An SDR-based FPV unmanned aerial vehicle precision detection system and method

Abstract

The application provides an FPV unmanned aerial vehicle accurate detection system and method based on SDR, and belongs to the technical field of radio detection, low-altitude security and anti-unmanned aerial vehicle.The method comprises the following steps: performing global radio frequency signal scanning and collecting on a target airspace through a spectrum scanning unit in an SDR radio frequency collecting module, adaptively matching the commonly used communication frequency bands of FPV unmanned aerial vehicle remote control and image transmission in the range of 20MHz to 6GHz frequency bands by the reconfigurable characteristics of the SDR radio frequency collecting module, dynamically switching the sampling rate by a signal processing module, synchronously capturing the weak radio frequency signals in the frequency hopping and fixed frequency working modes, and generating an original radio frequency signal data set; the application realizes the software reconfiguration of the radio frequency function by relying on the SDR technology, has wide frequency band coverage and strong adaptability, and can accurately identify various FPV unmanned aerial vehicles; by the radio frequency fingerprint identification technology, the false detection rate is greatly reduced, and the detection and identification accuracy can reach more than 95%.

Description

Technical Field

[0001] This invention proposes a precise detection system and method for FPV drones based on SDR, belonging to the fields of radio detection, low-altitude security and anti-drone technology. Background Technology

[0002] Currently, FPV drones are widely used in aerial photography and racing due to their small size, high maneuverability, low flight altitude, and agile obstacle avoidance. However, unauthorized flights, trespassing, and privacy breaches are frequent, posing serious security risks to key security areas such as airports, factories, and government parks. Existing drone detection technologies have significant shortcomings: traditional fixed radio frequency detection equipment uses fixed frequency bands and lacks versatility, making it difficult to adapt to the 2.4GHz and 5.8GHz frequency hopping communication bands commonly used by FPV drones; radar detection equipment is expensive, easily affected by urban building obstructions, and has low detection accuracy for small, lightweight FPV drones; conventional monitoring systems have weak anti-interference capabilities, are susceptible to clutter and civilian wireless signal interference in complex electromagnetic environments, cannot distinguish between ordinary civilian drones and FPV drones, and struggle to capture the weak radio frequency signals transmitted by FPV drones in high-definition image transmission, resulting in missed detections and false detections, and failing to achieve accurate detection and target tracing. Summary of the Invention

[0003] This invention provides an SDR-based FPV drone precision detection system and method to solve the problems of poor frequency band adaptability, weak anti-interference capability, low FPV drone identification accuracy, and high deployment cost of existing detection equipment.

[0004] This invention proposes a precise detection method for FPV drones based on SDR (Search and Detection), the method comprising:

[0005] S1. Perform full-domain radio frequency signal scanning and acquisition of the target airspace. Utilize the reconfigurable characteristics of the SDR radio frequency acquisition module to adaptively match the commonly used communication frequency bands for FPV UAV remote control and image transmission within the 20MHz to 6GHz frequency band. The signal processing module dynamically switches the sampling rate to simultaneously capture weak radio frequency signals in both frequency hopping and fixed frequency working modes, generating the original radio frequency signal dataset.

[0006] S2. The original radio frequency signal dataset is subjected to adaptive bandpass filtering by the signal filtering unit in the signal processing module to remove civilian spurious signal interference. The filtered signal is then enhanced by combining time-domain and frequency-domain dual noise reduction algorithms. The main control computing module extracts the time-frequency characteristic parameters of the FPV UAV pulse signal to generate preprocessed radio frequency characteristic data.

[0007] S3. The intelligent recognition algorithm unit in the main control computing module calls the radio frequency fingerprint feature library, compares the preprocessed radio frequency feature data with the feature parameters collected in the radio frequency fingerprint feature library using a lightweight deep learning algorithm, completes the signal category determination, distinguishes FPV drones from other civilian drones and wireless devices, and generates target recognition classification data.

[0008] S4. Using the trajectory calculation unit in the positioning calculation module, the FPV UAV signal confirmed in the target identification and classification data is analyzed by the time difference of arrival and signal strength joint positioning algorithm. Combined with the data fusion calculation of the multi-channel SDR radio frequency acquisition module, the real-time azimuth, distance and flight altitude of the UAV are calculated to generate target spatial position data.

[0009] S5. The target spatial location data is continuously accumulated over time through the early warning interaction module. The trajectory calculation unit dynamically draws the flight trajectory of the FPV UAV and generates trajectory tracking data. Abnormal behavior is judged from the trajectory tracking data. When it is determined to be an illegal FPV UAV, the early warning interaction module automatically triggers an audible and visual early warning signal and synchronously uploads the location, model, and signal frequency band data to the background management terminal to generate early warning management data.

[0010] This invention proposes a precision detection system for FPV unmanned aerial vehicles based on SDR (Self-Driving Detection and Recognition), the system comprising:

[0011] One or more processors;

[0012] Memory, used to store one or more programs;

[0013] Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0014] The beneficial effects of this invention are as follows: This invention utilizes SDR technology to achieve software-based reconfiguration of radio frequency functions, offering wide frequency band coverage and strong adaptability, enabling accurate identification of various FPV drones; through radio frequency fingerprint recognition technology, it significantly reduces the false detection rate, achieving a detection and identification accuracy rate of over 95%; it possesses excellent anti-interference capabilities, making it suitable for complex electromagnetic environments such as urban commercial districts, industrial parks, and airport perimeters; the equipment features high integration, low cost, and flexible deployment, requiring no complex debugging for rapid deployment; simultaneously, it possesses real-time positioning, trajectory recording, and data traceability functions, providing data support for low-altitude security enforcement and the management of illegal drones, effectively preventing security risks such as unauthorized flights, surreptitious filming, and intrusion, and possesses extremely high engineering application and market promotion value. Attached Figure Description

[0015] Figure 1 This is a diagram illustrating the steps of the method described in this invention;

[0016] Figure 2 This is a detailed flowchart of step S4 in this invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] One embodiment of the present invention, such as Figure 1 As shown, a precise detection method for FPV drones based on SDR is described, the method comprising:

[0019] S1. The SDR radio frequency acquisition module performs full-domain radio frequency signal scanning and acquisition of the target airspace through the spectrum scanning unit. Utilizing the reconfigurable characteristics of the SDR radio frequency acquisition module, it adaptively matches the commonly used communication frequency bands for FPV UAV remote control and image transmission within the 20MHz to 6GHz frequency band. The signal processing module dynamically switches the sampling rate to simultaneously capture weak radio frequency signals in both frequency hopping and fixed frequency working modes, generating the original radio frequency signal dataset.

[0020] S2. The original radio frequency signal dataset is subjected to adaptive bandpass filtering by the signal filtering unit in the signal processing module to remove interference from civilian spurious signals such as Bluetooth and WiFi. The filtered signal is then enhanced by combining time-domain and frequency-domain dual noise reduction algorithms. The main control computing module extracts the time-frequency characteristic parameters of the FPV UAV pulse signal to generate preprocessed radio frequency characteristic data.

[0021] S3. The intelligent recognition algorithm unit in the main control computing module calls the radio frequency fingerprint feature library, compares the pre-processed radio frequency feature data with the modulation mode, signal bandwidth, pulse interval and spectrum feature parameters included in the radio frequency fingerprint feature library using a lightweight deep learning algorithm, completes the signal category determination, distinguishes FPV drones from other civilian drones and wireless devices, and generates target recognition classification data.

[0022] S4. Using the trajectory calculation unit in the positioning calculation module, the FPV UAV signal confirmed in the target identification and classification data is analyzed by the time difference of arrival and signal strength joint positioning algorithm. Combined with the data fusion calculation of the multi-channel SDR radio frequency acquisition module, the real-time azimuth, distance and flight altitude of the UAV are calculated to generate target spatial position data.

[0023] S5. The target spatial location data is continuously accumulated over time through the early warning interaction module. The trajectory calculation unit dynamically draws the flight trajectory of the FPV UAV and generates trajectory tracking data. Abnormal behavior is judged from the trajectory tracking data. When it is determined to be an illegal FPV UAV, the early warning interaction module automatically triggers an audible and visual early warning signal and synchronously uploads the location, model, and signal frequency band data to the background management terminal to generate early warning management data.

[0024] The working principle and effects of the above technical solution are as follows: Through full-band adaptive radio frequency acquisition, the ability to capture weak signals from FPV drones is significantly improved, reducing the probability of missed detection caused by the fixed frequency bands of traditional equipment. Through layered filtering and dual noise reduction processing, the signal purity in complex electromagnetic environments is effectively enhanced, reducing misjudgments caused by civilian clutter interference. Through comparison of radio frequency fingerprints and deep learning, the identification accuracy of FPV drones is improved, avoiding misidentification caused by confusion between ordinary civilian equipment and FPV drones. Through joint positioning and multi-channel data fusion, the positioning accuracy of low-altitude small drones is improved, reducing positioning deviations caused by obstruction and multipath effects. This solution enables rapid detection and real-time trajectory tracking, while also allowing for low-cost and flexible deployment, avoiding the high cost and limited deployment of large radar equipment, and ensuring stable and reliable low-altitude security control in key areas.

[0025] In one embodiment of the present invention, S1 includes:

[0026] The S11 and SDR radio frequency acquisition modules start the full-domain detection working mode. The spectrum scanning unit continuously traverses the designated target airspace to complete the full-coverage scanning operation of radio frequency signals in all dimensions within the airspace, and continuously collects waveform and amplitude information of various wireless signals in the airspace to form the original signal sampling stream in the airspace.

[0027] The S12 and SDR radio frequency acquisition modules call their own reconfigurable operation mechanism to complete the frequency band adaptation and matching segment by segment in the overall frequency band range of 20MHz to 6GHz, accurately lock the mainstream communication frequency bands corresponding to the FPV drone remote control link and image transmission link, and complete the target frequency band directional locking.

[0028] S13. The signal processing module dynamically adjusts the sampling rate of the whole machine according to the frequency band matching results. It adapts the corresponding sampling rules for the common frequency hopping communication state and fixed frequency communication state of UAVs, and simultaneously captures the weaker covert radio frequency signals in the two working modes to complete the synchronous acquisition of multi-mode signals.

[0029] S14. The radio frequency acquisition link integrates the original information of various signals captured throughout the process, unifies the data format standardization and timing sorting, integrates the scattered single-channel sampling information into a complete data set, and summarizes it to form a full-domain original radio frequency signal dataset.

[0030] S15. The acquisition module monitors changes in airspace signal strength in real time, fine-tunes the scanning step and sampling duration according to the strength of environmental electromagnetic interference, performs preliminary timing calibration on the original radio frequency signal dataset, and ensures that the signal timing within the dataset is synchronized with the actual signal transmission status in the airspace.

[0031] The working principle and effects of the above technical solution are as follows: By using full-domain RF scanning and reconfigurable frequency band adaptation, the integrity of target airspace signal coverage is improved, reducing signal loss issues caused by fixed frequency bands in traditional equipment. Dynamic sampling rate adjustment and dual-mode signal capture enhance the ability to capture weak and concealed signals, reducing missed captures of frequency-hopping or fixed-frequency signals. Multi-channel signal integration and timing regularization improve the regularity and consistency of raw data, avoiding data chaos caused by scattered sampling. Real-time interference monitoring and parameter fine-tuning enhance the stability of data acquisition in complex electromagnetic environments, reducing data timing deviations caused by environmental interference. It can quickly adapt to various FPV UAV communication frequency bands and flexibly cope with changing electromagnetic environments, effectively ensuring the comprehensiveness, stability, and accuracy of raw RF data acquisition.

[0032] In one embodiment of the present invention, step S12 includes:

[0033] The original spatial signal sampling stream is processed frame by frame to analyze the spectrum features, extract multi-dimensional feature information such as signal frequency distribution, amplitude fluctuation law and time series change characteristics, remove invalid feature components corresponding to environmental noise, and generate refined spectrum feature analysis data.

[0034] The refined spectrum feature analysis data is subjected to full-band partition comparison processing, and the standard feature parameters in the 20MHz to 6GHz spectrum range are traversed to identify the unique feature attributes of the FPV UAV remote control and image transmission communication frequency bands, and generate frequency band feature identification data.

[0035] The frequency band feature identification data is processed for reconfigurable frequency band adaptation. The hardware reconfiguration capability of the SDR radio frequency acquisition module is used to complete the segmented frequency band calibration and adaptation. The data is accurately matched to the characteristics of the UAV communication frequency band and generates frequency band adaptation calibration data.

[0036] Redundant frequency bands are filtered out from the frequency band adaptation calibration data, eliminating frequency bands occupied by civilian conventional communication and blank frequency bands without effective characteristics, retaining the core frequency bands dedicated to UAV communication, and generating effective frequency band screening data;

[0037] The effective frequency band screening data is subjected to frequency band boundary fixing and locking processing to correct frequency band offset errors, fix the target frequency band range of the two types of communication links of FPV UAV, complete the target frequency band directional locking, and generate frequency band locking result data.

[0038] The working principle and effects of the above technical solution are as follows: By performing frame-by-frame spectrum analysis and multi-dimensional feature extraction, the precision of original signal feature recognition is improved, reducing feature ambiguity caused by background noise interference. Through full-band partition comparison and unique feature identification, the accuracy of FPV drone frequency band recognition is enhanced, reducing misjudgments caused by civilian frequency bands. Through segmented calibration and adaptation and redundant frequency band filtering, the targeting of frequency band matching is improved, reducing resource waste caused by invalid frequency band occupation. Through boundary solidification and error correction, the stability of frequency band locking is enhanced, avoiding target signal loss due to frequency band offset. It can quickly lock onto the drone's dedicated communication frequency band and accurately filter irrelevant signals, effectively improving the reliability and adaptation efficiency of frequency band recognition in complex electromagnetic environments.

[0039] In one embodiment of the present invention, S13 includes:

[0040] The signal processing module receives the output frequency band locking result data, combines it with the locked target communication frequency band parameters to generate adaptive sampling reference parameters, performs multi-dimensional parameter calibration processing on the sampling reference parameters, and generates dynamic sampling configuration parameter data.

[0041] The sampling rate of the dynamic sampling configuration parameter data is graded and controlled. Fixed sampling rate parameters are configured for the fixed frequency communication mode of FPV UAV, and dynamic adaptive sampling rate parameters are configured for the frequency hopping communication mode, generating dual-mode sampling rate configuration data.

[0042] The dual-mode sampling rate configuration data is encapsulated and adapted to match the sampling rules, and the signal sampling timing, sampling interval and signal capture threshold of the two communication modes are matched respectively to generate differentiated sampling execution rule data.

[0043] According to the differentiated sampling execution rules, multiple signal acquisition channels are opened synchronously to continuously collect and capture low-amplitude, concealed UAV radio frequency signals in the airspace, complete the synchronous acquisition of fixed-frequency and frequency-hopping signals, and generate multi-mode synchronous signal acquisition data.

[0044] The multi-mode synchronous signal acquisition data is processed by channel fusion and normalization to unify the signal output format of multiple acquisition channels, remove incomplete signal fragments generated during the acquisition process, and generate complete multi-mode synchronous acquisition result data.

[0045] The working principle and effects of the above technical solution are as follows: By calibrating the sampling reference parameters, the matching degree between the sampling parameters and the target frequency band is improved, reducing signal acquisition distortion caused by parameter deviation. By hierarchically adjusting the sampling rate, the adaptability of frequency hopping and fixed-frequency signals is enhanced, reducing signal loss caused by a single rate. By encapsulating differentiated sampling rules, the targeting of signal capture under two communication modes is improved, avoiding missed captures due to rule mismatch. By synchronously capturing multiple channels, the ability to capture low-amplitude concealed signals is enhanced, reducing the missed detection of weak signals. By channel fusion and regularization, the integrity and consistency of the acquired data are improved, avoiding subsequent processing errors caused by the disorder of multiple data streams. It can adapt to both communication modes of UAVs and stably capture weak signals, effectively improving the comprehensiveness and reliability of signal acquisition.

[0046] In one embodiment of the present invention, S2 includes:

[0047] S21. The signal filtering unit built into the signal processing module calls the adaptive bandpass filtering logic to perform hierarchical filtering operations on the raw radio frequency signal dataset produced in the previous stage, gradually filtering out spurious interference signals generated by conventional civilian wireless devices such as Bluetooth and WiFi in the airspace, and obtaining a basic filtered signal with improved purity.

[0048] S22. After the filtering unit completes interference removal, the signal processing module starts a dual noise reduction operation process from the time domain and frequency domain dimensions respectively. It suppresses random noise and environmental noise in the signal, amplifies the energy ratio of the effective signal of the UAV, and completes the overall signal feature enhancement.

[0049] S23. The main control computing module receives the signal data after filtering and noise reduction, analyzes the signal waveform change pattern frame by frame, continuously extracts the unique time domain parameters and frequency domain parameters of the FPV UAV pulse signal, and summarizes various feature information to form a discrete feature parameter group.

[0050] S24. The main control computing module performs correlation integration and dimension normalization processing on the discrete feature parameter group, unifies the parameter value range and data format, removes invalid data fragments with incomplete features, integrates all valid feature information, and generates standardized preprocessed radio frequency feature data.

[0051] S25. The module performs feature data stability detection, compares the fluctuation range of feature parameters over a continuous period, and performs secondary noise reduction correction on data segments with fluctuations exceeding the normal range to maintain the overall stability of preprocessed radio frequency feature data.

[0052] The working principle and effects of the above technical solution are as follows: Layered filtering removes civilian stray signals, improving the purity of the original radio frequency signal and reducing recognition interference from irrelevant signals. Dual noise reduction in the time and frequency domains enhances the effective signal strength of the UAV, reducing feature weakening caused by environmental noise. Frame-by-frame waveform analysis and extraction of specific parameters improve the accuracy of pulse signal feature extraction and reduce the inclusion of invalid features. Parameter normalization and incomplete data removal improve the regularity of feature data, avoiding data format chaos that affects subsequent comparisons. Stability detection and secondary correction enhance the reliability of preprocessed data, reducing recognition deviations caused by parameter fluctuations. This solution can deeply purify signals in complex electromagnetic environments and stably output high-quality feature data, effectively ensuring the accuracy of subsequent recognition processes.

[0053] In one embodiment of the present invention, step S3 includes:

[0054] S31. The intelligent identification algorithm unit in the main control computing module retrieves the locally stored radio frequency fingerprint feature library, fully loads the feature information of various wireless devices included in the library, and completes the local call and cache loading of feature library data.

[0055] S32. The intelligent identification algorithm unit reads the preprocessed radio frequency feature data, extracts the core feature parameters contained in the data such as modulation mode signal, bandwidth, and pulse interval spectrum, and splits out multiple sets of independent feature vectors that can be used for comparison.

[0056] S33. The algorithm unit runs a lightweight deep learning operation process, performing similarity matching operations between each of the split feature vectors and the standard parameters in the radio frequency fingerprint feature library, and matching each of the inherent signal features of various devices one by one.

[0057] S34. Combine the multi-dimensional similarity calculation results to complete the signal attribute distinction, and classify FPV drone signals, ordinary civilian drone signals and other types of civilian wireless device signals to complete the classification of all signals.

[0058] S35. The algorithm unit summarizes all signal category classification results, completes data classification and arrangement according to signal source and characteristic attributes, and organizes it into a clear target recognition classification data.

[0059] The working principle and effects of the above technical solution are as follows: By loading the feature library through local caching, the efficiency of feature data retrieval is improved, reducing recognition lag caused by real-time reading latency. By extracting multi-dimensional core parameters and splitting feature vectors, the comprehensiveness of feature comparison is improved, reducing misjudgments caused by single-dimensional comparison. Lightweight deep learning matching enhances the ability to identify complex features, reducing recognition errors caused by traditional comparison methods. Multi-dimensional similarity results distinguish signal attributes, improving the accuracy of FPV drone recognition and reducing confusion caused by signals from civilian equipment. Classifying and arranging recognition results improves the regularity of data output, avoiding classification confusion that affects subsequent positioning processing. It can quickly complete signal type determination and stably distinguish FPV drones from other devices, effectively improving recognition accuracy and overall processing efficiency.

[0060] In one embodiment of the present invention, S33 includes:

[0061] The algorithm unit reads each set of independent feature vectors, performs normalization correction on each feature parameter inside the vector, unifies the parameter value range and data precision, and generates standardized feature vector data.

[0062] Retrieve standard feature parameter sets of various devices stored in the radio frequency fingerprint feature database, compare the standardized feature vector data with the standard feature parameter sets dimension by dimension and item by item, and generate the initial feature comparison result data.

[0063] Initiate a lightweight deep learning computation process, perform similarity measurement calculations on the initial feature alignment results, statistically analyze the matching ratio and deviation values ​​of different feature dimensions, and generate multi-dimensional similarity measurement data.

[0064] The multi-dimensional similarity measurement data is weighted and fused to obtain the overall matching degree value by combining the matching weights of various features, and a global matching degree statistical data is generated.

[0065] Based on global matching degree statistics, the inherent signal characteristics of the equipment are benchmarked and verified, the matching objects and matching status corresponding to each set of feature vectors are recorded, and complete feature benchmarking result data is generated.

[0066] The working principle and effects of the above technical solution are as follows: By normalizing and correcting feature parameters, the comparability between feature vectors is improved, reducing comparison bias caused by differences in numerical ranges. Through dimension-by-dimensional and item-by-item comparison, the detail of feature matching is enhanced, reducing recognition errors caused by local feature mismatches. Through similarity quantification calculation, the measurability of feature matching results is improved, reducing result bias caused by subjective judgment. Through weighted fusion operations, the reliability of the overall matching result is improved, avoiding misjudgments caused by imbalances in the weight of a single dimension. By verifying and recording the matching status, the traceability of the recognition results is improved, reducing unreliable results caused by chaotic feature matching. This approach not only refines the feature comparison process but also accurately quantifies the degree of matching, effectively improving the accuracy and stability of signal feature recognition.

[0067] One embodiment of the present invention, such as Figure 2 As shown, S4 includes:

[0068] S41. The trajectory calculation unit in the positioning calculation module reads the target identification and classification data, filters out the valid signal data that is determined to be an FPV UAV, and separately divides the target signal data group to be located and calculated.

[0069] S42. The trajectory calculation unit starts the arrival time difference calculation logic and signal strength analysis logic, integrates the two types of positioning algorithms, and performs basic positioning parameter deduction based on signal propagation characteristics to obtain basic signal propagation calculation values.

[0070] S43. The multi-channel SDR radio frequency acquisition module summarizes the same source signal data returned from each acquisition channel, performs multi-source data fusion calculation, makes up for the deviation problem of single-channel data, and corrects the calculation error of basic positioning parameters.

[0071] S44. The fused and corrected parameters enter the comprehensive calculation process, and the straight-line distance of the target UAV's current azimuth and real-time flight altitude are calculated in sequence to obtain the UAV's three-dimensional spatial position information at a single moment.

[0072] S45. The positioning and calculation module continuously collects spatial location information from multiple time points, uniformly completes data encapsulation and formatting, integrates all location information, and generates continuous and complete target spatial location data.

[0073] The working principle and effects of the above technical solution are as follows: By filtering effective signals and dividing target data groups, the targeting of positioning calculations is improved, and computational redundancy caused by invalid signals is reduced. By fusing basic parameters through dual algorithms, the adaptability of the positioning logic is enhanced, reducing positioning inaccuracies caused by single algorithm defects. By fusing multi-channel data to correct errors, the accuracy of basic parameters is improved, and result deviations caused by single-channel biases are reduced. By comprehensively calculating three-dimensional parameters, the completeness of spatial position calculation is improved, reducing information loss caused by two-dimensional positioning. By integrating and encapsulating multi-node data, the continuity of position data is improved, avoiding trajectory breaks caused by single-point errors. It can quickly lock onto targets and output three-dimensional positions, and stably output continuous position data, effectively improving the positioning accuracy and reliability of low-altitude small UAVs.

[0074] In one embodiment of the present invention, S42 includes:

[0075] The trajectory calculation unit receives the target signal data group to be located and calculated, extracts multiple original propagation parameters such as signal transmission delay, signal field strength attenuation, and channel transmission loss, and generates original signal propagation parameter data.

[0076] Temporal noise reduction preprocessing is performed on the raw signal propagation parameter data to remove abnormal parameter values ​​that change instantaneously, smooth the overall fluctuation amplitude of the parameters, and generate stable propagation parameter data;

[0077] The trajectory calculation unit activates the arrival time difference calculation logic, uses the signal multi-path transmission delay difference to perform distance difference extrapolation calculation, and generates time delay positioning basic parameter data;

[0078] The trajectory calculation unit synchronously activates the signal strength analysis logic, and converts the received signal strength value by combining the spatial wireless signal attenuation law to generate the basic parameter data for field strength positioning.

[0079] The time-delay positioning basic parameter data and the field strength positioning basic parameter data are fused in parallel to complete the complementary superposition of the positioning parameters of the two algorithms, and the complete set of basic calculation value data corresponding to signal propagation is derived.

[0080] The working principle and effects of the above technical solution are as follows: By extracting multiple types of original propagation parameters, the comprehensiveness of the positioning data is improved, reducing insufficient positioning basis caused by single parameters. Temporal denoising preprocessing enhances the stability of the parameter sequence, reducing computational fluctuations caused by instantaneous abnormal values. Distance difference is extrapolated from time delay differences, improving the accuracy of positioning in the time dimension and reducing errors caused by signal transmission fluctuations. Intensity values ​​are converted from field strength attenuation laws, enhancing spatial attenuation adaptability and reducing intensity deviations caused by environmental occlusion. Parallel fusion of dual-algorithm parameters improves the reliability of basic computational values, avoiding inaccurate positioning caused by the defects of a single algorithm. It can support positioning from both time and field strength dimensions, and stably output reliable basic parameters, effectively improving the stability and accuracy of positioning operations in complex electromagnetic environments.

[0081] In one embodiment of the present invention, step S5 includes:

[0082] S51. The early warning interaction module continuously receives target spatial location data and performs continuous stacking and time-series accumulation of data in chronological order to form a location time-series data chain covering different time periods.

[0083] S52. The trajectory calculation unit reads the time-series position data chain, connects the spatial position information point by point according to the coordinate change law, dynamically outlines the motion route formed by the continuous flight of the FPV UAV, and generates the initial flight trajectory information.

[0084] S53. The trajectory calculation unit smooths and completes the initial flight trajectory information, eliminates trajectory jitter caused by single-point positioning error, and normalizes all trajectory information to generate standard trajectory tracking data.

[0085] S54. The early warning and interaction module retrieves the abnormal behavior judgment rules, analyzes the flight speed, flight area, motion attitude and other contents in the trajectory tracking data segment by segment, determines whether the target drone has illegal flight behavior, and marks the illegal target object.

[0086] The S55 early warning interaction module immediately activates the audible and visual warning device to output a warning signal for FPV drones that are determined to be in violation. At the same time, it extracts key information such as target location, equipment model, and signal frequency band, and transmits it to the background management terminal. It integrates all the information to generate the final early warning and management data.

[0087] The working principle and effects of the above technical solution are as follows: By stacking and accumulating location data in a time sequence, the continuity of location information is improved, reducing trajectory loss caused by scattered points. By connecting coordinates point by point to delineate the movement route, the intuitiveness of trajectory reconstruction is enhanced, avoiding route ambiguity caused by discrete locations. Through smoothing and trajectory completion, trajectory jitter caused by single-point errors is reduced, improving the regularity of trajectory data. By analyzing multi-dimensional parameters to determine abnormal behavior, the accuracy of violation identification is improved, reducing false positives and false negatives. Through real-time audible and visual warnings and synchronized information uploads, the timeliness of violation handling is improved, avoiding delays in control. It can completely reconstruct the drone's flight trajectory and quickly identify violations and trigger warnings, effectively ensuring the timeliness and effectiveness of low-altitude security control in key areas.

[0088] In one embodiment of the present invention, S51 includes:

[0089] The early warning interaction module continuously receives target spatial location data transmitted from multiple sources, caches the point coordinates and spatial parameter information output at different acquisition times, and forms a real-time location cache data stream;

[0090] The real-time location cache data stream is processed for time-series discrimination, and invalid data points with disordered time sequence and repeated superposition are removed, while valid location data with orderly time sequence is retained to generate time-series filtered location data.

[0091] The time-series filtered location data is continuously stacked, and the data is arranged sequentially according to the time nodes of data generation. A time-series arrangement framework for location data is built, and time-series stacked location data is generated.

[0092] Multi-dimensional parameter accumulation and integration processing is performed on time-series stacked location data to unify and standardize the parameter dimensions of single-point location data, supplement the missing micro-location information in adjacent time periods, and generate time-series accumulated and integrated data.

[0093] The time-series accumulated and integrated data is processed by link splicing to form a complete location time-series data chain covering the entire UAV flight process, connecting all effective location data points throughout the time period.

[0094] The working principle and effects of the above technical solution are as follows: By forming a real-time data stream through multi-channel data caching, the integrity of location data reception is improved, reducing the loss of data points caused by data dropouts. Invalid data points are eliminated through time-series screening, enhancing the orderliness of data timing and reducing time-series chaos caused by erroneous data. A layout framework is built through continuous stacking, improving the regularity of data arrangement and avoiding trajectory breaks caused by disordered arrangement. Minor information is supplemented through parameter accumulation and integration, enhancing the coherence of location data and reducing trajectory breaks caused by data loss. A complete data chain is formed through link splicing, improving the comprehensiveness of coverage during flight and avoiding incomplete trajectories caused by scattered data points. This ensures both correct timing and complete information of location data, and forms a continuous and reliable data chain, effectively improving the accuracy of trajectory reconstruction and the reliability of subsequent analysis.

[0095] One embodiment of the present invention provides a precision detection system for FPV unmanned aerial vehicles based on SDR, the system comprising:

[0096] One or more processors;

[0097] Memory, used to store one or more programs;

[0098] Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0099] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A precise detection method for FPV unmanned aerial vehicles based on SDR, characterized in that, The method includes: S1. Perform full-domain radio frequency signal scanning and acquisition of the target airspace. Utilize the reconfigurable characteristics of the SDR radio frequency acquisition module to adaptively match the commonly used communication frequency bands for FPV UAV remote control and image transmission within the 20MHz to 6GHz frequency band. The signal processing module dynamically switches the sampling rate to simultaneously capture weak radio frequency signals in both frequency hopping and fixed frequency working modes, generating the original radio frequency signal dataset. S2. The original radio frequency signal dataset is subjected to adaptive bandpass filtering by the signal filtering unit in the signal processing module to remove civilian spurious signal interference. The filtered signal is then enhanced by combining time-domain and frequency-domain dual noise reduction algorithms. The main control computing module extracts the time-frequency characteristic parameters of the FPV UAV pulse signal to generate preprocessed radio frequency characteristic data. S3. The intelligent recognition algorithm unit in the main control computing module calls the radio frequency fingerprint feature library, compares the preprocessed radio frequency feature data with the feature parameters collected in the radio frequency fingerprint feature library using a lightweight deep learning algorithm, completes the signal category determination, distinguishes FPV drones from other civilian drones and wireless devices, and generates target recognition classification data. S4. Using the trajectory calculation unit in the positioning calculation module, the FPV UAV signal confirmed in the target identification and classification data is analyzed by the time difference of arrival and signal strength joint positioning algorithm. Combined with the data fusion calculation of the multi-channel SDR radio frequency acquisition module, the real-time azimuth, distance and flight altitude of the UAV are calculated to generate target spatial position data. S5. The target spatial location data is continuously accumulated over time through the early warning interaction module. The trajectory calculation unit dynamically draws the flight trajectory of the FPV UAV and generates trajectory tracking data. Abnormal behavior is judged from the trajectory tracking data. When it is determined to be an illegal FPV UAV, the early warning interaction module automatically triggers an audible and visual early warning signal and synchronously uploads the location, model, and signal frequency band data to the background management terminal to generate early warning management data.

2. The method for precise detection of FPV UAVs based on SDR according to claim 1, characterized in that, S1 includes: The S11 and SDR radio frequency acquisition modules start the full-domain detection working mode. The spectrum scanning unit continuously traverses the designated target airspace to complete the full-coverage scanning operation of radio frequency signals in all dimensions within the airspace, and continuously collects waveform and amplitude information of various wireless signals in the airspace to form the original signal sampling stream in the airspace. The S12 and SDR radio frequency acquisition modules call their own reconfigurable operation mechanism to complete the frequency band adaptation and matching segment by segment in the overall frequency band range of 20MHz to 6GHz, accurately lock the mainstream communication frequency bands corresponding to the FPV drone remote control link and image transmission link, and complete the target frequency band directional locking. S13. The signal processing module dynamically adjusts the sampling rate of the whole machine according to the frequency band matching results. It adapts the corresponding sampling rules for the common frequency hopping communication state and fixed frequency communication state of UAVs, and simultaneously captures the weaker covert radio frequency signals in the two working modes to complete the synchronous acquisition of multi-mode signals. S14. The radio frequency acquisition link integrates the original information of various signals captured throughout the process, unifies the data format standardization and timing sorting, integrates the scattered single-channel sampling information into a complete data set, and summarizes it to form a full-domain original radio frequency signal dataset. S15. The acquisition module monitors changes in airspace signal strength in real time, fine-tunes the scanning step and sampling duration according to the strength of environmental electromagnetic interference, performs preliminary timing calibration on the original radio frequency signal dataset, and ensures that the signal timing within the dataset is synchronized with the actual signal transmission status in the airspace.

3. The method for precise detection of FPV UAVs based on SDR according to claim 2, characterized in that, S12 includes: The original spatial signal sampling stream is processed frame by frame to analyze the spectrum features, extract multi-dimensional feature information such as signal frequency distribution, amplitude fluctuation law and time series change characteristics, remove invalid feature components corresponding to environmental noise, and generate refined spectrum feature analysis data. The refined spectrum feature analysis data is subjected to full-band partition comparison processing, and the standard feature parameters in the 20MHz to 6GHz spectrum range are traversed to identify the unique feature attributes of the FPV UAV remote control and image transmission communication frequency bands, and generate frequency band feature identification data. The frequency band feature identification data is processed for reconfigurable frequency band adaptation. The hardware reconfiguration capability of the SDR radio frequency acquisition module is used to complete the segmented frequency band calibration and adaptation. The data is accurately matched to the characteristics of the UAV communication frequency band and generates frequency band adaptation calibration data. Redundant frequency bands are filtered out from the frequency band adaptation calibration data, eliminating frequency bands occupied by civilian conventional communication and blank frequency bands without effective characteristics, retaining the core frequency bands dedicated to UAV communication, and generating effective frequency band screening data; The effective frequency band screening data is subjected to frequency band boundary fixing and locking processing to correct frequency band offset errors, fix the target frequency band range of the two types of communication links of FPV UAV, complete the target frequency band directional locking, and generate frequency band locking result data.

4. The method for precise detection of FPV UAVs based on SDR according to claim 2, characterized in that, S13 includes: The signal processing module receives the output frequency band locking result data, combines it with the locked target communication frequency band parameters to generate adaptive sampling reference parameters, performs multi-dimensional parameter calibration processing on the sampling reference parameters, and generates dynamic sampling configuration parameter data. The sampling rate of the dynamic sampling configuration parameter data is graded and controlled. Fixed sampling rate parameters are configured for the fixed frequency communication mode of FPV UAV, and dynamic adaptive sampling rate parameters are configured for the frequency hopping communication mode, generating dual-mode sampling rate configuration data. The dual-mode sampling rate configuration data is encapsulated and adapted to match the sampling rules, and the signal sampling timing, sampling interval and signal capture threshold of the two communication modes are matched respectively to generate differentiated sampling execution rule data. According to the differentiated sampling execution rules, multiple signal acquisition channels are opened synchronously to continuously collect and capture low-amplitude, concealed UAV radio frequency signals in the airspace, complete the synchronous acquisition of fixed-frequency and frequency-hopping signals, and generate multi-mode synchronous signal acquisition data. The multi-mode synchronous signal acquisition data is processed by channel fusion and normalization to unify the signal output format of multiple acquisition channels, remove incomplete signal fragments generated during the acquisition process, and generate complete multi-mode synchronous acquisition result data.

5. The method for precise detection of FPV unmanned aerial vehicles based on SDR according to claim 1, characterized in that, S2 includes: S21. The signal filtering unit built into the signal processing module calls the adaptive bandpass filtering logic to perform hierarchical filtering operations on the raw radio frequency signal dataset produced in the previous stage, gradually filtering out spurious interference signals generated by conventional civilian wireless devices in the airspace, and obtaining a basic filtered signal with improved purity. S22. After the filtering unit completes interference removal, the signal processing module starts a dual noise reduction operation process from the time domain and frequency domain dimensions respectively. It suppresses random noise and environmental noise in the signal, amplifies the energy ratio of the effective signal of the UAV, and completes the overall signal feature enhancement. S23. The main control computing module receives the signal data after filtering and noise reduction, analyzes the signal waveform change pattern frame by frame, continuously extracts the unique time domain parameters and frequency domain parameters of the FPV UAV pulse signal, and summarizes various feature information to form a discrete feature parameter group. S24. The main control computing module performs correlation integration and dimension normalization processing on the discrete feature parameter group, unifies the parameter value range and data format, removes invalid data fragments with incomplete features, integrates all valid feature information, and generates standardized preprocessed radio frequency feature data. S25. The module performs feature data stability detection, compares the fluctuation range of feature parameters over a continuous period, and performs secondary noise reduction correction on data segments with fluctuations exceeding the normal range to maintain the overall stability of preprocessed radio frequency feature data.

6. The method for precise detection of FPV UAVs based on SDR according to claim 1, characterized in that, The S3 includes: S31. The intelligent identification algorithm unit in the main control computing module retrieves the locally stored radio frequency fingerprint feature library, fully loads the feature information of various wireless devices included in the library, and completes the local call and cache loading of feature library data. S32. The intelligent identification algorithm unit reads the preprocessed radio frequency feature data, extracts the core feature parameters contained in the data in sequence, and splits out multiple sets of independent feature vectors that can be used for comparison. S33. The algorithm unit runs a lightweight deep learning operation process, performing similarity matching operations between each of the split feature vectors and the standard parameters in the radio frequency fingerprint feature library, and matching each one against the inherent signal features of various devices. S34. Combine the multi-dimensional similarity calculation results to complete the signal attribute differentiation, and classify FPV drone signals, ordinary civilian drone signals and other types of civilian wireless device signals to complete the classification of all signals. S35. The algorithm unit summarizes all signal category classification results, completes data classification and arrangement according to signal source and characteristic attributes, and organizes it into a clear target recognition classification data.

7. The method for precise detection of FPV UAVs based on SDR according to claim 1, characterized in that, The S4 includes: S41. The trajectory calculation unit in the positioning calculation module reads the target identification and classification data, filters out the valid signal data that is determined to be an FPV UAV, and separately divides the target signal data group to be located and calculated. S42. The trajectory calculation unit starts the arrival time difference calculation logic and signal strength analysis logic, integrates the two types of positioning algorithms, and performs basic positioning parameter deduction based on signal propagation characteristics to obtain basic signal propagation calculation values. S43. The multi-channel SDR radio frequency acquisition module summarizes the same source signal data returned from each acquisition channel, performs multi-source data fusion calculation, and corrects the calculation error of the basic positioning parameters. S44. The fused and corrected parameters enter the comprehensive calculation process, and the straight-line distance of the target UAV's current azimuth and real-time flight altitude are calculated in sequence to obtain the UAV's three-dimensional spatial position information at a single moment. S45. The positioning and calculation module continuously collects spatial location information from multiple time points, uniformly completes data encapsulation and formatting, integrates all location information, and generates continuous and complete target spatial location data.

8. The method for precise detection of FPV UAVs based on SDR according to claim 7, characterized in that, S42 includes: The trajectory calculation unit receives the target signal data group to be located and calculated, extracts multiple original propagation parameters such as signal transmission delay, signal field strength attenuation, and channel transmission loss, and generates original signal propagation parameter data. Temporal noise reduction preprocessing is performed on the raw signal propagation parameter data to remove abnormal parameter values ​​that change instantaneously, smooth the overall fluctuation amplitude of the parameters, and generate stable propagation parameter data; The trajectory calculation unit activates the arrival time difference calculation logic, uses the signal multi-path transmission delay difference to perform distance difference extrapolation calculation, and generates time delay positioning basic parameter data; The trajectory calculation unit synchronously activates the signal strength analysis logic, and converts the received signal strength value by combining the spatial wireless signal attenuation law to generate the basic parameter data for field strength positioning. The time-delay positioning basic parameter data and the field strength positioning basic parameter data are fused in parallel to complete the complementary superposition of the positioning parameters of the two algorithms, and the complete set of basic calculation value data corresponding to signal propagation is derived.

9. The method for precise detection of FPV unmanned aerial vehicles based on SDR according to claim 1, characterized in that, The S5 includes: S51. The early warning interaction module continuously receives target spatial location data and performs continuous stacking and time-series accumulation of data in chronological order to form a location time-series data chain covering different time periods. S52. The trajectory calculation unit reads the time-series position data chain, connects the spatial position information point by point according to the coordinate change law, dynamically outlines the motion route formed by the continuous flight of the FPV UAV, and generates the initial flight trajectory information. S53. The trajectory calculation unit smooths and completes the initial flight trajectory information, eliminates trajectory jitter caused by single-point positioning error, and normalizes all trajectory information to generate standard trajectory tracking data. S54. The early warning and interaction module retrieves the abnormal behavior judgment rules, analyzes the content in the trajectory tracking data segment by segment, determines whether the target drone has violated flight behavior, and marks the violating target object. The S55 early warning interaction module immediately activates the audible and visual warning device to output a warning signal for FPV drones that are determined to be in violation. At the same time, it extracts key information and transmits it to the back-end management terminal, integrating all information to generate the final early warning and management data.

10. A precision detection system for FPV unmanned aerial vehicles based on SDR, characterized in that, The system includes: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 9.

Images