or atmospheric refraction. , ; S62. Construction of Spatial Deviation Field: Due to the spatial coherence of 6G high-frequency channels, the environmental influences on closely spaced targets are highly correlated. The residuals of multiple cooperative targets are utilized... A dynamic error function covering the probe space is fitted: in, Spatial location coordinates, It is an interpolation operator based on distance weights; S63. Coordinate Correction for Non-Cooperative Targets: For identified non-cooperative targets... Its initial fingerprint coordinates are Differential stripping is performed using spatial bias fields: Based on the coordinates of the non-cooperative target, the expected deviation value of that location is retrieved from the error map, and this deviation is subtracted from the initial perception result to obtain the corrected accurate location information.

8. The 6G integrated sensing low-altitude target monitoring algorithm with cooperative matching correction according to claim 5, characterized in that: In step S4, the optimal sparse solution is used. The process of mapping to obtain the parameters of the probe target includes the following steps: Target filtering: Set amplitude threshold From the optimal sparse solution Extract the satisfying Non-zero element index ; Index decoupling: The indexes are arranged according to a preset lexicographical order. Inverse mapping to distance index Speed ​​Index and angle index : in, This refers to the azimuth point. The pitch angle is the number of points; Physical parameter conversion: Calculate the target's physical parameter values ​​based on the raster resolution corresponding to each dimension. in, For detecting distance, For detecting speed, Detect pitch angle, Detect azimuth angle; At the speed of light, For time-delay resolution, For Doppler resolution, , The angle step size.

Images