Low-altitude unmanned aerial vehicle sensing method based on 5G signal and related equipment

Abstract

The invention discloses a low-altitude unmanned aerial vehicle sensing method based on a 5G signal and related equipment. The method comprises the following steps: acquiring a channel frequency response measurement value; performing static background elimination on the channel frequency response measurement value to obtain a dynamic channel component; obtaining dictionary incoherence and channel occupancy density of the channel segments to obtain channel scores; when the channel score is greater than or equal to a first preset threshold value, performing global estimation on the dynamic channel component to obtain a first target state estimation value at the current moment; when the channel score is smaller than a first preset threshold value and the target unmanned aerial vehicle does not exist at the previous moment, returning to the step of obtaining the channel frequency response measurement value; and when the channel score is smaller than a first preset threshold value and the target unmanned aerial vehicle exists at the previous moment, performing local estimation on the dynamic channel component to obtain a second target state estimation value at the current moment. The method can improve the continuity and accuracy of unmanned aerial vehicle perception, and can be widely applied to the technical field of unmanned aerial vehicle state estimation.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) state estimation technology, and in particular to a low-altitude UAV sensing method and related equipment based on 5G signals. Background Technology

[0002] The healthy development of the low-altitude economy depends on orderly operation and regulation, and the prerequisite for operation and regulation is the construction of low-altitude infrastructure. With the development of Integrated Sensing and Communication (ISAC) technology, the use of widely deployed 5G base stations (Next Generation NodeB, gNB) to monitor drones can bring many benefits: it does not require dedicated sensing systems, but can provide universal sensing capabilities using only existing communication infrastructure, saving the cost of deploying dedicated equipment; its wide-area coverage capability (including indoor and outdoor) can expand the monitoring range.

[0003] Currently, research on sensing fusion based on 5G New Radio (NR) and 4G Long Term Evolution (LTE) largely focuses on target path detection using reference signals. In LTE systems, a 1kHz Cell Reference Signal (CRS) is typically used to extract target distance and Doppler information. Although LTE CRS-based sensing schemes can acquire target distance and Doppler information, the CSI-RS configuration used in 5G NR to improve spectrum efficiency is sparser, limiting its sensing capabilities. Some studies have attempted to fuse multiple reference signals (such as CSI-RS, DM-RS, and PRS) to overcome these limitations, but DM-RS suffers from temporal unpredictability due to the coexistence of bursty data streams, and the availability of PRS depends on the user equipment's (UE) positioning request. Furthermore, real-time communication traffic fluctuations cause sensing performance to fluctuate. Existing strategies that use resource occupancy to assess UAV detection can mitigate the impact of communication fluctuations on sensing to some extent, but in actual gNB deployments, for a given gNB, the proportion of azimuth angles with downlink resource utilization density exceeding 7.1% is less than 5%. Strictly limiting the sensing function to high resource consumption ranges will lead to frequent and prolonged sensing interruptions, severely impacting the continuity and reliability of drone monitoring. Summary of the Invention

[0004] In view of this, the main objective of the embodiments of the present invention is to provide a low-altitude UAV sensing method and related equipment based on 5G signals, in order to solve at least one of the problems of the prior art. The present invention can improve the continuity and accuracy of UAV sensing.

[0005] To achieve the above objectives, one aspect of the present invention provides a low-altitude unmanned aerial vehicle (UAV) sensing method based on 5G signals, the method comprising: Acquire the reflected echo of the spontaneous signal, and obtain the channel frequency response measurement value based on the reflected echo; Static background elimination is performed on the channel frequency response measurements to obtain the dynamic channel components; Based on the channel segments of the dynamic channel components, the dictionary incoherence and channel occupancy density of the channel segments are obtained to obtain the channel score; The channel score is compared with a first preset threshold. When the channel score is greater than or equal to the first preset threshold, the dynamic channel component is globally estimated using a two-dimensional orthogonal matching pursuit algorithm to obtain the first target state estimate at the current moment. When the channel score is less than the first preset threshold and there was no target UAV in the previous moment, return to the step of obtaining the channel frequency response measurement value based on the reflected echo; When the channel score is less than the first preset threshold and the target UAV existed in the previous time, the dynamic channel components are locally estimated to obtain the second target state estimate at the current time.

[0006] In some embodiments, acquiring the reflected echo of the spontaneous signal and obtaining the channel frequency response measurement value based on the reflected echo includes the following steps: The reflected echo of the spontaneous signal is obtained by receiving the antenna; The reflected echo is subjected to frequency domain demodulation and synchronization processing to construct a received signal model; Based on the spontaneous signal and the received signal model, the channel frequency response measurement value is obtained.

[0007] In some embodiments, performing static background elimination on the channel frequency response measurement to obtain dynamic channel components includes the following steps: The channel frequency response measurement values ​​are subjected to mean filtering through a long time-domain window to obtain the static component; The dynamic channel component is obtained by subtracting the static component from the channel frequency response measurement.

[0008] In some embodiments, obtaining the dictionary incoherence and channel occupancy density of a channel segment based on the channel segment of the dynamic channel component to obtain a channel score includes the following steps: Obtain the channel segment of the dynamic channel component; Based on the resource grid corresponding to the channel segment, construct a distance and Doppler dictionary; The dictionary incoherence is obtained based on the distance and the Doppler dictionary; Obtain the channel occupancy density of the channel segment; The channel score is obtained by weighting the dictionary incoherence and the channel occupancy density.

[0009] In some embodiments, obtaining the dictionary incoherence based on the distance and the Doppler dictionary includes the following steps: Select a reference atom from the distance and Doppler dictionary; Obtain the first neighboring atom that is adjacent to the reference atom along the distance dimension; Obtain the second neighbor atom adjacent to the reference atom along the Doppler dimension; Obtain the absolute value of the first inner product between the reference atom and the first neighboring atom; Obtain the absolute value of the second inner product between the reference atom and the second neighboring atom; The dictionary incoherence is obtained by performing a maximum value function operation on the absolute values ​​of the first inner product and the second inner product.

[0010] In some embodiments, obtaining the channel occupancy density of the channel segment includes the following steps: Obtain the first number of non-empty resource elements in the channel segment; Obtain the second number of resource elements in the resource grid; The channel occupancy density is obtained by obtaining the ratio of the first quantity to the second quantity.

[0011] In some embodiments, comparing the channel score with a first preset threshold, and when the channel score is greater than or equal to the first preset threshold, performing a global estimation of the dynamic channel components using a two-dimensional orthogonal matching pursuit algorithm to obtain a first target state estimate at the current moment, includes the following steps: Obtain the third quantity of the dynamic channel components and construct the target path reconstruction-sparse compressed sensing problem; Based on the resource grid corresponding to the channel segment, construct a distance and Doppler dictionary; The channel frequency response measurements are masked, and the residual matrix is ​​initialized based on the masked channel frequency response measurements. In the distance and Doppler dictionary, select the first matching atom that has the greatest correlation with the current residual matrix, and add the index of the first matching atom to the support set; Based on the support set described above, the coefficient estimates of the sparse vectors are obtained by the least squares method; The residual matrix is ​​updated based on the coefficient estimates; When the residual change of the updated residual matrix is ​​greater than or equal to the second preset threshold and the maximum number of iterations has not been reached, return to the step of selecting the first matching atom with the greatest correlation to the current residual matrix in the distance and Doppler dictionary; When the residual change of the updated residual matrix is ​​less than the second preset threshold, or when the maximum number of iterations is reached, the reconstructed sparse matrix is ​​output, and the first target state estimate is obtained based on the reconstructed sparse matrix.

[0012] In some embodiments, when the channel score is less than the first preset threshold and the target UAV existed in the previous time, performing local estimation of the dynamic channel components to obtain a second target state estimate for the current time includes the following steps: Based on the resource grid corresponding to the channel segment, construct a distance and Doppler dictionary; Based on the estimated state of the third target at the previous moment, a pre-defined initial bounded region is set. Based on the movement direction of the target UAV, the initial bounded area is adjusted to obtain the target bounded area; Within the bounded region of the target, extract the corresponding subset of atoms from the distance and the Doppler dictionary; The correlation between the dynamic channel component and each candidate atom in the atomic subset is obtained, and the candidate atom with the highest correlation is selected to obtain the second matching atom; Based on the grid coordinates of the second matched atom, the first local state estimate at the current moment is obtained; The first local state estimate is processed by Kalman filtering to obtain the second local state estimate at the current time. The second local state estimate is calibrated by interval to obtain the second target state estimate at the current time.

[0013] To achieve the above objectives, another aspect of this invention provides a low-altitude unmanned aerial vehicle (UAV) sensing device based on 5G signals, the device comprising: The data acquisition module is used to acquire the reflected echo of the spontaneous signal and to acquire the channel frequency response measurement value based on the reflected echo. The preprocessing module is used to perform static background elimination on the channel frequency response measurement values ​​to obtain dynamic channel components; The channel score acquisition module is used to obtain the dictionary incoherence and channel occupancy density of the channel segments based on the channel segments of the dynamic channel component, and to obtain the channel score. The global estimation module is used to compare the channel score with a first preset threshold. When the channel score is greater than or equal to the first preset threshold, the dynamic channel component is globally estimated using a two-dimensional orthogonal matching pursuit algorithm to obtain the first target state estimate at the current time. The first local estimation module is used to return to the step of obtaining the channel frequency response measurement value based on the reflected echo when the channel score is less than the first preset threshold and there was no target UAV in the previous moment. The second local estimation module is used to perform local estimation of the dynamic channel components when the channel score is less than the first preset threshold and the target UAV existed in the previous time, so as to obtain the second target state estimate value at the current time.

[0014] To achieve the above objectives, another aspect of the present invention provides an electronic device, the electronic device including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method described above.

[0015] To achieve the above objectives, another aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods described above.

[0016] To achieve the above objectives, another aspect of the present invention provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions to cause the computer device to perform the aforementioned method.

[0017] The embodiments of the present invention include at least the following beneficial effects: The present invention provides a low-altitude UAV sensing method and related equipment based on 5G signals. This scheme acquires the reflected echo of spontaneous signals, obtains channel frequency response measurements based on the reflected echo, and provides a data basis for subsequent steps; static background elimination is performed on the channel frequency response measurements to obtain dynamic channel components, preparing for detecting the state of the target UAV; based on the channel segments of the dynamic channel components, the dictionary incoherence and channel occupancy density of the channel segments are obtained to obtain a channel score, which is used to measure the signal quality within the current data processing window, effectively mitigating the impact of communication data fluctuations; the channel score is then compared with a first preset threshold... In the comparison process, when the channel score is greater than or equal to the first preset threshold, the dynamic channel components are globally estimated using a two-dimensional orthogonal matching pursuit algorithm. This enables continuous target UAV perception even under conditions of non-uniform and fluctuating communication resource occupancy, obtaining the first target state estimate at the current moment. When the channel score is less than the first preset threshold and there was no target UAV in the previous moment, the process returns to the step of obtaining the channel frequency response measurement based on the reflected echo. When the channel score is less than the first preset threshold and there was a target UAV in the previous moment, the dynamic channel components are locally estimated to obtain the second target state estimate at the current moment. By narrowing the bounded search area, the accuracy of target UAV perception is improved. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the processing procedure of the low-altitude UAV perception method based on 5G signals provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of adjacent atoms on a distance-Doppler grid provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the distribution of target state estimation error with channel score according to an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating an application scenario of the low-altitude UAV sensing method based on 5G signals provided in an embodiment of the present invention. Figure 5 This is a flowchart of a low-altitude UAV perception algorithm based on 5G signals provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this invention; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this invention as detailed in the appended claims.

[0021] It should be noted that although functional modules are divided in the system diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first / S100" and "second / S200" in the specification, claims, and the foregoing drawings may be used herein to describe various concepts, but unless specifically stated otherwise, these concepts are not limited by these terms. These terms are used only to distinguish one concept from another. For example, first information may also be referred to as second information without departing from the scope of the embodiments of the invention, and similarly, second information may also be referred to as first information. Depending on the context, the words "if" or "when" as used herein may be interpreted as "when," "in response to a determination," or "in the event of a determination."

[0022] The terms “at least one,” “multiple,” “each,” “any,” etc., used in this invention, “at least one” includes one, two, or more than two; “multiple” includes two or more than two; “each” refers to each of the corresponding multiple; and “any” refers to any one of the multiple.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0024] Currently, research on the fusion of sensing and communication based on 5G New Radio (NR) and 4G Long Term Evolution (LTE) largely focuses on using reference signals for target path detection. In LTE systems, a 1 kHz Cell Reference Signal (CRS) is typically used to extract target distance and Doppler information. In contrast, 5G NR employs a more sparsely configured Channel State Information Reference Signal (CSI-RS) to improve spectral efficiency. With a 30 kHz subcarrier spacing configuration, the minimum time interval of CSI-RS is four time slots, and its maximum repetition frequency is 500 Hz.

[0025] In addition to CSI-RS, 5G NR systems also include downlink synchronization signals such as the Synchronization Signal Block (SSB), with a frequency range of up to 7.2 MHz and a repetition frequency typically of 50 Hz. However, the low repetition frequencies of these reference signals (e.g., 200 Hz for CSI-RS and 50 Hz for SSB) fundamentally limit the ability to track the Doppler shift of high-speed targets. To alleviate this limitation, 3GPP introduced Positioning Reference Signal (PRS) in Release 16, which provides higher time-frequency resolution and resource element density through diagonal or interleaved resource element (RE) configuration. However, the performance improvement of PRS comes at the cost of increased system resource overhead, reflecting the fundamental trade-off between spectral efficiency and sensing performance in NR system resource allocation strategies. Although periodically spaced reference signals facilitate sensing functions, they suffer from velocity and range ambiguity issues in high-speed or long-range target detection. Similarly, periodic intervals in the frequency domain can also lead to distance ambiguity. When the target distance is greater than the unambiguous time delay of a single-station sensing system, targets at greater distances will appear to fold into closer ranges, and target drones may be obscured by strong targets at close range. Although the defects of velocity and distance ambiguity can be mitigated by reducing the intervals of the reference signal in the time-frequency dimension, this will incur spectrum resource overhead. Therefore, this reflects a fundamental conflict in sensing fusion systems: communication prioritizes resource efficiency, while improving sensing performance requires more resources with a wide and dense time-frequency distribution.

[0026] In view of this, this invention provides a low-altitude UAV sensing method and related equipment based on 5G signals. This method proposes an incoherence... - A density score quantification metric is used to evaluate channel quality. Then, based on the instantaneous channel score, an adaptive hybrid estimation algorithm for the channel based on two-dimensional orthogonal matching pursuit is designed, which can achieve continuous target UAV perception under non-uniform and fluctuating communication resource occupancy.

[0027] Figure 1 This is an optional flowchart of a low-altitude UAV sensing method based on 5G signals provided in an embodiment of the present invention. Figure 1 The method may include, but is not limited to, steps S100 to S600: Step S100: Obtain the reflected echo of the spontaneous signal, and obtain the channel frequency response measurement value based on the reflected echo; Step S200: Static background elimination is performed on the channel frequency response measurement to obtain the dynamic channel components; Step S300: Based on the channel segments of the dynamic channel components, obtain the dictionary incoherence and channel occupancy density of the channel segments to obtain the channel score; Step S400: Compare the channel score with the first preset threshold. When the channel score is greater than or equal to the first preset threshold, perform a global estimation of the dynamic channel components using the two-dimensional orthogonal matching pursuit algorithm to obtain the first target state estimate at the current moment. Step S500: When the channel score is less than the first preset threshold and there was no target UAV in the previous moment, return to the step of obtaining the channel frequency response measurement value based on the reflected echo. Step S600: When the channel score is less than the first preset threshold and there was a target UAV in the previous moment, perform local estimation of the dynamic channel components to obtain the second target state estimate value at the current moment.

[0028] In some embodiments, step S100 may include, but is not limited to, steps S110 to S130: Step S110: Obtain the reflected echo of the spontaneous signal through the receiving antenna; Step S120: Perform frequency domain demodulation and synchronization processing on the reflected echo to construct a received signal model; Step S130: Based on the spontaneous signal and received signal model, obtain the channel frequency response measurement value.

[0029] In step S110 of some embodiments, when the base station is conducting downlink communication with surrounding users, the gNB uses an additional receiving antenna to capture the reflected echo of spontaneous signals.

[0030] In step S120 of some embodiments, after the received signal (i.e., the reflected echo) is demodulated in the frequency domain and synchronized with the Synchronization Signal Block (SSB), a received signal model can be obtained. For example, the received signal model is as follows: ; in, Indicates the first The subcarrier, the first The received signal at each Orthogonal Frequency Division Multiplexing (OFDM) symbol; This represents the true value of the channel frequency response; This refers to the reference signal or data signal transmitted by the base station, i.e., the spontaneously generated signal; This represents additive white Gaussian noise.

[0031] In step S130 of some embodiments, under the architecture of a single-site sensing base station, since the signals emitted by all base stations themselves (i.e., spontaneous signals) are known information, the base station can obtain the channel frequency response values ​​of all non-empty resource elements (including pilots and data units) within the scheduling bandwidth. For example, after least squares (LS) channel estimation processing, the channel frequency response (CFR) measurement value can be obtained. Due to the presence of additive white Gaussian noise, the channel frequency response measurement value is a noisy version of the true value, and the formula used includes: ; in, This represents the measured value of the channel frequency response.

[0032] In some embodiments, step S200 may include, but is not limited to, steps S210 to S220: Step S210: The channel frequency response measurement is subjected to mean filtering through a long time-domain window to obtain the static component; Step S220: Subtract the static component from the channel frequency response measurement to obtain the dynamic channel component.

[0033] In step S210 of some embodiments, the channel frequency response measurement is mean-filtered through a long time-domain window (e.g., 100ms). Within this long time-domain window, the multipath components of the moving target UAV will experience more than The phase change of the phase component and its interference effect are averaged over time, while the static component (such as the static reflection component of the ground and buildings) remains stable, so the static component can be extracted.

[0034] In step S220 of some embodiments, for each subcarrier, the static component is calculated as the arithmetic mean of all occupied resource elements within a long time-domain window. By subtracting the static component from the channel frequency response measurement, the time-varying component containing the UAV target moving in the surrounding environment, i.e., the dynamic channel component, can be obtained.

[0035] In some embodiments, step S300 may include, but is not limited to, steps S310 to S350: Step S310: Obtain the channel segment of the dynamic channel component; Step S320: Construct a distance and Doppler dictionary based on the resource grid corresponding to the channel segment; Step S330: Obtain dictionary incoherence based on distance and Doppler dictionary; Step S340: Obtain the channel occupancy density of the channel segment; Step S350: Obtain the weighted sum of dictionary incoherence and channel occupancy density to get the channel score.

[0036] In step S310 of some embodiments, a channel segment is obtained from the channel frequency response measurement value after subtracting the static component (i.e., the dynamic channel component). The channel segment is a data block composed of multiple consecutive channel snapshots. A channel snapshot is a channel state at a "point in time" and has instantaneous time characteristics. A channel snapshot is a column in the CFR matrix.

[0037] In step S320 of some embodiments, a range and Doppler grid can be generated based on the dimension parameters of the time-frequency resource grid corresponding to the channel segment and a preset system sensing range. Then, a corresponding theoretical signal atom is generated for each point in the range and Doppler grid, and all theoretical signal atoms constitute a range and Doppler dictionary.

[0038] In some embodiments, the distribution of resource elements (REs) in the time-frequency channel resources determines the achievable range and Doppler resolution. Because the RE distribution is irregular and non-uniform, the classical Nyquist criterion is no longer applicable for evaluating the resolution of two targets. Therefore, embodiments of the present invention redefine this problem as a test of the incoherence of the range-Doppler dictionary: in a compressed sensing framework, the lower the mutual coherence between atoms, the higher the reliability of the algorithm in separating different scattering paths. Incoherence Defined as the inner product between all distinct pairs of atoms The maximum value of is given by the following formula: ; Although mutual coherence can be used as a direct metric for channel reconstruction quality, it has high computational complexity due to the need to calculate the correlation between all pairs of atoms, making it unsuitable for real-time applications.

[0039] In step S330 of some embodiments, based on the properties of the Fourier basis—the mutual coherence of adjacent atoms and the Fourier translation (shift) invariance—an equivalent alternative method is proposed, reducing computational complexity and making it suitable for real-time applications. For example, regarding the mutual coherence of adjacent atoms, consider two atoms along the distance or Doppler dimension, such as... Figure 2 Atom pairs shown and The distance and the step size of the Doppler grid define the finest distance and Doppler shift that the system can resolve. The more precisely these atoms can separate adjacent delays or Doppler shifts, the higher their accuracy in resolving adjacent delays or Doppler shifts, and the smaller their mutual coherence (i.e., the magnitude of the inner product) should be. For Fourier translation invariance, any two atoms with the same phase difference have the same phase shift, and thus the same inner product magnitude. Based on the properties of these two Fourier bases, for a given reference atom, by obtaining the first nearest neighbor atom along the delay (distance) dimension and the second nearest neighbor atom along the Doppler dimension, the inner product magnitude between the reference atom and the first neighbor atom, and the inner product magnitude between the reference atom and the second neighbor atom, are calculated respectively. The larger of these two inner product magnitudes is selected as the dictionary incoherence of the current channel snapshot.

[0040] In some embodiments, step S330 may include, but is not limited to, steps S331 to S336: Step S331: Select a reference atom from the distance and Doppler dictionary; Step S332: Obtain the first neighbor atom adjacent to the reference atom along the distance dimension; Step S333: Obtain the second neighbor atom adjacent to the reference atom along the Doppler dimension; Step S334: Obtain the absolute value of the first inner product between the reference atom and its first neighboring atom; Step S335: Obtain the absolute value of the second inner product between the reference atom and the second neighboring atom; Step S336: Perform a maximum value function operation on the absolute values ​​of the first inner product and the second inner product to obtain the dictionary incoherence.

[0041] In steps S331 to S333 of some embodiments, an atom is selected as a reference atom from a distance and Doppler dictionary. On the distance and Doppler grid, the atom immediately adjacent to the reference atom along the distance dimension is found and designated as the first neighbor atom. On the distance and Doppler grid, the atom immediately adjacent to the reference atom along the Doppler dimension is found and designated as the second neighbor atom. For example, as... Figure 2 As shown, reference atoms are selected from the distance and Doppler dictionary. ,in, These represent the distance index and Doppler index of the reference atom on the distance and Doppler grid, respectively; then, the first neighbor atom adjacent to the reference atom along the distance dimension (i.e., the direction in which the m-index increases) is obtained in the distance and Doppler grid. And obtain the second neighbor atom adjacent to the reference atom along the Doppler dimension (i.e., the direction of increasing n-index). .

[0042] In step S334 of some embodiments, the inner product of the reference atom and the first neighboring atom is obtained, and then the absolute value of the inner product is obtained to obtain the absolute value of the first inner product. For example, the reference atom... With the first neighbor atom The inner product is The absolute value of the first inner product is... .

[0043] In step S335 of some embodiments, the inner product of the reference atom and the second neighboring atom is obtained, and then the absolute value of the inner product is obtained to obtain the absolute value of the second inner product. For example, the reference atom... With the second neighbor atom The inner product is The absolute value of the first inner product is... .

[0044] In step S336 of some embodiments, the dictionary incoherence can be obtained based on the absolute value of the first inner product and the absolute value of the second inner product. The dictionary incoherence estimation formula used includes:

[0045] in, The operator for the maximum value function; The dictionary incoherence is represented by the larger of the absolute values ​​of the first and second inner products as the incoherence of the current channel snapshot. The smaller the value, the higher the separability of adjacent atoms and the lower the mutual coherence.

[0046] In step S340 of some embodiments, the transmit power of the base station (gNB) may vary with the number of scheduled resource blocks. When the power of each resource block is constant, the more resource blocks occupied, the higher the pulse integral energy generated, thus achieving a stronger echo signal-to-noise ratio. Therefore, channel occupancy density is used as a key indicator to measure the potential capability or quality of the current communication signal for sensing. When channel occupancy density and dictionary incoherence are combined to form the channel score, channel occupancy density ensures that the score considers not only the quality of the signal processing algorithm (incoherence) but also the strength of the signal energy at the physical layer, thereby providing a more comprehensive assessment of the quality of sensing opportunities.

[0047] In some embodiments, step S340 may include, but is not limited to, steps S341 to S343: Step S341: Obtain the first number of non-empty resource elements in the channel segment; Step S342: Obtain the second number of resource elements in the resource grid; Step S343: Obtain the ratio of the first quantity to the second quantity to obtain the channel occupancy density.

[0048] In steps S341 to S342 of some embodiments, a first number of non-empty resource elements in the channel segment is obtained. And obtain the total number of resource elements in the resource grid, i.e., the second quantity. .

[0049] In step S343 of some embodiments, the channel occupancy density can be obtained based on a first number of non-empty resource elements in the current channel segment and a second number of resource elements in the resource grid, using formulas including: ; in, This represents channel occupancy density. Channel occupancy density is actually proportional to signal power. In channel ranges with higher transmit power, the robustness of target detection results improves due to the increased signal-to-noise ratio. By introducing signal power as a constraint, the system can avoid channel ranges where the transmit power is insufficient for reliable sensing and instead choose to use a higher transmit power range for sensing.

[0050] In step S350 of some embodiments, in order to filter out segments from the transmitted channel segments that simultaneously have good distance and Doppler resolution and sufficient signal transmission energy, the incoherence-density channel score is calculated using dictionary incoherence. and channel occupancy density The weighted sum, used to reflect the combined effect of the two factors, includes the following formulas: ; in, Indicates the channel score; Indicates the weighting coefficient. It can be adjusted based on processing priorities; for example, if transmit power or signal-to-noise ratio is emphasized, a higher priority can be used. If speed and range resolution are prioritized, then a lower resolution should be used. .

[0051] In some embodiments, step S400 may include, but is not limited to, steps S410 to S490: Step S410: Obtain the third number of dynamic channel components and construct the target path reconstruction-sparse compressed sensing problem; Step S420: Construct a distance and Doppler dictionary based on the resource grid corresponding to the channel segment; Step S430: Mask the channel frequency response measurement values ​​and initialize the residual matrix based on the masked channel frequency response measurement values; Step S440: In the distance and Doppler dictionary, select the first matching atom that has the greatest correlation with the current residual matrix, and add the index of the first matching atom to the support set; Step S450: Based on the current support set, obtain the coefficient estimates of the sparse vectors using the least squares method; Step S460: Update the residual matrix based on the coefficient estimates; Step S470: When the residual change of the updated residual matrix is ​​greater than or equal to the second preset threshold and the maximum number of iterations has not been reached, return to the step of selecting the first matching atom with the greatest correlation to the current residual matrix in the distance and Doppler dictionary. Step S480: When the residual change of the updated residual matrix is ​​less than the second preset threshold, or when the maximum number of iterations is reached, output the reconstructed sparse matrix, and obtain the first target state estimate based on the reconstructed sparse matrix.

[0052] In step S410 of some embodiments, after static background suppression is completed, only the channel frequency response measurement values ​​are retained. (in (For unknown parameters) the main dynamic channel components, and These represent the number of points in the inverse Fourier transform in the frequency dimension and the Fourier transform in the time dimension, respectively. The dynamic channel components can be represented as a set of complex gains. For dimension 1 Distance - Doppler grid , No. The complex gain of a path can be approximated by grid points. Complex gain at the point (or reindex as) ), corresponding to the target distance and Doppler frequency shift parameters. This embodiment of the invention is based on dynamic channel components (whose third quantity is...). Model the problem as -Sparse compressed sensing problem, i.e., target path reconstruction -sparse compressed sensing problem, so as to achieve... Distance-Doppler parameter estimation along the target path.

[0053] For example, the original problem is transformed into a classic compressed sensing signal reconstruction problem, and the formula used is as follows: ; in, Represents the estimated sparse complex gain vector, and represents the reconstructed target reflection coefficient; Let L0 be the pseudonorm, and let vector be the vector. The quantity of non-empty resource elements in the middle and lower reaches of the earth is used to promote sparsity; This represents the measured channel frequency response after masking. Distance and Doppler dictionary; Represents a sparse complex gain vector; This indicates a preset error threshold, used to control reconstruction errors; This represents the L2 norm. The goal of this compressed sensing signal reconstruction problem is to reconstruct signals using a limited number of observations. Reconstructing an approximate distance-Doppler spectrum.

[0054] In some embodiments, a Let represent the complete channel frequency response measurement under full resource unit allocation, and represent the channel response under ideal conditions. Their weighted sum under the Fourier basis can be expressed as: ; in, The Fourier basis matrix representing the frequency dimension; Fourier basis matrix representing the time dimension; Represents the complex gain matrix on the range-Doppler grid; and Each is a matrix The Columns and matrices The Column vector; Represents grid points Complex gain at the location; This represents the conjugate transpose operation; Let be the identity basis matrix in the channel frequency domain, defined as , representing the grid points in the distance-Doppler domain The amplitude at that point is 1, and it is subject to complex gain. Weighted; Indicates the index of the distance dimension, corresponding to latency or distance; An index representing the Doppler dimension, corresponding to velocity or Doppler frequency shift; The number of grid points representing the distance dimension; This represents the number of grid points in the Doppler dimension.

[0055] In some embodiments, step S420 is described with reference to the aforementioned embodiment of step S320, and will not be repeated here.

[0056] In step S430 of some embodiments, a binary sampling mask is introduced. By performing element-wise multiplication, the complete channel frequency response measurement under full resource unit allocation can be extracted. The measured values ​​are used to obtain the masked channel frequency response measurement values. This can be expanded as follows: ; in, This represents the measured channel frequency response after masking. The binary sampling mask is a binary sampling mask matrix with elements consisting only of 0 or 1, and has dimensions of . ; Represents the set of complex numbers; This represents the element-wise multiplication operator; The identity basis matrix representing the channel frequency domain; The distance is represented by the first character in the Doppler dictionary. One atom; Distances represented by atoms and a Doppler dictionary (each atom is...) Matrix), defined as ; This represents a sparse complex gain vector.

[0057] In the expansion formula of the masked channel frequency response measurement, from equation two... To Equation 3 This involves reindexing the double summation. It is converted into a single-index form. The introduced binary sampling mask has non-empty resource elements located only at the positions of occupied resource elements, thus representing... The distribution characteristics of occupied resource units (REs) are determined. After obtaining the masked channel frequency response measurements, the residual matrix is ​​initialized using these measurements. Then there is To solve the problem using the two-dimensional orthogonal matching pursuit (2D-OMP) algorithm. - Sparse compressed sensing provides initial data.

[0058] In step S440 of some embodiments, the atom with the highest correlation to the current residual matrix is ​​selected from the distance and Doppler dictionary as the first matching atom, and the index of the first matching atom is added to the support set.

[0059] In step S450 of some embodiments, the coefficient estimates of the sparse complex gain vector are calculated using the least squares method based on all atoms in the current support set.

[0060] In step S460 of some embodiments, the residual matrix is ​​updated using the coefficient estimates, and the update formula is: ; in, Represents the updated residual matrix; Represents the support set; This represents a sub-dictionary composed of atoms indexed by the support set.

[0061] In some embodiments, steps S470 to S480 involve the following steps: when the residual change of the updated residual matrix is ​​greater than or equal to a preset error threshold... (i.e., the second preset threshold), and the maximum number of iterations has not yet been reached. When the time is reached, return to the step of selecting the first matching atom with the highest correlation to the current residual matrix from the distance and Doppler dictionary. When the residual change of the updated residual matrix is ​​less than the second preset threshold or the maximum number of iterations is reached, the two-dimensional orthogonal matching pursuit algorithm terminates and outputs the reconstructed sparse matrix. Extracting the largest magnitude from the reconstructed sparse matrix. This component will The distance and velocity parameters corresponding to each component are used as the first target state estimate at the current moment.

[0062] In some embodiments, for high-resolution channel bands, although reconstruction capabilities are strong, long-range, low radar cross-section (RCS) UAV targets are relatively weak compared to strong low-speed clutter and transceiver self-leakage. To avoid real targets being obscured by the residual energy of strong interference sources, global estimation does not reconstruct the range-Doppler map from a complete dictionary, but rather atomically isolates the range-Doppler coordinates within the low-speed / short-range region, as this region is typically primarily affected by strong clutter and transceiver self-leakage. Figure 3 As shown in part (b), the interference area is located in , The range. Figure 3 Part (c) demonstrates the advantages of reconstruction after interval isolation. (Compared to...) Figure 3 Compared to part (a) in the previous section, it can be seen that by isolating the low-speed and ultra-close target range regions from the global estimation, the present invention can prevent strong interference from masking the weak echoes of long-range, small RCS UAVs.

[0063] In step S500 of some embodiments, when the channel score is less than a first preset threshold and there was no target UAV at the previous moment, the step of obtaining the channel frequency response measurement value based on the reflected echo is returned.

[0064] In some embodiments, step S600 may include, but is not limited to, steps S610 to S680: Step S610: Construct a distance and Doppler dictionary based on the resource grid corresponding to the channel segment; Step S620: Based on the estimated state of the third target at the previous moment, a pre-set initial bounded region is established; Step S630: Adjust the initial bounded area according to the movement direction of the target UAV to obtain the target bounded area; Step S640: Within the bounded region of the target, extract the corresponding subset of atoms from the distance and Doppler dictionary; Step S650: Obtain the correlation between the dynamic channel component and each candidate atom in the atomic subset, and select the candidate atom with the highest correlation to obtain the second matching atom; Step S660: Based on the grid coordinates of the second matching atom, obtain the first local state estimate at the current moment; Step S670: Perform Kalman filtering on the first local state estimate to obtain the second local state estimate at the current time. Step S680: Perform interval calibration on the second local state estimate to obtain the second target state estimate at the current time. In some embodiments, step S610 is described with reference to the embodiment of step S320 described above, and will not be repeated here.

[0065] In step S620 of some embodiments, at least one third target state estimate obtained at the previous time step is used as the search center at the current time step, wherein each third target state estimate includes a distance estimate and a velocity estimate. Based on each search center, an initial bounded search region is defined on the range-Doppler grid, the initial bounded search region having a range in the distance dimension of [missing information]. ,in, For distance, For Doppler frequency, This is the preset distance offset. This is the preset speed offset.

[0066] In step S630 of some embodiments, the initial bounded area is dynamically adjusted according to the movement direction of the target drone to obtain the target bounded search area. For example, if the target drone is moving away from the base station (…), If the value is negative, it means the distance to the base station is increasing, thus further narrowing the search space to... If the target drone is approaching the base station ( If it is a positive number, then the space direction If the direction is moved, then the first... In the processing snapshot, the first Bounded search area for each target It can be represented as: ; in, Indicates the first Each target is a bounded area; Represents the distance variable; Represents the velocity variable; Indicates the previous moment. Distance estimates for each target; Indicates the previous moment. Estimated velocity values ​​for each target; This indicates the search range and represents the offset in the distance direction. This indicates the velocity search range, which is the offset in the velocity direction; This represents the total number of target state estimates in the last estimation.

[0067] In steps S640 to S660 of some embodiments, within the bounded region of the target, the corresponding subset of atoms is extracted from the range-Doppler dictionary, the correlation between the dynamic channel component and each atom in the subset of atoms is calculated, and the atom with the largest correlation coefficient amplitude is selected as the second matching atom. The range coordinates and velocity coordinates corresponding to the second matching atom on the range-Doppler grid are used as the range estimate and velocity estimate of the target UAV at the current moment, respectively, and the complex gain amplitude corresponding to the second matching atom is extracted to jointly constitute the first local state estimate of the target UAV at the current moment.

[0068] In step S670 of some embodiments, after obtaining a preliminary result from the local estimation, namely the first local state estimate, a state optimization and smoothing process based on dynamic confidence is introduced. This process uses adaptive weighting to obtain a better and more stable state estimate, namely the second local state estimate. Optionally, the first local state estimate is subjected to Kalman filtering. This Kalman filtering employs an adaptively adjusted measurement noise covariance matrix, the value of which is determined by the channel quality score at the current time. Channel quality (quantified by the "channel score") is time-varying, which directly determines the reliability of the local estimate. Therefore, embodiments of the present invention use the measurement noise covariance matrix... Designed for channel score function For example, the first The expression for the measurement noise covariance matrix at time t is: ; In the formula, This represents the adaptively adjusted measurement noise covariance matrix; The normalized channel score, used to measure the quality of the sensing results at the current time step, is defined as follows: , ,in It is the first The raw score of the channel interval at time t. It is a preset high-quality perception threshold; Indicates the upper limit of the measurement noise covariance; This represents the lower limit of the measurement noise covariance.

[0069] In high-quality channels, i.e. At this point, the local estimate has high reliability. Approaching With less measurement noise, the filter will assign higher weights to the measurements, making the output state closer to the current local estimation result and enabling a faster response to real changes in the target. In low-quality channels, i.e. At this point, local estimates may be unreliable. Approaching With significant measurement noise, the filter reduces confidence in the measured values ​​and relies more on the predictions from the system's motion model. This effectively suppresses outliers or estimation biases caused by poor signal quality, maintaining the smoothness and plausibility of the state trajectory.

[0070] In step S680 of some embodiments, the predicted velocity at the current moment, obtained from the Kalman filter prediction step and extrapolated from historical trajectories, is used as the theoretical value. The velocity in the second local state estimate is used as the observed value. The predicted velocity is compared with the observed velocity, and the distance estimate is calibrated based on the comparison result. A second target state estimate is then output, thereby solving the problem that the local search area may completely deviate from the true target due to the accumulation of estimation errors under continuous low-quality channels. For example, if the difference between the velocity predicted by the Kalman filter and the observed velocity in the second local state estimate is less than a preset threshold, it is considered that the prediction and observation are basically consistent, and the locally estimated distance value is reliable and no adjustment is needed. If the velocity difference is greater than the preset threshold, it indicates that tracking may drift. In this case, if the predicted velocity is greater than the measured velocity, the current target distance is increased by 1 to 3 distance units depending on the magnitude of the excess. If the predicted velocity is less than the measured velocity, the current target distance is decreased by 1 to 3 distance units.

[0071] In some embodiments, if the current time The state estimate of a low-altitude UAV based on 5G signals is a second target state estimate obtained after local estimation, Kalman filtering, and interval calibration. Therefore, in the next processing time... When the system re-enters the local estimation branch, based on the time... The second target state estimate is used to update the bounded region, that is: when the system re-enters the local estimation branch, the time step is changed to... The second target state estimate is used as the third target state estimate of the previous moment. The process of returning the third target state estimate of the previous moment and setting the initial bounded region is completed, forming a complete feedback loop.

[0072] In some embodiments, such as Figure 4As shown, taking a 5G base station (gNB) simulating a single-site radar with one transmit link and one auxiliary receive link as an example, the base station, in downlink communication, receives reflected echoes from the environment and performs sensing signal processing. In uplink communication, it is responsible for communication signal processing. The gNB sends OFDM communication data packets to mobile users through the downlink. After the transmitted signal is reflected by surrounding objects such as drones and buildings, the auxiliary receive link captures the reflected echoes, demodulates them, and obtains the channel frequency response measurement value, thereby achieving joint estimation of target distance and velocity.

[0073] For example, a low-altitude drone perception algorithm based on 5G signals, such as Figure 5 As shown, the algorithm's processing steps include: Step 1, Data Collection: When a gNB conducts downlink communication between the base station and surrounding users, it uses an additional receiving antenna to capture the reflected echoes of spontaneously emitted signals. The received signal is then demodulated in the frequency domain and synchronized with a synchronization signal block to obtain the received signal model. After least-squares estimation, the channel frequency response measurement can be obtained.

[0074] Step 2: Static background removal: To detect the range-velocity of targets such as drones, it is necessary to first suppress strong static background reflections from the ground, surrounding buildings, etc. To separate static and dynamic channel components, this system uses a long time-domain window (e.g., 100ms) to perform mean filtering on the CFR to extract the static background. Specifically, for each subcarrier, the static component is calculated as the arithmetic mean of all occupied resource units within the window. Subtracting this static component from the original channel frequency response measurement yields the time-varying component containing moving targets in the surrounding environment.

[0075] Step 3, Incoherence-Density Score: A quantitative index, "Incoherence-Density Score," is proposed to evaluate the quality of signal intervals. This index considers two features that jointly determine target range and velocity estimation: (1) Dictionary incoherence. Based on the properties of two Fourier bases (mutual coherence between adjacent atoms, Fourier translation invariance), for a given atom It is only necessary to calculate the inner product magnitude between the channel and its nearest neighbor atoms along the delay (distance) and Doppler dimensions. The larger of the inner product magnitudes in the two dimensions is taken as the incoherence of the current channel snapshot.

[0076] (2) Channel occupancy density. The channel occupancy density is obtained by the ratio of the number of non-empty resource elements to the total number of resource elements in the current channel snapshot.

[0077] The incoherence-density score is calculated by weighting the dictionary incoherence and channel occupancy density to reflect the combined effect of the two factors. This allows for the selection of segments from the transmitted channel segments that simultaneously possess good delay-Doppler resolution and sufficient signal transmission energy.

[0078] Step 4: Resource density assessment and target existence verification: (1) If the current The channel score within the time window is greater than or equal to the first preset threshold. If so, then global estimation is triggered; (2) If the current The channel score within the time window is less than the first preset threshold. ,and If a target is always present, a global estimation is triggered; (3) Otherwise, continue to collect the window at the next moment.

[0079] Step 5, Global Estimation (when the channel score is greater than or equal to) (Triggered at time) Within the signal processing interval, high-scoring signal segments will be used to generate high-confidence target range-velocity estimates via a two-dimensional orthogonal matching pursuit (2D-OMP) algorithm, serving as anchor points to initialize continuous state estimation. For example, to achieve... Distance-Doppler parameter estimation along each target path, modeling the problem as - Sparse Compressed Sensing Problem. This optimization problem is solved using a two-dimensional orthogonal matching pursuit algorithm. The algorithm initializes the residual matrix using a binary sampling mask followed by a CFR (Computed Regression Function) and then... The following three steps are performed in this iteration: 1) Select the atom with the highest correlation to the current residual. and update the support set. ; 2) Estimate the coefficients using the least squares method. ; 3) Update residuals .

[0080] When the residual change is lower than the second preset threshold The algorithm terminates when the maximum number of iterations is reached, and outputs the reconstructed sparse matrix. .

[0081] Step 6, Local Estimation (triggered when resource density reaches a threshold): For low-resolution signal segments, it cannot provide enough measurement points (due to low density) or sufficient inter-atomic incoherence. This limits the reliability of perception using traditional OMP-based methods. However, it can be observed that low-resolution signal segments do not hinder the feasibility of continuous range / velocity estimation for UAVs: if the search space is limited to the local vicinity of the target, the UAV's range-Doppler characteristics are statistically distinguishable from environmental interference. Range / velocity information from previous high-resolution signal segments can be used to guide range / velocity estimation in the current frame, thereby improving overall stability. The specific steps are as follows: 1) Assume the grid coordinates of the previous high-resolution segment on the distance-Doppler map. A dominant component is generated at the location ( For distance, (Doppler frequency). The continuity of the drone's motion means that the target is very likely in the current segment. It persists in the vicinity of the local area.

[0082] 2) Restrict the dictionary's search space to a bounded region. Inside. The search space can be further narrowed by considering the target's direction of movement. If the target is moving away from the base station ( If the value is negative, it means the distance to the base station is increasing. Therefore, the search space is further narrowed down to... If the target is approaching the base station ( If it is a positive number, then the space direction If the direction is moved, then the first... In the processing snapshot, the first Target Area It can be represented as: .

[0083] Optionally, select It defines a bounding box [3.49m, 1.03m / s]. Each bounded block region will be independently evaluated for the atom with the maximum coefficient. The grid coordinates associated with the maximum coefficient magnitude are selected as the estimated target state (range, velocity, and amplitude) for the current segment. , , In subsequent segments, the above operation will be repeated using the new target coordinates as the center. In this way, the target state within the lower segments can be continuously updated.

[0084] Step 7, Kalman filtering: While local estimation enhances the robustness of target range and velocity estimation by limiting the solution range of the target location, repeated estimation errors can cause the solution range of the local estimation to shift, even gradually moving the solution interval away from the true target location, resulting in erroneous estimations. To mitigate such error propagation, this invention employs a method with a dynamic measurement noise matrix. The Kalman filter, based on channel fluctuations, weighs the degree of trust between the predicted and estimated values: if the current channel incoherence-density score is high, the estimated value of the system is trusted more; otherwise, the predicted value is trusted more. The Kalman filter uses the... Measurement noise covariance matrix at time 1 At the two upper and lower boundaries and It adapts and adjusts itself accordingly.

[0085] Step 8, Interval Calibration: When the channel occupancy score is low across multiple consecutive processing windows, the system may accumulate range and velocity estimation errors due to low signal-to-noise ratio, resulting in an incorrect local estimation range—meaning the target's true range and velocity may not be present within the local estimation range. To avoid this, interval calibration compares the velocity value predicted by the Kalman filter for each target trajectory with the current locally estimated velocity value. If the difference is less than a preset threshold, the center of the next local estimation is the current target range and velocity; if the predicted velocity is greater than the locally estimated velocity, the current target range is increased by 1 to 3 range units depending on the magnitude of the excess; conversely, it is decreased by 1 to 3 range units. If the next processing window is also a local estimation, the calibrated target position is used as the center for the local estimation.

[0086] This invention also provides a low-altitude UAV sensing device based on 5G signals, which can implement the above-mentioned low-altitude UAV sensing method based on 5G signals. The device includes: The data acquisition module is used to acquire the reflected echo of spontaneous signals and obtain the channel frequency response measurement value based on the reflected echo. The preprocessing module is used to perform static background elimination on the channel frequency response measurements to obtain dynamic channel components; The channel score acquisition module is used to obtain the dictionary incoherence and channel occupancy density of channel segments based on the channel segments of dynamic channel components, and then obtain the channel score. The global estimation module is used to compare the channel score with a first preset threshold. When the channel score is greater than or equal to the first preset threshold, the dynamic channel components are globally estimated using a two-dimensional orthogonal matching pursuit algorithm to obtain the first target state estimate at the current moment. The first local estimation module is used to return the step of obtaining the channel frequency response measurement value based on the reflected echo when the channel score is less than the first preset threshold and there was no target UAV in the previous moment. The second local estimation module is used to perform local estimation of the dynamic channel components when the channel score is less than the first preset threshold and the target UAV existed in the previous time step, so as to obtain the second target state estimate value at the current time step.

[0087] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0088] This invention also provides an electronic device, which includes a processor and a memory. The memory stores a computer program, and the processor executes the computer program to implement the above-described method. This electronic device can be any smart terminal, including a tablet computer, an in-vehicle computer, or similar device.

[0089] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0090] refer to Figure 6 , Figure 6 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 701 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention. The memory 702 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 702 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 702 and is called and executed by the processor 701. The input / output interface 703 is used to implement information input and output; The communication interface 704 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 705 transmits information between various components of the device (e.g., processor 701, memory 702, input / output interface 703, and communication interface 704); The processor 701, memory 702, input / output interface 703, and communication interface 704 are connected to each other within the device via bus 705.

[0091] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0092] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0093] This invention also provides a computer program product or computer program that includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions to cause the computer device to perform the aforementioned method.

[0094] In summary, the low-altitude UAV sensing method and related equipment based on 5G signals according to embodiments of the present invention have the following advantages: 1. The incoherence proposed in the embodiments of the present invention - Density score measures the signal quality within the current data processing window, and the Kalman filter adaptively adjusts the weights between the measurement matrix and the noise matrix based on the score, effectively mitigating the impact of communication data fluctuations.

[0095] 2. This invention presents an adaptive hybrid estimation algorithm based on two-dimensional orthogonal matching pursuit (2D-OMP), enabling continuous target perception even under conditions of non-uniform and fluctuating communication resource occupancy. The system uses the non-uniform resource occupancy as a mask matrix, allowing for the selection of globally estimated targets and locally tracked targets based on different resource occupancy densities. Within time windows where the score is greater than a threshold, the estimated distance and velocity values ​​serve as anchor points for initial and stable tracking trajectories. For time windows where the score is less than the threshold, this invention restricts the 2D-OMP search space to a local region surrounding the previous estimates, adjusting the range of this local region based on the target's instantaneous velocity. This design effectively amplifies the signal-to-noise ratio of target detection. Even in cases of sparse or ill-conditioned measurement matrices, this reduced search space enables 2D-OMP to provide accurate tracking results.

[0096] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and sub-operations described as part of a larger operation are executed independently.

[0097] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the described functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0098] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0100] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0101] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0102] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0103] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0104] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A low-altitude unmanned aerial vehicle (UAV) sensing method based on 5G signals, characterized in that, Includes the following steps: Acquire the reflected echo of the spontaneous signal, and obtain the channel frequency response measurement value based on the reflected echo; Static background elimination is performed on the channel frequency response measurements to obtain the dynamic channel components; Based on the channel segments of the dynamic channel components, the dictionary incoherence and channel occupancy density of the channel segments are obtained to obtain the channel score; The channel score is compared with a first preset threshold. When the channel score is greater than or equal to the first preset threshold, the dynamic channel components are globally estimated using a two-dimensional orthogonal matching pursuit algorithm to obtain the first target state estimate at the current moment. When the channel score is less than the first preset threshold and there was no target UAV in the previous moment, return to the step of obtaining the channel frequency response measurement value based on the reflected echo; When the channel score is less than the first preset threshold and the target UAV existed in the previous time, the dynamic channel components are locally estimated to obtain the second target state estimate at the current time.

2. The method according to claim 1, characterized in that, The process of acquiring the reflected echo of the spontaneous signal and obtaining the channel frequency response measurement value based on the reflected echo includes the following steps: The reflected echo of the spontaneous signal is obtained by receiving the antenna; The reflected echo is subjected to frequency domain demodulation and synchronization processing to construct a received signal model; Based on the spontaneous signal and the received signal model, the channel frequency response measurement value is obtained.

3. The method according to claim 1, characterized in that, The process of statically eliminating the background of the channel frequency response measurement to obtain the dynamic channel components includes the following steps: The channel frequency response measurement values ​​are subjected to mean filtering through a long time-domain window to obtain the static component; The dynamic channel component is obtained by subtracting the static component from the channel frequency response measurement.

4. The method according to claim 1, characterized in that, The step of obtaining the dictionary incoherence and channel occupancy density of a channel segment based on the channel segment of the dynamic channel component, and thus obtaining the channel score, includes the following steps: Obtain the channel segment of the dynamic channel component; Based on the resource grid corresponding to the channel segment, construct a distance and Doppler dictionary; The dictionary incoherence is obtained based on the distance and the Doppler dictionary; Obtain the channel occupancy density of the channel segment; The channel score is obtained by weighting the dictionary incoherence and the channel occupancy density.

5. The method according to claim 4, characterized in that, The step of obtaining the dictionary incoherence based on the distance and the Doppler dictionary includes the following steps: Select a reference atom from the distance and Doppler dictionary; Obtain the first neighboring atom that is adjacent to the reference atom along the distance dimension; Obtain the second neighbor atom adjacent to the reference atom along the Doppler dimension; Obtain the absolute value of the first inner product between the reference atom and the first neighboring atom; Obtain the absolute value of the second inner product between the reference atom and the second neighboring atom; The dictionary incoherence is obtained by performing a maximum value function operation on the absolute values ​​of the first inner product and the second inner product.

6. The method according to claim 4, characterized in that, Obtaining the channel occupancy density of the channel segment includes the following steps: Obtain the first number of non-empty resource elements in the channel segment; Obtain the second number of resource elements in the resource grid; The channel occupancy density is obtained by obtaining the ratio of the first quantity to the second quantity.

7. The method according to claim 1, characterized in that, The step of comparing the channel score with a first preset threshold, and when the channel score is greater than or equal to the first preset threshold, performing a global estimation of the dynamic channel components using a two-dimensional orthogonal matching pursuit algorithm to obtain the first target state estimate at the current moment, includes the following steps: Obtain the third quantity of the dynamic channel components and construct the target path reconstruction-sparse compressed sensing problem; Based on the resource grid corresponding to the channel segment, construct a distance and Doppler dictionary; The channel frequency response measurements are masked, and the residual matrix is ​​initialized based on the masked channel frequency response measurements. In the distance and Doppler dictionary, select the first matching atom that has the greatest correlation with the current residual matrix, and add the index of the first matching atom to the support set; Based on the support set described above, the coefficient estimates of the sparse vectors are obtained by the least squares method; The residual matrix is ​​updated based on the coefficient estimates; When the residual change of the updated residual matrix is ​​greater than or equal to the second preset threshold and the maximum number of iterations has not been reached, return to the step of selecting the first matching atom with the greatest correlation to the current residual matrix in the distance and Doppler dictionary; When the residual change of the updated residual matrix is ​​less than the second preset threshold, or when the maximum number of iterations is reached, the reconstructed sparse matrix is ​​output, and the first target state estimate is obtained based on the reconstructed sparse matrix.

8. The method according to claim 1, characterized in that, When the channel score is less than the first preset threshold and the target UAV existed in the previous time step, the dynamic channel components are locally estimated to obtain the second target state estimate for the current time step, which includes the following steps: Based on the resource grid corresponding to the channel segment, construct a distance and Doppler dictionary; Based on the estimated state of the third target at the previous moment, a pre-defined initial bounded region is set. Based on the movement direction of the target UAV, the initial bounded area is adjusted to obtain the target bounded area; Within the bounded region of the target, extract the corresponding subset of atoms from the distance and the Doppler dictionary; The correlation between the dynamic channel component and each candidate atom in the atomic subset is obtained, and the candidate atom with the highest correlation is selected to obtain the second matching atom; Based on the grid coordinates of the second matched atom, the first local state estimate at the current moment is obtained; The first local state estimate is processed by Kalman filtering to obtain the second local state estimate at the current time. The second local state estimate is calibrated by interval to obtain the second target state estimate at the current time.

9. A low-altitude unmanned aerial vehicle (UAV) sensing device based on 5G signals, characterized in that, include: The data acquisition module is used to acquire the reflected echo of the spontaneous signal and to acquire the channel frequency response measurement value based on the reflected echo. The preprocessing module is used to perform static background elimination on the channel frequency response measurement values ​​to obtain dynamic channel components; The channel score acquisition module is used to obtain the dictionary incoherence and channel occupancy density of the channel segments based on the channel segments of the dynamic channel component, and to obtain the channel score. The global estimation module is used to compare the channel score with a first preset threshold. When the channel score is greater than or equal to the first preset threshold, the dynamic channel component is globally estimated using a two-dimensional orthogonal matching pursuit algorithm to obtain the first target state estimate at the current time. The first local estimation module is used to return to the step of obtaining the channel frequency response measurement value based on the reflected echo when the channel score is less than the first preset threshold and there was no target UAV in the previous moment. The second local estimation module is used to perform local estimation of the dynamic channel components when the channel score is less than the first preset threshold and the target UAV existed in the previous time, so as to obtain the second target state estimate value at the current time.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 8.

Images