Method, system and device for lateral positioning of a drone based on spectrum sounding

Abstract

The application is suitable for the technical field of unmanned aerial vehicle radio spectrum detection, and provides an unmanned aerial vehicle lateral positioning method, system and device based on spectrum detection, which comprises the following steps: collecting radio spectrum signals in a space through a distributed antenna array; identifying a target frequency band and corresponding signal characteristics of a suspected unmanned aerial vehicle from the radio spectrum signals; extracting amplitude information and phase information of the target frequency band at different antenna nodes, and performing dimension normalization processing; constructing a lateral positioning model based on the normalized relative characteristic variables, calculating a lateral azimuth angle, and combining baseline topological information of the antenna array to solve two-dimensional lateral position coordinates of the unmanned aerial vehicle. After the position coordinates are solved under continuous time stamps, an extended Kalman filter is introduced to establish a motion state model of the unmanned aerial vehicle, so as to adaptively smooth and eliminate position mutations caused by spectrum multipath effects or instantaneous shielding.

Description

Technical Field

[0001] This invention belongs to the field of UAV radio spectrum detection technology, and particularly relates to a UAV lateral positioning method, system and device based on spectrum detection. Background Technology

[0002] Traditional drone detection and positioning methods mainly rely on active radar or automatic dependent surveillance broadcast (ADS-B). However, active radar has blind spots for detecting low, slow, and small targets and its electromagnetic radiation can easily reveal its own position. ADS-B is only applicable to aircraft that actively broadcast cooperative signals and is completely ineffective against "black flight" drones that turn off their transponders.

[0003] Existing spectrum detection lateral localization methods mainly estimate the direction of arrival by the amplitude or phase difference of the target signal received by multiple antenna nodes. Typical techniques include amplitude-to-arrival (AMA) localization and phase-to-arrival (PCO) localization. However, these methods face significant challenges in practical engineering applications.

[0004] The extraction and fusion of amplitude and phase information lacks dimensional normalization. Hardware differences such as receiving gain, cable loss, and analog-to-digital conversion quantization errors in different antenna channels cause the absolute measured values ​​of amplitude and phase in each channel to deviate from the theoretical values. Existing methods directly substitute the absolute amplitude or phase difference into the positioning formula, failing to eliminate hardware gain and dimensional differences. This results in poor consistency of positioning results in different environments or on different devices, and lateral accuracy is severely limited by the level of hardware calibration. Furthermore, the direction-finding fusion model does not incorporate signal-to-noise ratio (SNR) as a confidence weight into the multi-node fusion mechanism. In actual electromagnetic environments, due to differences in spatial distribution, the SNR of the target signal received by different antenna nodes fluctuates significantly. Nodes closer to the target or in directions with less interference should receive higher fusion weights. However, existing methods often use equal-weighted averaging or simple unweighted least squares algorithms, leading to measurement biases from low SNR nodes that severely drag down overall positioning accuracy. A robust fusion strategy based on SNR weighting is lacking.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] The purpose of this invention is to provide a UAV lateral positioning method based on spectrum detection, which aims to solve the above-mentioned problems.

[0007] This invention is implemented as follows: it provides a UAV lateral positioning method based on spectrum detection, comprising the following steps: S1. Acquire radio spectrum signals in the airspace through a distributed antenna array; S2. Extract features from the radio spectrum signal to identify the target frequency band of the suspected UAV and the corresponding signal features; S3. Extract the amplitude and phase information of the target frequency band at different antenna nodes, and perform dimension normalization processing to extract the relative feature variables that eliminate the hardware gain and dimension difference. S4. Based on the normalized relative feature variables, construct a lateral positioning model and calculate the lateral azimuth angle of the UAV; S5. Combining the multiple sets of lateral azimuth angles with the baseline topology information of the antenna array, the two-dimensional lateral position coordinates of the UAV are calculated.

[0008] As a further aspect of the present invention, the step S1 of acquiring radio spectrum signals in the airspace through a distributed antenna array specifically includes: Deploy at least three non-collinear broadband receiving antennas to form a distributed array; The predicted UAV image transmission frequency band and remote control communication frequency band are polled and scanned at a preset Nyquist sampling frequency to obtain a time-domain discrete complex baseband sequence containing environmental noise and target signals.

[0009] As a further aspect of the present invention, the feature extraction in step S2 specifically includes: The time-domain discrete complex baseband sequence is subjected to a fast Fourier transform to obtain the frequency domain energy distribution; the background noise is filtered out using an adaptive energy threshold method, and the center frequency, bandwidth, and frequency hopping period features corresponding to the energy peak are extracted; the frequency hopping period features are matched with a preset UAV communication protocol feature library, and if the similarity exceeds a preset threshold, the corresponding signal is confirmed as a target frequency band signal.

[0010] As a further aspect of the present invention, the construction of the lateral positioning model and the calculation of the lateral azimuth angle of the UAV in step S4 are specifically achieved through the following fusion model formula: First, for each effective detection node pair in the antenna array, calculate the proportion of the measured signal-to-noise ratio (SNR) of the received signal of that node pair to the sum of the measured SNR of all node pairs, and use this as the SNR weighting coefficient for that node pair. Then, calculate the normalized amplitude difference coefficient of the node pair multiplied by the sine value of the normalized relative phase offset to obtain the first product term, and the normalized amplitude difference coefficient multiplied by the cosine value of the normalized relative phase offset to obtain the second product term. The signal-to-noise ratio weighting coefficient of the node pair is multiplied by the first and second product terms mentioned above to obtain the directional component contribution value of the node pair. Then, the first product contribution values ​​of all node pairs are summed to obtain the directional sine sum term. The second product contribution values ​​of all node pairs are summed to obtain the directional cosine sum term. Finally, the arctangent value of the ratio obtained by dividing the directional sine sum term by the directional cosine sum term is calculated as the estimated value of the UAV lateral azimuth angle.

[0011] As a further aspect of the present invention, the normalized amplitude difference coefficient and the normalized relative phase offset are calculated using the following formulas: The normalized amplitude difference coefficient is calculated as follows: the difference between the absolute values ​​of the amplitudes of the target frequency band signals received by the first antenna and the second antenna at the same node is used as the numerator, the sum of the absolute values ​​of the amplitudes of the two channels plus a preset minimum positive number is used as the denominator, and the numerator is divided by the denominator to obtain the normalized amplitude difference coefficient. The normalized relative phase offset is calculated as follows: First, divide the center wavelength of the target frequency band by the physical baseline length between the node pairs to obtain the wavelength baseline ratio; then divide the original phase difference measurement between the first antenna and the second antenna in the node pair by the wavelength baseline ratio to obtain the preliminary phase offset; finally, multiply the preliminary phase offset by the array hardware phase calibration constant to obtain the normalized relative phase offset.

[0012] As a further aspect of the present invention, the step S5 of calculating the two-dimensional lateral position coordinates of the UAV specifically includes: Obtain the lateral azimuth angles output by at least two independent antenna subarrays distributed at different physical locations; Using the physical coordinates of each antenna subarray as the origin, a set of multi-directional intersecting ray equations is constructed in combination with the corresponding lateral azimuth angles; The equations of the multi-directional intersecting rays are solved using the least squares optimization algorithm to obtain the optimal intersection point coordinates that minimize the sum of the squares of the spatial distances of each directional ray, which are then used as the two-dimensional lateral position coordinates of the UAV.

[0013] As a further aspect of the present invention, after calculating the two-dimensional lateral position coordinates, the method further includes: Establish the state transition equation and observation equation for the two-dimensional planar motion of the UAV; input the two-dimensional lateral position coordinates calculated at multiple consecutive time stamps into the extended Kalman filter; The extended Kalman filter is used to smooth and remove outlier position coordinate changes caused by spectral multipath effects or instantaneous occlusion, and output a continuous and stable UAV lateral motion trajectory.

[0014] The present invention also provides a UAV lateral positioning system based on spectrum detection, the system being used to implement the aforementioned UAV lateral positioning method based on spectrum detection, the system comprising: The acquisition module is used to acquire radio spectrum signals in the airspace through a distributed antenna array; The feature extraction module is used to extract features from the radio spectrum signal and identify the target frequency band and corresponding signal features of the suspected UAV. The normalization module is used to extract the amplitude and phase information of the target frequency band at different antenna nodes, and perform dimensional normalization processing to extract the relative feature variables that eliminate the hardware gain and dimensional differences. The azimuth angle determination module is used to construct a lateral positioning model based on the normalized relative feature variables and calculate the lateral azimuth angle of the UAV. The positioning module is used to calculate the two-dimensional lateral position coordinates of the UAV by combining multiple sets of the lateral azimuth angles and the baseline topology information of the antenna array.

[0015] As a further aspect of the present invention, the acquisition module's method of acquiring radio spectrum signals in the airspace via a distributed antenna array specifically includes: Deploy at least three non-collinear broadband receiving antennas to form a distributed array; The predicted UAV image transmission frequency band and remote control communication frequency band are polled and scanned at a preset Nyquist sampling frequency to obtain a time-domain discrete complex baseband sequence containing environmental noise and target signals.

[0016] As a further aspect of the present invention, the feature extraction in the feature extraction module specifically includes: The time-domain discrete complex baseband sequence is subjected to a fast Fourier transform to obtain the frequency domain energy distribution; the background noise is filtered out using an adaptive energy threshold method, and the center frequency, bandwidth, and frequency hopping period features corresponding to the energy peak are extracted; the frequency hopping period features are matched with a preset UAV communication protocol feature library, and if the similarity exceeds a preset threshold, the corresponding signal is confirmed as a target frequency band signal.

[0017] The present invention also provides an apparatus comprising one or more processors and one or more memories, wherein at least one piece of program code is stored in the one or more memories, and when the program code is loaded and executed by the one or more processors, the UAV lateral positioning method based on spectrum detection is implemented.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: In this invention, during the lateral azimuth angle calculation, the proportion of the measured signal-to-noise ratio (SNR) of each node to the sum of the SNRs of all node pairs is used as a weighting coefficient to perform weighted fusion of the normalized amplitude difference and phase offset. This mechanism makes the measurement contribution of high SNR nodes stand out, while the interference of low SNR nodes is effectively suppressed, significantly improving the direction finding accuracy and noise resistance in harsh electromagnetic environments such as multipath interference.

[0019] In this invention, the lateral azimuth angles output by multiple independent subarrays in a distributed antenna array are used to construct a multi-directional intersecting ray equation system in combination with the physical coordinates of each subarray. The optimal intersection point coordinates are then solved by least squares optimization, which improves the positioning result from a single-direction line to two-dimensional planar coordinates. This solves the defect of traditional methods that cannot provide specific location information and meets the precise positioning requirements of actual countermeasure deployment.

[0020] In this invention, after calculating the position coordinates under continuous timestamps, an extended Kalman filter is introduced to establish the motion state model of the UAV. Adaptive smoothing and elimination are performed on the position changes caused by spectral multipath effects or instantaneous occlusion, and a continuous and stable motion trajectory is output, which significantly improves the reliability and trajectory continuity of dynamic tracking. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0022] Figure 1 This is a flowchart of a UAV lateral localization method based on spectrum detection.

[0023] Figure 2 This is a block diagram of the composition of a UAV lateral positioning system based on spectrum detection. Detailed Implementation

[0024] 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 and not intended to limit the invention.

[0025] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0026] Figure 1 The flowchart illustrates a spectrum-based UAV lateral localization method, which, according to an embodiment of the present invention, includes the following steps: S1. Acquire radio spectrum signals in the airspace using a distributed antenna array. This distributed antenna array can consist of multiple broadband receiving antennas deployed in different physical locations; for example, two or three antenna elements can be simply placed on the ground, ensuring a certain physical spacing between them. These antenna elements are configured to continuously monitor preset radio frequency bands; for example, a fixed frequency range can be set for scanning, or specific frequency points can be manually set for monitoring. The acquired signals can be raw analog radio frequency signals, which are then converted into digital signals by an analog-to-digital converter for further processing.

[0027] S2. Extract features from the radio spectrum signal to identify the target frequency bands and corresponding signal characteristics of suspected drones. After receiving a large amount of radio spectrum data, it is necessary to distinguish signals related to drones. One approach is to perform spectrum analysis on the acquired signals, such as obtaining the frequency components of the signal through a simple Fourier transform, then manually observing the spectrum graph, judging based on experience whether there are abnormal energy peaks, and recording the center frequency and approximate bandwidth corresponding to these peaks. These identified frequency bands and their corresponding energy intensity, duration, and other characteristics are initially considered as signal characteristics of suspected drones.

[0028] S3. Extract the amplitude and phase information of the target frequency band at different antenna nodes, and perform dimensional normalization processing to extract relative feature variables that eliminate hardware gain and dimensional differences. After identifying the target frequency band, it is necessary to obtain its amplitude and phase information from the signals received by different antennas. One implementation is to directly measure the absolute amplitude and original phase values ​​of the target signal received by each antenna channel. For example, a peak detector can be used to measure the amplitude, and a phase meter can be used to measure the phase. Then, in order to eliminate possible hardware gain differences between different antenna channels (e.g., different losses of different antennas or cables) and differences in measurement units, these original measurements can be simply linearly scaled, for example, by dividing all amplitude values ​​by a fixed reference value, or by converting the phase values ​​to a uniform unit of radians. Thus, feature variables reflecting the relative differences of the signals can be obtained.

[0029] S4. Based on the normalized relative feature variables, construct a lateral positioning model and calculate the lateral azimuth angle of the UAV. After obtaining the relative feature variables, these variables need to be used to estimate the direction of arrival of the UAV. One implementation method is to construct a geometric model, for example, based on the amplitude difference or phase difference of the signals received by the two antennas, and directly calculate a rough azimuth angle using a lookup table or a preset empirical formula. This model can be a pre-trained lookup table that maps specific amplitude differences or phase differences to corresponding azimuth angles.

[0030] S5. Combining the multiple sets of lateral azimuth angles with the baseline topology information of the antenna array, the two-dimensional lateral position coordinates of the UAV are calculated. After obtaining the lateral azimuth angle of a single antenna array, information from multiple observation points is needed to determine the specific two-dimensional position of the UAV. One implementation is to deploy multiple independent lateral positioning systems, each calculating an azimuth angle independently. Then, these azimuth angles are plotted as simple straight lines on a map, and the intersection of these lines is visually determined as the estimated position of the UAV. This method relies on manual judgment, and it is difficult to determine an optimal intersection point when multiple rays do not intersect precisely.

[0031] This embodiment overcomes the shortcomings of existing technologies in UAV lateral positioning in terms of accuracy, robustness, and two-dimensional coordinate output by introducing dimensional normalization processing, signal-to-noise ratio weighted fusion, and baseline topology-based two-dimensional intersection calculation, and provides a more accurate, stable, and practical UAV lateral positioning method.

[0032] In a preferred embodiment of the present invention, the step S1 of acquiring radio spectrum signals in the airspace via a distributed antenna array specifically includes: Deploy at least three non-collinear broadband receiving antennas to form a distributed array; The predicted UAV image transmission frequency band and remote control communication frequency band are polled and scanned at a preset Nyquist sampling frequency to obtain a time-domain discrete complex baseband sequence containing environmental noise and target signals.

[0033] In this embodiment, deploying "at least three non-collinear broadband receiving antennas" to form a distributed array means spatially configuring at least three broadband receiving antennas that are not located on the same straight line as each other.

[0034] "At least three" antenna nodes can provide enough measurement baselines to achieve a unique localization solution in a two-dimensional plane, avoiding the ambiguity that may occur when using only two antennas.

[0035] The "non-collinear" deployment method ensures that the antenna array has a good geometric configuration, enabling it to receive signals from different angles, thereby improving direction finding accuracy and positioning robustness, and avoiding the direction finding blind spots in certain directions of collinear arrays.

[0036] A "wideband receiving antenna" refers to an antenna that can cover a wide frequency range to accommodate various image transmission and remote control communication frequency bands that drones may use, ensuring the capture of target signals. For example, a wideband receiving antenna can take the form of a log-periodic antenna, an ultra-wideband antenna, or a multi-band antenna.

[0037] "Distributed array" refers to antenna nodes that are spaced apart rather than tightly packed. This deployment method allows for a larger effective aperture, thereby improving direction-finding accuracy and resistance to multipath interference. Furthermore, distributed arrays can cover a wider area, increasing the ability to detect long-range UAVs.

[0038] Polling the known UAV image transmission and remote control communication bands using a preset Nyquist sampling frequency means that during analog-to-digital conversion, the sampling frequency must be at least twice the highest frequency component of the signal to ensure complete signal reconstruction. Using a preset Nyquist sampling frequency ensures that spectral aliasing does not occur when converting analog radio spectrum signals to digital signals, thus preserving the complete information of the signal and providing a high-quality data foundation for subsequent feature extraction and localization.

[0039] "Predicted UAV image transmission and remote control communication frequency bands" refers to the specific frequency ranges that a UAV may use, based on existing UAV communication protocols and standards. Scanning these predicted frequency bands can effectively focus on target signals, avoiding a large amount of unnecessary data collection and processing on irrelevant frequency bands, thereby improving system efficiency and resource utilization.

[0040] "Polling scan" is an efficient spectrum monitoring strategy that involves the system sequentially scanning and collecting data from multiple target frequency bands in a pre-defined order. This approach allows the system to monitor multiple potential UAV communication frequency bands simultaneously with limited hardware resources, without the need to deploy a large number of parallel receivers, thus reducing system cost and complexity.

[0041] Ultimately, "obtaining a time-domain discrete complex baseband sequence containing environmental noise and the target signal" refers to the digital signal representation obtained after down-conversion and analog-to-digital conversion.

[0042] "Complex baseband" indicates that the signal contains both real (I-path) and imaginary (Q-path) parts, fully representing the signal's amplitude and phase information, which is crucial for subsequent phase information extraction and localization. "Discrete in time" indicates that the signal is a sequence of sampling points that is discontinuous in time, representing a fundamental form of digital signal processing. This sequence contains both the target UAV signal and unavoidable environmental noise, requiring subsequent processing to extract target signal features.

[0043] In a preferred embodiment of the present invention, the feature extraction in step S2 specifically includes: Perform a Fast Fourier Transform on the time-domain discrete complex baseband sequence to obtain the frequency-domain energy distribution; An adaptive energy thresholding method is used to filter out background noise and extract the center frequency, bandwidth and frequency hopping period features corresponding to the energy peak. The frequency hopping period feature is matched with a preset UAV communication protocol feature library. If the similarity exceeds a preset threshold, the corresponding signal is confirmed as a target frequency band signal. The preset threshold ranges from 0.7 to 0.95.

[0044] In this embodiment, a Fast Fourier Transform (FFT) is performed on the time-domain discrete complex baseband sequence to obtain the frequency-domain energy distribution. This process converts the original time-domain signal into a frequency-domain representation, enabling the analysis of the signal's frequency characteristics and energy distribution. This can be achieved by executing FFT algorithm library functions through a digital signal processor or a general-purpose processor.

[0045] An adaptive energy thresholding method is employed to filter out background noise, extracting the center frequency, bandwidth, and frequency hopping period characteristics corresponding to the energy peak. Its function is to effectively suppress random noise and interference in the environment, highlighting the energy peak of the target signal, thereby accurately extracting the key frequency characteristics of the UAV signal. This can be achieved by calculating the average noise power or noise power spectral density over a period of time, and then dynamically setting the threshold according to a preset signal-to-noise ratio requirement. For example, a sliding window averaging method can be used to estimate the noise floor, and then multiplied by a factor to obtain the dynamic threshold.

[0046] The frequency hopping cycle feature is matched against a preset UAV communication protocol feature library. If the similarity exceeds a preset threshold, the corresponding signal is confirmed as a target frequency band signal. The preset threshold ranges from 0.7 to 0.95. Its purpose is to further accurately identify and confirm UAV signals by comparing the extracted signal features with the features of known UAV communication protocols, thereby reducing the false alarm rate. This can be achieved by calculating the correlation coefficient or Euclidean distance between the extracted frequency hopping cycle feature and the features of each protocol in the feature library as a similarity metric. The preset threshold of 0.7 to 0.95 controls the strictness of the matching to ensure the accuracy of identification.

[0047] In a preferred embodiment of the present invention, the construction of the lateral positioning model and the calculation of the lateral azimuth angle of the UAV in step S4 are specifically achieved through the following fusion model formula: in, This represents the estimated lateral azimuth angle of the UAV calculated from the solution. The total number of valid detection node pairs in the antenna array is a positive integer; The signal-to-noise ratio weight coefficient of the i-th node pair is dimensionless. ,in Let be the measured signal-to-noise ratio of the received signal at the i-th node; Represents the normalized amplitude difference coefficient of the i-th node with respect to the received signal, which is dimensionless; This represents the normalized relative phase offset of the i-th node to the received signal, with its dimension in radians; all input parameters in the formula have been normalized to ensure the consistency and convergence of the computational dimensions.

[0048] The signal-to-noise ratio of the received signal at the i-th node The method is as follows: within the target frequency band, the ratio of the peak power of the measured signal to the average noise power in the adjacent no-signal frequency band is used, and the result is smoothed by using a sliding window averaging.

[0049] In this embodiment, the above-mentioned fusion model formula aims to use mathematical methods to comprehensively process the normalized amplitude difference coefficients and normalized relative phase offsets obtained from multiple antenna nodes, so as to obtain a unified and accurate lateral azimuth angle estimate.

[0050] in, This represents the estimated angle of the UAV relative to the antenna array baseline or reference direction, and is a key parameter in the positioning process.

[0051] This refers to the number of effective antenna node pairs participating in the calculation, and its value usually depends on the array configuration and signal detection conditions. As a signal-to-noise ratio (SNR) weighting coefficient, its function is to assign different weights to each node pair based on the quality (SNR) of the received signal. Node pairs with higher SNR will receive greater weights, thus playing a more important role in azimuth angle calculation. The calculation method of this weighting coefficient ensures that the sum of the weights of all effective node pairs is 1, achieving an adaptive weighted average. and These represent the normalized amplitude difference and normalized relative phase offset of the i-th node pair, respectively. They are pre-processed (e.g., dimensionally normalized) characteristic variables used to characterize the amplitude and phase differences of the signal arriving at different antennas. These differences are directly related to the incident direction of the signal. All input parameters are normalized before being substituted into the formula, which is crucial for eliminating the influence of different physical dimensions and ensuring the stability and convergence of the calculation. For example, besides the signal-to-noise ratio weighting coefficient... In addition to the calculation method, other weight allocation strategies can also be adopted, such as weighting based on signal strength, channel quality indicators, or fixed preset values.

[0052] The scheme in this application uses the aforementioned fusion model formula to weightedly fuse the normalized amplitude difference coefficients and normalized relative phase offsets extracted from multiple node pairs of a distributed antenna array. Specifically, for each node pair, its normalized amplitude difference coefficients... and normalized relative phase offset Used to calculate the horizontal and vertical components of a direction vector (i.e. and These components are then weighted according to the signal-to-noise ratio of the node pair. Weighting is performed. The weighted components of all valid node pairs are summed to form a combined horizontal and vertical component. Finally, these two combined components are converted into an estimate of the UAV's lateral azimuth using the arctangent function (arctan). This weighted fusion mechanism enables the system to fully utilize multi-source information, and when signal quality is uneven, it uses signal-to-noise ratio weighting coefficients. Prioritizing high-quality measurement data effectively suppresses the impact of noise and interference on direction-finding accuracy. In this way, the model can comprehensively determine the lateral bearing of the UAV from multiple angles and dimensions, significantly improving the accuracy and robustness of azimuth angle calculation and solving the challenge of accurate direction finding in complex electromagnetic environments.

[0053] As a preferred embodiment of the present invention, the normalized amplitude difference coefficient and normalized relative phase offset It is obtained by calculation using the following formula: ; in, and These represent the absolute values ​​of the target frequency band signal amplitudes received by the first antenna and the second antenna at the i-th node, respectively, and both have the same voltage or power dimensions. The original phase difference measurement between the first antenna and the second antenna is expressed in radians. The center wavelength of the target frequency band is represented by meters. Represents the physical baseline length between the i-th node pair, in meters; This is the array hardware phase calibration constant, dimensionless. Used to correct inherent phase deviations between different channels; For a preset minimum positive number, This represents the equivalent phase offset after baseline length normalization and hardware calibration.

[0054] In this embodiment, the normalized amplitude difference coefficient It is a relative quantity used to characterize the difference in amplitude of the signals received by the two antennas in the i-th node pair. It is calculated by taking the absolute values ​​of the amplitudes of the target frequency band signals received by the two antennas. and The difference is divided by its sum, and a preset minimum positive number is added. This calculation method effectively eliminates the influence of factors such as transmit power and propagation distance attenuation on the absolute signal strength, focusing instead on the relative amplitude difference of the signal at different antennas. This provides a more direct indication of the signal's incident direction. For example, the analog signal can be converted to a digital signal using an analog-to-digital converter at the receiver's front end, and then the absolute amplitude value can be obtained through digital signal processing algorithms (such as root mean square value calculation), and then substituted into the formula for calculation. Alternatively, the signal can be processed using a logarithmic amplifier, followed by differential and normalization operations.

[0055] Normalized relative phase offset This is a relative quantity characterizing the phase difference between the received signals from the two antennas in the i-th node pair, after baseline length normalization and hardware calibration. This calculation method uses the original phase difference measurement value... Divide by (the center wavelength of the target frequency band) With physical baseline length The ratio), then multiplied by the array hardware phase calibration constant. This processing method converts the raw phase difference into a physical quantity directly related to the signal incident angle, while eliminating the inherent phase deviation between different antenna channels and the influence of the physical baseline length on phase measurement. For example, the raw phase difference can be obtained through digital down-conversion and phase detection algorithms (such as phase extraction after fast Fourier transform or arctangent calculation after IQ demodulation), then normalized according to the known center wavelength and baseline length, and a pre-calibrated hardware phase calibration constant can be applied. Alternatively, the phase difference can be measured using high-precision phase-locked loop technology, and then subsequent normalization and calibration can be performed by a digital signal processor.

[0056] and These represent the absolute amplitude values ​​of the target frequency band signals received by the first and second antennas at the i-th node, respectively, both having the same voltage or power dimensions. They are the fundamental inputs for calculating the normalized amplitude difference coefficients. For example, the received signal's amplitude can be detected by a detector circuit or digital signal processor in the receiver to obtain its instantaneous or average amplitude value. Alternatively, it can be obtained by energy integration or peak detection of the received time-domain signal.

[0057] This represents the original phase difference measurement between the first and second antennas, measured in radians. It is a key input for calculating the normalized relative phase offset. For example, the time delay corresponding to the peak value can be found by performing cross-correlation on the signals received by the two antennas, and then converted into a phase difference. Alternatively, the I / Q components can be obtained by quadrature demodulating the two signals, then the phase can be calculated using the arctangent function, and finally the phase difference can be obtained by subtraction.

[0058] The center wavelength represents the target frequency band and is measured in meters. It is an important parameter in the phase normalization process, used to convert the phase difference into a quantity related to physical distance. For example, it can be calculated using the formula of the speed of light divided by the frequency, based on the center frequency of the identified target frequency band. Alternatively, during system design, the corresponding center wavelength value can be pre-stored for a preset UAV communication frequency band.

[0059] This represents the physical baseline length between the i-th node pair, measured in meters. It is used in phase normalization to correlate the phase difference with the antenna spacing. For example, in antenna array deployment, the physical distance between each antenna node can be accurately measured and recorded using high-precision measuring tools such as laser rangefinders or GPS RTK modules.

[0060] This is a dimensionless phase calibration constant for the array hardware. It is used to correct inherent phase deviations between different channels, ensuring the accuracy of phase measurements. For example, before the system is put into use, this calibration constant can be calibrated and stored by placing a reference signal source at a known location and measuring the phase response of each receiving channel. Alternatively, adaptive calibration techniques can be used to dynamically adjust this constant by analyzing the received known reference signal or a stable signal in the environment.

[0061] This is a preset, extremely small positive number. Its purpose is to prevent the denominator from being zero when calculating the normalized amplitude difference coefficient, thereby avoiding calculation errors and improving numerical stability. For example, in software implementation, it can be set to a very small floating-point number, such as 1e-9 or 1e-12, to ensure that the denominator is not zero when the amplitude values ​​are extremely small or equal.

[0062] In a preferred embodiment of the present invention, the step S5 of calculating the two-dimensional lateral position coordinates of the UAV specifically includes: Obtain the lateral azimuth angles output by at least two independent antenna subarrays distributed at different physical locations; Using the physical coordinates of each antenna subarray as the origin, a set of multi-directional intersecting ray equations is constructed in combination with the corresponding lateral azimuth angles; The equations of the multi-directional intersecting rays are solved using the least squares optimization algorithm to obtain the optimal intersection point coordinates that minimize the sum of the squares of the spatial distances of each ray. , which serves as the two-dimensional lateral position coordinates of the UAV.

[0063] In this embodiment, acquiring the lateral azimuth angles output by at least two independent antenna subarrays distributed at different physical locations refers to the need to obtain direction-finding information from multiple spatially separated observation points in order to accurately determine the two-dimensional position of the UAV. Each independent antenna subarray can independently perform steps such as spectrum signal acquisition, feature extraction, and lateral azimuth angle calculation, thereby providing independent direction-finding data for subsequent positioning. These subarrays can be independent physical units composed of several antenna nodes, each with independent signal processing capabilities; or, a large distributed antenna array can be logically divided into multiple functionally independent subarrays as needed, with each subarray responsible for the direction-finding task within its coverage area.

[0064] Using the physical coordinates of each antenna subarray as the origin, and combining the corresponding lateral azimuth angles, a system of multi-directional intersecting ray equations is constructed. This involves transforming the physical deployment location (i.e., its coordinates) of each independent antenna subarray and its measured UAV lateral azimuth angle into a mathematical model. The physical coordinates of each subarray serve as the starting point of the ray, while the measured lateral azimuth angle defines the direction of the ray in the two-dimensional plane.

[0065] The equations of the multi-directional intersecting rays are solved using the least squares optimization algorithm to obtain the optimal intersection point coordinates that minimize the sum of the squares of the spatial distances of each ray. The two-dimensional lateral position coordinates of the UAV refer to the fact that, due to unavoidable errors in actual measurements, multiple direction-finding rays often cannot precisely intersect at a single point. To obtain the most reliable positioning results, this application employs a least-squares optimization algorithm. This algorithm finds an optimal two-dimensional coordinate point. This minimizes the sum of squares of the spatial distances from the point to all the directional rays. The coordinates corresponding to this minimum sum of squares are considered the optimal two-dimensional lateral position estimate for the UAV, thus effectively reducing the impact of measurement errors on positioning accuracy.

[0066] In a preferred embodiment of the present invention, after calculating the two-dimensional lateral position coordinates, the method further includes: Establish the state transition equation and observation equation for the two-dimensional planar motion of the UAV; The two-dimensional lateral position coordinates calculated at multiple consecutive timestamps are input into the extended Kalman filter; The extended Kalman filter is used to smooth and remove outlier position coordinate changes caused by spectral multipath effects or instantaneous occlusion, and output a continuous and stable UAV lateral motion trajectory.

[0067] In this embodiment, establishing the state transition equations and observation equations for the UAV's two-dimensional planar motion is fundamental to the Kalman filter. The state transition equations describe how the UAV's position, velocity, and other state variables evolve over time in the two-dimensional plane, typically based on the UAV's kinematic model. For example, it can be assumed that the UAV moves at a constant velocity in a straight line for a short time; its state vector can contain position and velocity components, and the state transition equations define the relationship between these components within a time step. The observation equations describe how to predict the two-dimensional lateral position coordinates calculated using the aforementioned method from the UAV's actual state. For example, the observation equations can directly use the position component in the state vector as the observed value. The establishment of these equations provides a dynamic model for subsequent filter processing.

[0068] Subsequently, the two-dimensional lateral position coordinates calculated at multiple consecutive time stamps are input into the extended Kalman filter. This step is crucial for combining real-time measurement data with a dynamic model. The extended Kalman filter is a recursive filter that can optimally estimate the system state using historical data and current measurements. By continuously inputting position coordinates at multiple time stamps, the filter can continuously update its estimate of the UAV's current state and predict its future state.

[0069] The extended Kalman filter (EPF) is used to smooth and eliminate outlier position coordinate abrupt changes caused by spectral multipath effects or momentary occlusion, outputting a continuous and stable UAV lateral motion trajectory. The EPF operates through a prediction-update loop. In the prediction phase, the filter predicts the current state based on the state transition equation and the state estimate from the previous time step. In the update phase, the filter compares the actual measured value (i.e., the calculated two-dimensional lateral position coordinates) with the predicted value and corrects the state estimate based on measurement noise and model uncertainty. When outliers appear in the measured values, i.e., when there is a significant deviation from the predicted value, the EPF automatically reduces the weight of the outlier measurement value according to its internal statistical model, thereby effectively suppressing its influence on the state estimate, achieving trajectory smoothing and outlier removal.

[0070] As a preferred embodiment of the technical solution of the present invention, the method further includes: Based on the signal-to-noise ratio, received signal strength, and historical direction finding stability of each distributed antenna node, calculate the node quality score for each antenna node. Based on the node quality score, each antenna node is sorted, and combined with the spatial distribution relationship between each antenna node, a set of effective positioning nodes participating in the current positioning calculation is selected. The lateral positioning calculations in steps S3 to S5 are performed using only the amplitude and phase information corresponding to the set of effective positioning nodes, and the antenna nodes that have not entered the set of effective positioning nodes are controlled to enter a low-power listening state. Predict the target area of ​​the UAV at the next moment based on the continuous positioning results, pre-activate the antenna nodes near the target area, and release the antenna nodes far away from the target area. The process of repeatedly performing node quality scoring, effective positioning node screening, and dynamic updating of positioning nodes continues until the UAV target leaves the monitoring area.

[0071] In one example of the technical solution of this invention, based on the completion of basic spectrum positioning, the antenna nodes participating in positioning are dynamically filtered to reduce system resource consumption and improve positioning efficiency. Traditional methods usually assume that all distributed antennas always participate in signal acquisition and positioning calculation. However, in practical applications, not all antennas can receive high-quality UAV signals. For example, some antennas are far away from the UAV or are affected by factors such as building obstruction and electromagnetic interference, resulting in poor signal quality. If all these antennas are still used for positioning, it will not only increase the amount of computation but may also reduce the accuracy of the positioning results.

[0072] Furthermore, the quality of the target signal currently received by each antenna node is evaluated. The evaluation can comprehensively consider the strength of the received signal, the signal-to-noise ratio, and the stability of the node during the historical positioning process, so as to determine whether each antenna node is suitable for participating in the positioning. Subsequently, antenna nodes with better quality are selected based on the evaluation results, and further screening is carried out based on the location distribution of these nodes in the monitoring area. The purpose of this is not only to select nodes with good signal quality, but also to avoid all nodes being concentrated in the same area, but to distribute them as much as possible in different locations around the target, so as to receive target signals from multiple directions and improve the stability of subsequent lateral positioning calculations.

[0073] Once the nodes involved in the positioning are identified, only these nodes are kept in normal working order, while the remaining nodes enter a low-power listening state or reduce their sampling frequency, retaining only basic monitoring capabilities. As the drone moves, the system predicts its next possible location based on the most recent positioning results and activates antenna nodes near the target in advance, while shutting down nodes far from the target area, so that the devices involved in the positioning can continuously adjust as the target moves.

[0074] In this way, the system always uses only a small number of high-quality antennas with reasonable spatial distribution to complete the positioning calculation, without all devices needing to maintain high load operation, thereby reducing device power consumption and computing resource consumption, while maintaining good positioning accuracy and real-time performance.

[0075] As a preferred embodiment of the technical solution of the present invention, the method further includes: The positioning stability corresponding to the current positioning result is calculated based on the changes in lateral azimuth angle and two-dimensional lateral position coordinates obtained from multiple consecutive positioning cycles. When the positioning stability is lower than a preset threshold, the target frequency band signal is reacquired for each antenna node involved in the positioning, and the normalized amplitude difference coefficient and the normalized relative phase offset are recalculated. Based on the recalculated amplitude information, phase information, and node quality scores corresponding to each antenna node, the UAV's lateral azimuth angle is recalculated, and the two-dimensional lateral position coordinates are updated. When the positioning stability is higher than the preset threshold, maintain the current positioning result and reduce the target frequency band scanning frequency; otherwise, continue to perform the remeasurement process until the positioning stability reaches the preset requirement.

[0076] In one embodiment of the technical solution of this invention, the spectrum acquisition and positioning calculation process is further dynamically adjusted according to the stability of the UAV positioning results to avoid unnecessary recalculation and improve system operating efficiency. In the actual environment, radio signals are easily affected by factors such as building reflection, tree obstruction and interference from other wireless devices, which can cause large fluctuations in the calculated azimuth or position coordinates at certain times. If the system maintains the same scanning frequency and calculation method regardless of whether the positioning is stable, it will not only increase the equipment burden, but may also process a large amount of invalid data.

[0077] To address the above situation, the positioning results obtained from multiple consecutive positioning cycles are first compared to determine whether the UAV's position changes are continuous and stable. If the changes in the continuous positioning results are small, it indicates that the current positioning is relatively stable, and the existing measurement results can be considered to have high reliability. In this case, there is no need to frequently remeasure. The target frequency band scanning frequency can be appropriately reduced to reduce the number of data acquisitions and calculations, thereby reducing system power consumption.

[0078] Conversely, when the continuous positioning results fluctuate significantly, it indicates that the current positioning may be affected by environmental interference or the target may be moving rapidly. At this time, the system re-acquires the target frequency band signal and recalculates the amplitude information, phase information, and lateral azimuth angle to correct the positioning results in a timely manner. At the same time, it can also combine the signal quality of each antenna node to re-complete the positioning calculation, thereby improving the reliability of the positioning results.

[0079] By repeatedly judging positioning stability, adjusting scanning frequency, and re-measuring when necessary, the system can automatically select the appropriate working mode according to the actual positioning situation. When positioning is stable, it reduces resource consumption; when positioning is unstable, it improves measurement accuracy, achieving a dynamic balance between positioning performance and system power consumption, thereby improving the operating efficiency and environmental adaptability of the entire UAV spectrum detection and positioning system.

[0080] Figure 2 The diagram shows the structural composition of a spectrum-detection-based UAV lateral positioning system. A spectrum-detection-based UAV lateral positioning system is also provided to implement the aforementioned spectrum-detection-based UAV lateral positioning method. The system includes: Acquisition module 11 is used to acquire radio spectrum signals in the airspace through a distributed antenna array; Feature extraction module 12 is used to extract features from the radio spectrum signal and identify the target frequency band and corresponding signal features of the suspected UAV. The normalization module 13 is used to extract the amplitude and phase information of the target frequency band at different antenna nodes, and perform dimensional normalization processing to extract the relative feature variables that eliminate the hardware gain and dimensional differences. The azimuth angle determination module 14 is used to construct a lateral positioning model based on the normalized relative feature variables and calculate the lateral azimuth angle of the UAV. The positioning module 15 is used to calculate the two-dimensional lateral position coordinates of the UAV by combining multiple sets of the lateral azimuth angles and the baseline topology information of the antenna array.

[0081] Furthermore, the acquisition module 11, which acquires radio spectrum signals in the airspace via a distributed antenna array, specifically includes: Deploy at least three non-collinear broadband receiving antennas to form a distributed array; The predicted UAV image transmission frequency band and remote control communication frequency band are polled and scanned at a preset Nyquist sampling frequency to obtain a time-domain discrete complex baseband sequence containing environmental noise and target signals.

[0082] Specifically, the feature extraction described in feature extraction module 12 includes: The time-domain discrete complex baseband sequence is subjected to a fast Fourier transform to obtain the frequency domain energy distribution; the background noise is filtered out using an adaptive energy threshold method, and the center frequency, bandwidth, and frequency hopping period features corresponding to the energy peak are extracted; the frequency hopping period features are matched with a preset UAV communication protocol feature library, and if the similarity exceeds a preset threshold, the corresponding signal is confirmed as a target frequency band signal.

[0083] In this embodiment, the core innovation lies in the dimensional normalization of amplitude and phase information to eliminate hardware gain and dimensional differences. Combined with the signal-to-noise ratio weighted lateral positioning model introduced in the positioning module 15 and the two-dimensional coordinate calculation mechanism based on baseline topology, the key problems of hardware gain mismatch, low signal-to-noise ratio interference and two-dimensional positioning loss in multi-node fusion are effectively overcome, thereby achieving the effect of improving lateral positioning accuracy and robustness.

[0084] Specifically, the normalization module 13 converts the original measurements of different antenna nodes into dimensionless relative characteristic variables by calculating the normalized amplitude difference coefficient and the normalized relative phase offset, thus avoiding absolute measurement deviations caused by receiver gain, cable loss, or analog-to-digital conversion errors. The positioning module 15 further constructs an azimuth angle calculation formula based on the normalized amplitude difference coefficient, the normalized relative phase offset, and the signal-to-noise ratio weighting coefficient, so that the measurement results of high signal-to-noise ratio nodes receive higher weight in the fusion process, suppressing the interference of low signal-to-noise ratio nodes. At the same time, the positioning module 15 obtains multiple sets of lateral azimuth angles and combines them with the baseline topology information of the antenna array, uses the least squares optimization algorithm to solve the multi-directional intersection ray equation system, and outputs the optimal intersection point coordinates that minimize the sum of the squares of the spatial distances of each directional ray, thereby accurately calculating the two-dimensional lateral position coordinates of the UAV.

[0085] As a preferred embodiment of the technical solution of the present invention, an apparatus is also provided, the apparatus including one or more processors and one or more memories, wherein at least one piece of program code is stored in the one or more memories, and when the program code is loaded and executed by the one or more processors, the UAV lateral positioning method based on spectrum detection is implemented.

[0086] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for lateral localization of unmanned aerial vehicles based on spectrum detection, characterized in that, Includes the following steps: S1. Acquire radio spectrum signals in the airspace through a distributed antenna array; S2. Extract features from the radio spectrum signal to identify the target frequency band of the suspected UAV and the corresponding signal features; S3. Extract the amplitude and phase information of the target frequency band at different antenna nodes, and perform dimension normalization processing to extract the relative feature variables that eliminate the hardware gain and dimension difference. S4. Based on the normalized relative feature variables, construct a lateral positioning model and calculate the lateral azimuth angle of the UAV; S5. Combining the multiple sets of lateral azimuth angles with the baseline topology information of the antenna array, the two-dimensional lateral position coordinates of the UAV are calculated.

2. The UAV lateral positioning method based on spectrum detection as described in claim 1, characterized in that, Step S1, which involves acquiring radio spectrum signals in the airspace using a distributed antenna array, specifically includes: Deploy at least three non-collinear broadband receiving antennas to form a distributed array; The predicted UAV image transmission frequency band and remote control communication frequency band are polled and scanned at a preset Nyquist sampling frequency to obtain a time-domain discrete complex baseband sequence containing environmental noise and target signals.

3. The UAV lateral positioning method based on spectrum detection as described in claim 1, characterized in that, The feature extraction described in step S2 specifically includes: The time-domain discrete complex baseband sequence is subjected to a fast Fourier transform to obtain the frequency domain energy distribution; the background noise is filtered out using an adaptive energy threshold method, and the center frequency, bandwidth, and frequency hopping period features corresponding to the energy peak are extracted; the frequency hopping period features are matched with a preset UAV communication protocol feature library, and if the similarity exceeds a preset threshold, the corresponding signal is confirmed as a target frequency band signal.

4. The UAV lateral positioning method based on spectrum detection according to claim 1, characterized in that, The construction of the lateral positioning model and the calculation of the lateral azimuth angle of the UAV in step S4 are specifically achieved through the following fusion model formula: First, for each effective detection node pair in the antenna array, calculate the proportion of the measured signal-to-noise ratio (SNR) of the received signal of that node pair to the sum of the measured SNR of all node pairs, and use this as the SNR weighting coefficient for that node pair. Then, calculate the normalized amplitude difference coefficient of the node pair multiplied by the sine value of the normalized relative phase offset to obtain the first product term, and the normalized amplitude difference coefficient multiplied by the cosine value of the normalized relative phase offset to obtain the second product term. The signal-to-noise ratio weighting coefficient of the node pair is multiplied by the first and second product terms mentioned above to obtain the directional component contribution value of the node pair. Then, the first product contribution values ​​of all node pairs are summed to obtain the directional sine sum term. The second product contribution values ​​of all node pairs are summed to obtain the directional cosine sum term. Finally, the arctangent value of the ratio obtained by dividing the directional sine sum term by the directional cosine sum term is calculated as the estimated value of the UAV lateral azimuth angle.

5. The UAV lateral positioning method based on spectrum detection according to claim 3, characterized in that, The normalized amplitude difference coefficient and the normalized relative phase offset are calculated using the following formulas: The normalized amplitude difference coefficient is calculated as follows: the difference between the absolute values ​​of the amplitudes of the target frequency band signals received by the first antenna and the second antenna at the same node is used as the numerator, the sum of the absolute values ​​of the amplitudes of the two channels plus a preset minimum positive number is used as the denominator, and the numerator is divided by the denominator to obtain the normalized amplitude difference coefficient. The normalized relative phase offset is calculated as follows: First, divide the center wavelength of the target frequency band by the physical baseline length between the node pairs to obtain the wavelength baseline ratio; then divide the original phase difference measurement between the first antenna and the second antenna in the node pair by the wavelength baseline ratio to obtain the preliminary phase offset; finally, multiply the preliminary phase offset by the array hardware phase calibration constant to obtain the normalized relative phase offset.

6. The UAV lateral positioning method based on spectrum detection according to claim 5, characterized in that, Step S5, which involves calculating the two-dimensional lateral position coordinates of the UAV, specifically includes: Obtain the lateral azimuth angles output by at least two independent antenna subarrays distributed at different physical locations; Using the physical coordinates of each antenna subarray as the origin, a set of multi-directional intersecting ray equations is constructed in combination with the corresponding lateral azimuth angles; The equations of the multi-directional intersecting rays are solved using the least squares optimization algorithm to obtain the optimal intersection point coordinates that minimize the sum of the squares of the spatial distances of each directional ray, which are then used as the two-dimensional lateral position coordinates of the UAV.

7. The UAV based on spectrum detection according to claim 5 The lateral positioning method is characterized by, After calculating the two-dimensional lateral position coordinates, the method further includes: Establish the state transition equation and observation equation for the two-dimensional planar motion of the UAV; input the two-dimensional lateral position coordinates calculated at multiple consecutive time stamps into the extended Kalman filter; The extended Kalman filter is used to smooth and eliminate outlier position coordinate changes caused by spectral multipath effects or instantaneous occlusion, and output a continuous and stable UAV lateral motion trajectory.

8. A spectrum-detection-based unmanned aerial vehicle (UAV) lateral positioning system, the system being used to implement the spectrum-detection-based UAV positioning as described in any one of claims 1 to 7. The lateral positioning method is characterized by, The system includes: The acquisition module is used to acquire radio spectrum signals in the airspace through a distributed antenna array; The feature extraction module is used to extract features from the radio spectrum signal and identify the target frequency band and corresponding signal features of the suspected UAV. The normalization module is used to extract the amplitude and phase information of the target frequency band at different antenna nodes, and perform dimensional normalization processing to extract the relative feature variables that eliminate the hardware gain and dimensional differences. The azimuth angle determination module is used to construct a lateral positioning model based on the normalized relative feature variables and calculate the lateral azimuth angle of the UAV. The positioning module is used to calculate the two-dimensional lateral position coordinates of the UAV by combining multiple sets of the lateral azimuth angles and the baseline topology information of the antenna array.

9. The UAV lateral positioning system based on spectrum detection as described in claim 8, characterized in that, The acquisition module describes acquiring radio spectrum signals in the airspace via a distributed antenna array, specifically including: Deploy at least three non-collinear broadband receiving antennas to form a distributed array; The predicted UAV image transmission frequency band and remote control communication frequency band are polled and scanned at a preset Nyquist sampling frequency to obtain a time-domain discrete complex baseband sequence containing environmental noise and target signals.

10. An apparatus, characterized in that, The device includes one or more processors and one or more memories, wherein at least one piece of program code is stored in the one or more memories, and when the program code is loaded and executed by the one or more processors, it implements the UAV lateral positioning method based on spectrum detection as described in any one of claims 1 to 7.

Images