A method and system for positioning low-altitude unmanned aerial vehicles in urban environments based on wireless signals

Abstract

The present disclosure is suitable for the field of unmanned aerial vehicle positioning, and provides a low-altitude unmanned aerial vehicle positioning method and system in urban environment based on wireless signals. By using sensing nodes distributed in different positions in the city, the wireless signals broadcasted by the unmanned aerial vehicle for image transmission are passively received, and the positioning of the non-cooperative target can be realized without the active cooperation of the unmanned aerial vehicle. The limitation of the traditional monitoring method based on long-range identification broadcast, which requires the unmanned aerial vehicle to actively report information, is overcome. At the same time, by extracting the time difference of arrival and the angle of arrival to fuse the positioning, compared with the existing method which only relies on a single parameter, the positioning result can be complementary in different signal environments, and the reliability and robustness of the positioning result can be improved. At the same time, the interpolation processing technology is introduced at the signal simulation level, the discrete sampling signal is interpolated to restore the continuous signal amplitude, and the problem of insufficient distance difference resolution granularity caused by the sampling period limitation of the actual sampling device is overcome.

Description

Technical Field

[0001] This application belongs to the field of unmanned aerial vehicle (UAV) positioning, and specifically relates to a method for positioning low-altitude UAVs in urban environments based on wireless signals. Background Technology

[0002] With the widespread application of drones in logistics delivery, urban inspection, security monitoring, and emergency rescue, drone activity in low-altitude airspace is becoming increasingly frequent. However, the improper use of drones has also brought serious safety hazards and privacy issues, such as entering other people's residences without permission, filming sensitive areas, and even intruding into no-fly zones such as airports and nuclear power plants, posing a threat to public safety.

[0003] To address this issue, various drone monitoring methods have been developed, primarily including optical monitoring, acoustic monitoring, radar monitoring, and broadcast monitoring based on Remote Identification (RID). Optical monitoring methods identify drones by capturing images with cameras, but are easily obstructed by opaque buildings in urban environments. Furthermore, due to their small size and varied appearance, drones are difficult to distinguish from birds and other flying objects, and monocular vision alone cannot provide accurate three-dimensional spatial location information. Acoustic monitoring methods identify drones by collecting the noise emitted by their rotors, but are easily interfered with by background noise in noisy urban environments, and their effective monitoring range is very limited. Radar monitoring methods detect targets by emitting electromagnetic waves and receiving reflected echoes, but due to their high transmission power and strong electromagnetic radiation, they are unsuitable for deployment in densely populated urban environments. Additionally, drones typically have a small radar cross-section, making it difficult to reliably distinguish them from birds and other objects. The RID-based broadcast monitoring method requires drones to broadcast their identity information, location coordinates, and other data periodically during flight for the ground monitoring system to receive. This method relies on the active cooperation of drones. However, it will be completely ineffective for older drones or drones from small brands that do not have RID modules installed, as well as for non-cooperative drones that maliciously tamper with broadcast information.

[0004] In summary, existing drone monitoring methods have significant limitations when dealing with non-cooperative drones in complex urban environments. Optical and radar methods are easily obstructed and have difficulty distinguishing targets, acoustic methods are severely affected by environmental interference and have short detection ranges, while RID broadcasting methods rely on drones to actively cooperate and cannot cope with non-cooperative scenarios.

[0005] Therefore, how to achieve passive, high-precision positioning of low-altitude drones without relying on the active cooperation of drones, and how to effectively evaluate the location layout and algorithm performance of sensing nodes before actual deployment, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] To address the aforementioned issues, this application provides a method for locating low-altitude unmanned aerial vehicles (UAVs) in urban environments based on wireless signals. This method overcomes the limitations of traditional monitoring methods based on remote identification and broadcasting, which require UAVs to actively report information.

[0007] A method for locating low-altitude unmanned aerial vehicles (UAVs) in urban environments based on wireless signals includes the following steps: Multiple distributed sensing nodes are set up and deployed in different locations in the urban space to receive wireless signals broadcast by drones. Each distributed sensing node contains an array antenna and has time synchronization capability. Obtain pre-configured simulation scene parameters, including the fixed spatial coordinates of each distributed sensing node and the true spatial coordinates of the simulated drone's location; Based on the simulation scenario parameters, a modulated wireless signal broadcast by the simulated drone for transmitting images is set, and the propagation process of the wireless signal in space is simulated to obtain the received signals of each sensing node. Based on the received signals from each sensing node, signal processing is used to obtain the time difference of the simulated drone arriving at different sensing nodes and the angle of arrival of the wireless signal broadcast by the simulated drone as measured by a single sensing node. The estimated position of the simulated drone is calculated based on the time difference of arrival and / or angle of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node.

[0008] Furthermore, by simulating the propagation process of wireless signals in space, the received signals of each sensing node are obtained, including: The continuous signal received by each sensing node is calculated based on the spatial transmission distance from the simulated drone to each sensing node. Set the sampling period for each sensing node, sample the continuous signal according to the sampling period, and obtain the discrete signal sequence of each sensing node. Interpolation is performed on the discrete signal sequence to obtain a signal sequence with improved time resolution, which is then output as the received signal.

[0009] Furthermore, the interpolation process is a multiple interpolation process, which includes: Based on a preset interpolation factor, new sampling points are inserted between adjacent sampling points to obtain a signal sequence with improved time resolution and interpolation factor, which is then used as the received signal output.

[0010] Furthermore, the time differences between the simulated drone's arrival at different sensing nodes, obtained through signal processing, include: Select a sensing node as a reference node, and perform cross-correlation operations between the received signals of the other sensing nodes and the received signals of the reference node. The cross-correlation operation is performed by sliding the received signals of the reference node and the received signals of any sensing node relative to each other in the time dimension and calculating the sum of the products of the overlapping parts, so as to obtain the sequence of cross-correlation function changes with the relative sliding time. The relative sliding time position corresponding to the peak of the sequence of cross-correlation function changes with relative sliding time is selected, and the relative sliding time corresponding to the relative sliding time position is used as the time difference between the simulated UAV and the sensing node.

[0011] Furthermore, the angle of arrival of the simulated UAV broadcast wireless signal measured by a single sensing node, obtained through signal processing, includes: At each sensing node, the phase difference of the signals received by the array antenna is used to divide the received data space into mutually orthogonal signal subspace and noise subspace through feature decomposition; By utilizing the orthogonality of the two subspaces, a spatial spectral function is constructed, and a spectral peak search is performed in the entire angle space. The angles corresponding to the spectral peaks are the azimuth and elevation angles of the incoming wave direction of the received signal.

[0012] Furthermore, based on the time difference of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node, the estimated position of the simulated drone is calculated, including: Based on the time difference between the simulated UAV arriving at the sensing node and the reference node, candidate positions are obtained by solving a system of nonlinear equations. Select a sensing node as the decision node, calculate the theoretical distance difference between the candidate position and the reference node and the decision node respectively, compare the theoretical distance difference with the distance difference obtained by actual measurement based on the time difference of arrival, and select the candidate position that is closest to the actual value as the initial estimated position. Using the initial estimated position as the initial value for iteration, the nonlinear time difference of arrival observation equation is expanded in first order Taylor at the initial value. After linearization, the position correction is solved by weighted least squares method, and the estimated position is updated. Repeat the above process until the iteration step size is less than the preset threshold or the preset number of iterations is reached, then stop the iteration and use the iteration result at this time as the final estimated position.

[0013] Furthermore, the iterative process of the Taylor series expansion method includes: The observation equation for the time difference of arrival is expanded in first order Taylor at the initial estimated position to establish the error equation. The position correction is solved by weighted least squares method, and the estimated position is updated. The above process is repeated until convergence.

[0014] Furthermore, based on the angle of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node, the estimated position of the simulated drone is calculated, including: A spatial ray is constructed based on the azimuth and elevation angles of the incoming wave direction of the signal received by the sensing node; The least squares method is used to solve for the intersection of spatial rays. By minimizing the sum of squared vertical distances from all rays to the point to be estimated, a system of linear equations is established and solved to obtain the estimated position of the UAV.

[0015] A low-altitude unmanned aerial vehicle (UAV) positioning system for urban environments based on wireless signals, comprising: Distributed sensing nodes are deployed in different locations in the urban space to receive wireless signals broadcast by drones. Each distributed sensing node contains an array antenna and has time synchronization capabilities. The parameter configuration module is used to obtain pre-configured simulation scene parameters, which include the fixed spatial coordinates of each distributed sensing node and the true spatial coordinates of the simulated UAV's location. The signal simulation module is used to set the modulated wireless signal broadcast by the simulated drone for image transmission according to the simulation scene parameters, and to simulate the propagation process of the wireless signal in space to obtain the received signals of each sensing node. The parameter extraction module is used to obtain the time difference of the simulated drone arriving at different sensing nodes and the angle of arrival of the wireless signal broadcast by the simulated drone measured by a single sensing node, based on the received signals of each sensing node. The position calculation module is used to calculate the estimated position of the simulated drone based on the time difference of arrival and / or angle of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node.

[0016] A computer program product includes computer instructions that, when executed by a processor, specifically perform the steps of any of the methods described above.

[0017] Compared with the prior art, this application has the following advantages: This application employs distributed sensing nodes deployed in different urban locations to passively receive wireless signals broadcast by drones for image transmission. This enables the location of non-cooperative targets without relying on the active cooperation of drones, overcoming the limitations of traditional monitoring methods based on remote identification broadcasts that require drones to actively report information. Furthermore, by extracting two types of parameters—time difference of arrival and angle of arrival—for fusion positioning, compared to existing methods that rely on only a single parameter, this approach can complement each other under different signal environments, improving the reliability and robustness of the positioning results.

[0018] Meanwhile, this application introduces interpolation processing technology at the signal simulation level. By interpolating discrete sampled signals to restore the amplitude of continuous signals, it overcomes the problem of insufficient distance difference resolution granularity caused by the sampling period limitation of actual sampling equipment, making the simulation results closer to the real deployment scenario.

[0019] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart of a method according to an embodiment of this application is shown; Figure 2 A system block diagram according to an embodiment of this application is shown; Figure 3 A simulation interface diagram according to an embodiment of this application is shown; Figure 4 The diagram shows the IQ signal waveforms received by four sensing nodes at different locations according to an embodiment of this application. Figure 5 The cross-correlation function diagrams of four sensing nodes and reference nodes at different locations according to embodiments of this application are shown. Figure 6 The image shows a spatial spectrum energy map of the angle of arrival estimated using a two-dimensional MUSIC algorithm according to an embodiment of this application; Figure 7 The following diagram illustrates the positioning error performance under different signal-to-noise ratio environments with different sampling interpolation numbers according to embodiments of this application; Figure 8 The following illustrations show the positioning error performance under different signal-to-noise ratio environments, comparing the addition of Taylor optimization and the use of only the Chan algorithm according to embodiments of this application. Figure 9 The positioning error performance of the angle of arrival estimation algorithm according to the embodiments of this application is shown under different signal-to-noise ratio environments. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] This application is primarily applied to passive positioning scenarios for low-altitude drones in urban environments. With the increasing prevalence of drones in urban logistics, inspection, and security, unauthorized "black flights" pose a serious threat to public safety. Existing monitoring methods, such as optical, acoustic, radar, and remote identification broadcasting, have significant limitations in dealing with non-cooperative drones in complex urban environments: optical methods are easily obstructed and difficult to measure distances; acoustic methods are easily interfered with by environmental noise; radar has strong radiation and is difficult to distinguish small targets; and remote identification broadcasting relies on the active cooperation of drones.

[0024] Therefore, how to achieve high-precision positioning using the autonomously transmitted wireless image transmission signals of drones without relying on their cooperation, and how to effectively simulate and evaluate the layout of sensing nodes and algorithm performance before actual deployment, has become a technical problem that urgently needs to be solved in this field.

[0025] To address the aforementioned issues, this application employs distributed sensing nodes deployed at different altitudes within the city to passively receive wireless signals broadcast by drones for image transmission. Simultaneously, interpolation technology is introduced at the simulation level to overcome the resolution limitations of sampling devices, and two types of parameters, time difference of arrival and angle of arrival, are fused for positioning. Furthermore, a fifth node is introduced to eliminate fuzzy solutions, thereby significantly improving positioning accuracy and reliability.

[0026] The simulation positioning method of this application will be described in detail step by step below with reference to the accompanying drawings. Please refer to the attached drawings. Figure 1 The steps in this embodiment include: S1. Simulation scenario parameter configuration and signal propagation simulation.

[0027] S1-1, Set up multiple distributed sensing nodes.

[0028] Distributed signal monitoring technology, represented by distributed sensing nodes, is already a mature and widely used existing technology in this field. For example, in multimedia sensor networks, a distributed sensing system composed of sensor nodes with computing, storage and communication capabilities in a self-organizing manner typically includes a three-layer structure of sensor nodes, aggregation nodes and control center. The nodes are responsible for collecting data such as audio, video and images and transmitting them to the aggregation center.

[0029] These sensing nodes are deployed at fixed locations at different heights in the urban space to receive wireless signals broadcast by drones for image transmission. To ensure the extraction of subsequent positioning parameters, each distributed sensing node includes an array antenna and has high-precision time synchronization capability. The array antenna is used to estimate the angle of arrival of the signal, and the high-precision time synchronization capability is used to measure the time difference of arrival between different nodes.

[0030] S1-2, Simulation of UAV signal propagation.

[0031] This step aims to simulate the true space coordinates of the UAV's location based on pre-configured simulation scenario parameters, serving as a benchmark for subsequent algorithm verification and evaluating the accuracy of the positioning results.

[0032] S1-2-1. Pre-configure simulation scene parameters.

[0033] The pre-configured simulation scenario parameters include the fixed spatial coordinates of each distributed sensing node and the ground-value spatial coordinates of the simulated UAV's location, such as... Figure 3 As shown, it is a simulation interface diagram of three-dimensional time difference of arrival localization (3DTDOA Localization) in an embodiment of this application, which intuitively shows the position distribution of five sensing nodes (5 sensing nodes s) and the UAV in three-dimensional space.

[0034] The spatial coordinates of the five distributed sensing nodes are set as follows: Sensing Node 1 (BS0): :0m, :0m, : 20m.

[0035] Sensing Node 2 (BS1): 300m :0m, : 80m.

[0036] Sensing Node 3 (BS2): 300m 300m : 40m.

[0037] Sensing Node 4 (BS3): :0m, 300m : 20m.

[0038] Sensing Node 5 (BS4): 150m 150m :0m.

[0039] The true space coordinates of the simulated UAV location are: 89m 135m : 40m.

[0040] Where x and y are horizontal coordinates, and z is the height coordinate.

[0041] Subsequently, the simulated drone was set to operate at a speed of 1×10 per second. 6 The symbol rate transmits a broadcast signal using Quadrature Phase Shift Keying (QPSK) modulation. The bit sequence is generated completely randomly, the signal transmission time is set to time 0, and the speed of light is 3 × 10⁻⁶. 8 m / s.

[0042] S1-2-2. Based on the pre-configured simulation scene parameters, set the modulated wireless signal broadcast by the simulated drone for image transmission. In this embodiment, the drone is set to transmit the wireless signal at a rate of 1×10⁻⁶ Hz. 6 The symbol rate of the symbol is used to transmit a quadrature phase shift keying (QPSK) modulated broadcast signal. The bit sequence is generated completely randomly, and the signal transmission time is set to time 0.

[0043] Since the propagation of analog signals in space requires consideration of time delays caused by distance propagation and free-space transmission losses, the following formula is used to reflect the physical laws governing signal amplitude attenuation with distance and propagation delay. Assuming the original signal is s(t), the signal s received by the i-th sensing node is calculated. i (t), the formula is as follows:

[0044] in, For reference range, it can be set to 1; t is a continuous time variable, representing the change of the signal on the time axis; c is the speed of light, taken as 3 × 10⁻⁶. 8 m / s; α is the path fading constant, which is taken as 2 in this embodiment; Let be the spatial transmission distance from the UAV to the i-th sensing node; N i (t) represents the Gaussian noise received by the i-th sensing node at time t.

[0045] Since actual receiving devices can only sample in a discrete manner, it is necessary to discretize the continuous signal.

[0046] The sampling period of each sensing node is set to T. s(This embodiment corresponds to a sampling bandwidth of 100MHz, T) s =10ns), after a distance d i After transmission, the discrete signal sequence S acquired by the i-th sensing node i [n] is represented as:

[0047] in, For the nth sampling time (n is an integer, T) s (Sampling period).

[0048] d i / c represents the propagation delay.

[0049] s interp This is the original signal after interpolation.

[0050] N i [n] represents the discrete Gaussian noise superimposed on the i-th sensing node at the n-th sampling time.

[0051] See also Figure 4 This is a waveform diagram of the IQ signal received by four sensing nodes (sensing node 1 to sensing node 4) at different locations. The horizontal axis represents the sampling time, and the vertical axis represents the signal amplitude. It shows the signal amplitude changes over the first 1000 sampling points, reflecting the time offset and amplitude difference of the signal propagation to nodes at different distances. Based on the coordinates of each node, the spatial transmission distance d from the UAV to each sensing node can be calculated using the Euclidean distance formula. i and based on d i Calculate the propagation delay for each sensing node.

[0052] according to Figure 4 The data shows that the waveform characteristics of each node are closely related to its measured distance: Node 1 (coordinates 0, 0, 20) is approximately 163.6 meters away from the drone, making it the closest of all nodes. Therefore, it receives the largest signal amplitude, and due to its propagation delay of approximately 0.545 μs, the shortest among the four nodes, its waveform appears earliest on the time axis.

[0053] Node 2 (coordinates 300, 0, 80) and Node 4 (coordinates 0, 300, 20) are approximately 253.1 meters and 187.7 meters away from the UAV, respectively. The waveform amplitude and occurrence time of both nodes fall between those of Node 1 and Node 3. Among them, Node 4, being closer, should have a significantly stronger signal amplitude than Node 2.

[0054] Node 3 (coordinates 300, 300, 40) is approximately 267.8 meters away from the UAV, making it the farthest of all nodes. Due to path loss, it receives the smallest signal amplitude and has the longest propagation delay, approximately 0.893 μs, with the waveform exhibiting the most significant time lag.

[0055] S1-2-3, Perform interpolation processing on discrete signal sequences.

[0056] Since directly using discrete sampling sequences will limit the resolution granularity of distance differences and make it impossible to accurately simulate continuous propagation delay, in order to solve this problem, it is necessary to perform interpolation processing on discrete signal sequences to restore the amplitude of continuous signals.

[0057] Specifically, the interpolation process is a multi-fold interpolation process. Based on a preset interpolation factor k, new sampling points are inserted between adjacent sampling points to obtain a signal sequence with a time resolution correspondingly increased by the interpolation factor. The interpolation process can employ one of the following methods: linear interpolation, cubic spline interpolation, or Singer interpolation. Cubic spline interpolation achieves a good balance between computational complexity and accuracy. Its formula is as follows:

[0058] In some embodiments, the discrete signal sequence can be subjected to 32x cubic spline interpolation, inserting 31 new points between every two original sampling points to obtain a signal sequence with improved time resolution of approximately 0.3125 ns, which is then used as the received signal output.

[0059] See also Figure 7 This is a comparison chart of positioning error performance under different sampling interpolation multiples and different signal-to-noise ratio environments. The horizontal axis of the chart represents the signal-to-noise ratio (SNR), ranging from -20dB to 20dB; the vertical axis represents the positioning error log, in meters, ranging from 0 to 4 meters. The chart contains four curves: the red line corresponds to 1x interpolation (no interpolation), the green line corresponds to 4x interpolation, the orange line corresponds to 8x interpolation, and the blue line corresponds to 32x interpolation.

[0060] Looking at the overall trend of the curves, the positioning error of all curves decreases as the signal-to-noise ratio (SNR) increases. The error decreases more rapidly in the low SNR region (-20dB to 0dB), while it tends to stabilize in the high SNR region (10dB to 20dB). In a horizontal comparison, at the same SNR, the higher the interpolation factor, the smaller the positioning error. The 32x interpolation curve is always at the bottom, the 1x interpolation curve is at the top, and the 4x and 16x interpolation curves are in the middle. For example, at a 10dB SNR, the 1x interpolation error is approximately 0.3 meters, the 4x interpolation error is approximately -0.1 meters, the 8x interpolation error is approximately -0.6 meters, and the 32x interpolation error is approximately -0.7 meters. The spacing between the curves gradually decreases as the interpolation factor increases, indicating that the accuracy improvement brought by interpolation has a marginal effect. This figure demonstrates the effectiveness of interpolation processing in improving the accuracy of time difference of arrival (TDOA) estimation and thus reducing positioning error, reflecting the positive correlation between the interpolation factor and positioning accuracy.

[0061] S2. Calculation of arrival time difference and arrival angle.

[0062] (a) Calculation of arrival time difference.

[0063] S2-1. Utilizing the high-precision time synchronization capability of the distributed sensing nodes, a sensing node i is selected as the reference node j (sensing node 1 is selected in this embodiment). The received signals of the remaining sensing nodes are cross-correlated with the received signals of the reference node, wherein the cross-correlation function is defined as:

[0064] Among them, S i (t) and S j (t) represents the continuous signal forms of the i-th sensing node and the reference node j after interpolation, respectively; S j （k） [n+m] represents the (n+m)th sampling point of the signal sequence after k-fold interpolation of the reference node j, i.e., the reference node sequence relative to S. i The sequence was offset by m sampling points.

[0065] S2-2. Cross-correlation operation involves sliding two signals relative to each other over time and summing the products of the overlapping portions to obtain a sequence of cross-correlation functions as a function of relative sliding time. The search then finds the relative displacement corresponding to the maximum value (peak position) of the cross-correlation function. ,Right now:

[0066] Among them, R ij [m] represents the value of the cross-correlation function between the i-th sensing node and the j-th sensing node at the m-th displacement point; S2-3, relative displacement Converted to the distance difference between the signal source and the two sensing nodes, i and j. The formula is as follows:

[0067] Where k represents the interpolation factor.

[0068] Figure 5 This is a cross-correlation function graph of the four sensing nodes (nodes 1 to 4) and the reference node in this embodiment. The time difference corresponding to the peak position in the graph is the arrival time difference between each node and the reference node. It shows the waveform result after cross-correlation operation between the received signals of the four sensing nodes and the reference signal. The horizontal axis is the number of lag sampling points (Lag), and the vertical axis is the cross-correlation amplitude. Specifically: The peak value of sensing node 1 is located at approximately 80 sampling points, with an amplitude of approximately 0.025. The peak value of sensing node 2 shifts to the right to approximately 85-90 sampling points, and the amplitude decreases to approximately 0.022; The peak value of sensing node 3 is located at approximately 60-65 sampling points, with an amplitude of approximately 0.034. The peak value of sensing node 4 is located at approximately 40-45 sampling points, with an amplitude of approximately 0.085.

[0069] The data in the figure shows that the differences in physical distance or propagation path of the four sensing nodes lead to a significant time delay distribution. The signal matching quality of sensing node 4 and sensing node 3 is significantly better than that of the first two base stations. Sensing node 4 has the shortest overall spatial distance and the least signal attenuation. Although sensing node 3 is farther in the horizontal distance, its height (40m) is exactly the same as the height of the UAV, eliminating vertical path loss. This conforms to the law that signal strength is inversely proportional to the square of the distance, and shows a higher signal-to-noise ratio and correlation amplitude.

[0070] (ii) Calculation of arrival angle.

[0071] S2`-1. Utilizing the array antennas contained in the distributed sensing nodes, at each sensing node, the direction of arrival of the wave is estimated by using the two-dimensional MUSIC (Multiple Signal Classification) algorithm based on the phase difference of the received signal through the array antenna. This algorithm divides the received data space into mutually orthogonal signal subspace and noise subspace through feature decomposition.

[0072] S2`-2. Construct a spatial spectrum function using the orthogonality of the two subspaces, and perform a peak search in the entire angle space. The angles corresponding to the peaks are the azimuth and elevation angles of the signal wave direction.

[0073] Preferably, a two-dimensional MUSIC (Multiple Signal Classification) algorithm can also be used to calculate the azimuth and elevation angles of the signal wave direction.

[0074] Figure 6 The image shows the spatial spectrum energy map estimated by the two-dimensional MUSIC algorithm. The true arrival directions of the three signal sources are marked by green circles. The azimuth and elevation angles of these three signal sources are (-30°, 10°), (60°, 20°), and (120°, 60°), respectively. Meanwhile, the red crosses accurately indicate the corresponding peak positions estimated by the algorithm. It can be clearly observed from the image that the three red crosses and the corresponding green circles highly overlap in the bright area of ​​the spatial spectrum (yellow / green high-energy area), indicating that the algorithm accurately captured the azimuth and elevation angle information of the three signal sources.

[0075] S3. UAV position calculation based on time difference of arrival and angle of arrival Based on the two types of parameters extracted from S2, the drone's position is calculated separately to achieve multi-parameter fusion positioning.

[0076] (a) Position calculation based on Time Difference of Arrival (TDOA).

[0077] S3-1. Select one of the five sensing nodes as a reference node (sensing node 1 can be selected), select three as anchor nodes (sensing nodes 2, 3 and 4 can be selected in this embodiment), and use the remaining sensing node 5 as a decision node.

[0078] S3-2, Preliminary estimation of the drone's location.

[0079] Obtain the arrival time difference between the drone and the three anchor nodes and the reference node: The time difference between node 2 and node 1 is ; The time difference between node 3 and node 1 is ; The time difference between node 4 and node 1 is ; The time difference between node 5 and node 1 is .

[0080] Combining the three-dimensional coordinates of each sensing node, we substitute them into the Chan algorithm to solve the hyperbola equations. Since the Chan algorithm involves quadratic equations in the solution process, it will produce two possible solutions, namely two candidate positions P1 and P2, one of which is the true position and the other is a false root.

[0081] Calculate the theoretical distance differences between P1 and P2 and reference node 1 and decision node 5 respectively, and compare them with the distances calculated based on the theoretical distance differences between P1 and P2 and reference node 1 and decision node 5 respectively. By comparing the measured distance differences, the candidate location that is closer to the measured value is selected as the initial estimated location. Its three-dimensional coordinates are marked ,in, Represents the x-axis coordinate. Represents the y-axis coordinate. This represents the z-axis coordinate.

[0082] In this step, the introduction of decision nodes effectively solves the problem of fuzzy solutions in the traditional Chan algorithm, ensuring that the starting point for subsequent optimization iterations is correct.

[0083] S3-3. Using the initial estimated position as the initial value for iteration, the Taylor series expansion method is used for iterative optimization. Specifically, the nonlinear time difference of arrival observation equation is expanded to the first order Taylor at the initial value, linearized, and then the position correction is solved by the weighted least squares method to update the estimated position.

[0084] S3-4. Repeat the above iterative process until the iteration step size is less than the preset threshold or the preset number of iterations is reached. Then stop and use the iteration result at this point as the final estimated position. Its three-dimensional coordinates are marked ,in, Represents the x-axis coordinate. Represents the y-axis coordinate. This represents the z-axis coordinate.

[0085] Taylor's algorithm can quickly converge to a more accurate solution near the initial value, and its performance is particularly outstanding in high signal-to-noise ratio environments.

[0086] Figure 8The graph compares the positioning error performance of Taylor optimization and Chan algorithm alone under different signal-to-noise ratio (SNR) environments. The horizontal axis represents SNR, ranging from -20dB to 20dB, and the vertical axis represents positioning error in meters, ranging from 0 to 3.5 meters. Two curves are plotted: the red line represents the error curve using only the Chan algorithm, and the blue line represents the error curve after the initial estimation using the Chan algorithm followed by Taylor optimization iterations. The curve distribution shows that in the low SNR region (-20dB to -15dB), the two curves largely overlap, indicating that Taylor optimization has limited accuracy improvement under high noise levels. As the SNR increases, the two curves gradually diverge. In the 5dB to 10dB range, the Chan+Taylor curve becomes significantly lower than the Chan curve. In the high SNR region (15dB to 20dB), the Chan+Taylor curve error stabilizes at around 0.2 meters, while the Chan curve error is approximately 0.4 meters, a difference of about 0.2 meters. The distance between the two curves increases with the increase of signal-to-noise ratio, indicating that Taylor optimization is sensitive to the quality of the initial value and can effectively improve positioning accuracy under high signal-to-noise ratio. This figure verifies that Taylor iterative optimization can make up for the inaccuracy of the closed-form solution of the Chan algorithm, especially in high signal-to-noise ratio environments.

[0087] (ii) Position calculation based on the angle of arrival (AOA).

[0088] S3`-1. After estimating the azimuth and elevation angles of the incoming wave direction at each sensing node using the two-dimensional MUSIC (Multiple Signal Classification) algorithm, a spatial ray emanating from that node's location can be constructed. For ease of subsequent least squares calculations, the azimuth and elevation angles are calculated as follows: and The unit direction vector of the incoming wave signal measured by the i-th observation node. :

[0089] S3`-2. Due to the existence of angle measurement errors, multiple rays usually do not intersect at a single point. Therefore, the least squares method is used to estimate the most likely location of the signal source, that is, to minimize the sum of the squares of the perpendicular distances from all rays to the location to be estimated. The formula is as follows:

[0090] Where p is the UAV position vector to be estimated; s i Let be the position vector of the i-th sensing node; T represents the vector transpose; I is the identity matrix; This represents the projection matrix, used to project a vector onto a plane perpendicular to direction d. i On the plane, that is, calculate the perpendicular vector from the point to the ray; S3`-3, Taking the derivative with respect to p and setting the gradient to zero, we obtain the linear equation:

[0091] By summing the constraints of all observation nodes, a least-squares closed-form solution is obtained, which is then used as the final estimated position of the UAV. The calculation formula is:

[0092] This step aims to address the problem that multiple space rays typically do not intersect precisely due to angular measurement errors.

[0093] See also Figure 9 The graph shows the positioning error performance of the angle of arrival estimation algorithm under different signal-to-noise ratio (SNR) environments. The horizontal axis represents the SNR (-20dB~20dB), and the vertical axis represents the positioning error (0~16 meters). The curves correspond to the AOA positioning errors under different configurations (such as the number of array elements and nodes). The curves decrease as the SNR increases, decreasing more rapidly in the low SNR region (-20dB~-0dB) and stabilizing in the high SNR region (0dB~20dB). Compared with the TDOA method, AOA has a significant advantage in low SNR conditions, but its accuracy is slightly lower in high SNR conditions, verifying the complementary characteristics of the two methods.

[0094] By employing the two calculation methods described above, which are based on time difference of arrival and angle of arrival, this application can select or fuse the two positioning results to achieve complementary advantages depending on different signal environments. The time difference of arrival method has higher accuracy under high signal-to-noise ratio, while the angle of arrival method has better robustness under low signal-to-noise ratio.

[0095] Finally, this embodiment can select the position calculation result based on the time difference of arrival of the received signal or the position calculation result based on the angle of arrival of the received signal, or combine the two, according to actual needs, to obtain the final estimated position of the simulated UAV, and compare it with the true coordinates to verify the accuracy of the algorithm. Through the above steps, high-precision simulation positioning of low-altitude UAVs in urban environments is achieved.

[0096] Please see Figure 2 In response to the above method, this application also discloses a low-altitude unmanned aerial vehicle (UAV) positioning system for urban environments based on wireless signals, comprising: Distributed sensing nodes are deployed in different locations in the urban space to receive wireless signals broadcast by drones. Each distributed sensing node contains an array antenna and has time synchronization capabilities. The parameter configuration module is used to obtain pre-configured simulation scene parameters, which include the fixed spatial coordinates of each distributed sensing node and the true spatial coordinates of the simulated UAV's location. The signal simulation module is used to set the modulated wireless signal broadcast by the simulated drone for image transmission according to the simulation scene parameters, and to simulate the propagation process of the wireless signal in space to obtain the received signals of each sensing node. The parameter extraction module is used to obtain the time difference of the simulated drone arriving at different sensing nodes and the angle of arrival of the wireless signal broadcast by the simulated drone measured by a single sensing node, based on the received signals of each sensing node. The position calculation module is used to calculate the estimated position of the simulated drone based on the time difference of arrival and / or angle of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node.

[0097] This application also discloses an electronic device, which includes at least one processor and at least one memory, the memory being data-connected to the processor, wherein the memory stores instructions executable by at least one processor, the instructions being executed by at least one processor to enable at least one processor to perform any of the methods described above.

[0098] This application also discloses a computer-storeable medium storing computer instructions, which, when executed by a processor, specifically perform the steps of any of the methods described above.

[0099] This application also discloses a computer program product, including computer instructions, which, when executed by a processor, specifically perform the steps of any of the methods described above.

[0100] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0101] Although this application 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; and these 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 this application.

Claims

1. A method for locating low-altitude unmanned aerial vehicles (UAVs) in urban environments based on wireless signals, characterized in that: Includes the following steps: Obtain pre-configured simulation scene parameters, which include the pre-set fixed spatial coordinates of each distributed sensing node and the true spatial coordinates of the simulated UAV's location. Based on the simulation scenario parameters, a modulated wireless signal broadcast by the simulated drone for image transmission is set, and the propagation process of the wireless signal in space is simulated to obtain the received signals of each sensing node. Based on the received signals from each sensing node, signal processing is used to obtain the time difference between the arrival of the simulated drone at different sensing nodes and the angle of arrival of the wireless signal broadcast by the simulated drone as measured by a single sensing node. The estimated position of the simulated drone is calculated based on the time difference between the arrival times of the simulated drone at different sensing nodes extracted by the sensing nodes and the angle of arrival of the wireless signal broadcast by the simulated drone measured by a single sensing node.

2. The method according to claim 1, characterized in that, Simulating the propagation process of wireless signals in space, the received signals from each sensing node are obtained, including: Based on the spatial transmission distance from the simulated UAV to each sensing node, calculate the continuous signal received by each sensing node; Set the sampling period for each sensing node, and sample the continuous signal according to the sampling period to obtain the discrete signal sequence of each sensing node; The discrete signal sequence is interpolated to obtain a signal sequence with improved time resolution, which is then used as the received signal output.

3. The method according to claim 2, characterized in that, The interpolation process is a multiple interpolation process, which includes: According to a preset interpolation factor, new sampling points are inserted between adjacent sampling points to obtain a signal sequence in which both the time resolution and the interpolation factor are improved. The improved signal sequence is then output as the received signal.

4. The method according to claim 1, characterized in that, The time difference between the simulated drone's arrival at different sensing nodes, obtained through signal processing, includes: Select one sensing node as the reference node, and perform cross-correlation operations between the received signals of the other sensing nodes and the received signals of the reference node to obtain a sequence of cross-correlation functions as a function of relative sliding time. The relative sliding time position corresponding to the peak of the sequence of cross-correlation function changes with relative sliding time is selected, and the relative sliding time corresponding to the relative sliding time position is used as the time difference between the simulated UAV and the sensing node.

5. The method according to claim 4, characterized in that, The cross-correlation operation is performed by sliding the received signal of the reference node relative to the received signal of any sensing node in the time dimension and calculating the sum of the products of the overlapping parts, thus obtaining a sequence of cross-correlation functions changing with the relative sliding time.

6. The method according to claim 1, characterized in that, The angle of arrival of the simulated drone broadcast wireless signal, measured by a single sensing node, obtained through signal processing, includes: At each sensing node, the phase difference of the signals received by the array antenna is used to divide the received data space into mutually orthogonal signal subspace and noise subspace through feature decomposition; A spatial spectral function is constructed using the orthogonality of the two subspaces. Spectral peaks are searched throughout the entire angular space, and the azimuth and elevation angles of the received signal are determined based on the angles corresponding to the spectral peaks.

7. The method according to claim 1, characterized in that, Based on the time difference of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node, the estimated position of the simulated drone is calculated, including: Based on the time difference between the simulated UAV arriving at the sensing node and the reference node, candidate positions are obtained by solving a system of nonlinear equations. Select a sensing node as the decision node, calculate the theoretical distance difference between the candidate position and the reference node and the decision node respectively, compare the theoretical distance difference with the distance difference obtained by actual measurement based on the time difference of arrival, and select the candidate position that is closest to the actual value as the initial estimated position. Using the initial estimated position as the initial value for iteration, the nonlinear time difference of arrival observation equation is expanded in first order Taylor at the initial value. After linearization, the position correction is solved by weighted least squares method, and the estimated position is updated. Repeat the above process until the iteration step size is less than the preset threshold or / and the preset number of iterations is reached, then stop the iteration and use the iteration result at this point as the final estimated position.

8. The method according to claim 7, characterized in that, The iterative process of the Taylor series expansion method includes: The observation equation for the time difference of arrival is expanded in first order Taylor at the initial estimated position to establish the error equation. The position correction is solved by weighted least squares method, and the estimated position is updated. The above process is repeated until convergence.

9. The method according to claim 6, characterized in that, Based on the angle of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node, the estimated position of the simulated drone is calculated, including: A spatial ray is constructed based on the azimuth and elevation angles of the incoming wave direction of the signal received by the sensing node; The least squares method is used to solve for the intersection point of spatial rays. By minimizing the sum of squared vertical distances from all rays to the point to be estimated, a system of linear equations is established and solved to obtain the estimated position of the UAV.

10. The method according to claim 1, characterized in that, Before acquiring the pre-configured simulation scenario parameters, multiple distributed sensing nodes are set up. These distributed sensing nodes are deployed in different locations in the urban space to receive wireless signals broadcast by drones. Each distributed sensing node contains an array antenna and has time synchronization capabilities.

11. A low-altitude unmanned aerial vehicle (UAV) positioning system for urban environments based on wireless signals, characterized in that, include: Distributed sensing nodes are deployed in different locations in the urban space to receive wireless signals broadcast by drones. Each distributed sensing node contains an array antenna and has time synchronization capability. The parameter configuration module is used to obtain pre-configured simulation scene parameters, which include the fixed spatial coordinates of each distributed sensing node and the true spatial coordinates of the simulated UAV's location. The signal simulation module is used to set the modulated wireless signal broadcast by the simulated drone for transmitting images according to the simulation scenario parameters, and to simulate the propagation process of the wireless signal in space to obtain the received signals of each sensing node. The parameter extraction module is used to obtain the time difference of the simulated drone arriving at different sensing nodes and the angle of arrival of the wireless signal broadcast by the simulated drone measured by a single sensing node, based on the received signals of each sensing node. The position calculation module is used to calculate the estimated position of the simulated drone based on the time difference of arrival and / or angle of arrival of the wireless signal broadcast by the simulated drone extracted by the sensing node.

12. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they specifically perform the steps in the method as described in any one of claims 1-10.

Images