Unmanned aerial vehicle anti-decoy method and device based on cross validation and medium

Abstract

The invention provides an unmanned aerial vehicle anti-decoy method and device based on cross validation and a medium. The method comprises the steps that a local timestamp T1 is obtained based on a high-precision local crystal oscillator integrated in an unmanned aerial vehicle, standard time T2 provided by a GNSS navigation system is obtained, and the time deviation delta T between the local timestamp T1 and the standard time T2 is calculated; an autonomous positioning result P1 of the unmanned aerial vehicle is obtained through a visual SLAM module, a real-time positioning result P2 of a GNSS system is obtained at the same time, and the spatial deviation delta P of the positioning result P1 and the positioning result P2 is calculated; comparing the time deviation delta T with a preset time deviation threshold value Tthr, comparing the space deviation delta P with a preset space deviation threshold value Pthr, and judging whether the unmanned aerial vehicle is decoy or not according to a comparison result; and triggering a preset safety response mechanism according to the judgment result. According to the technical scheme provided by the invention, the navigation safety and the control reliability in a high-risk environment can be improved.

Description

TECHNICAL FIELD

[0001] The present document relates to the technical field of UAV anti-deception, and particularly relates to a cross-verification UAV anti-deception method, device and medium. BACKGROUND

[0002] At present, the navigation and positioning of a UAV highly depends on a GNSS system, which plays an important role in fields such as civilian aerial photography, logistics transportation, surveying and mapping, and the like. However, in special scenarios such as border control and military operations, GNSS signals are extremely vulnerable to attacks and are easily interfered with or deceived. Once the GNSS signals are tampered with, the UAV may deviate from the flight path, enter a forbidden area, or even be controlled by the enemy, which may result in serious consequences such as task failure, and may also cause major risks such as political crisis and military secret leakage. In order to deal with the safety problem of GNSS signals, there are various anti-deception means in the prior art, such as encryption processing of GNSS signals, use of frequency hopping communication technology to avoid interference, use of an AI anomaly recognition model to detect signal anomalies, and use of a geographic fence to limit the flight range. However, the prior art has the following problems: 1. Most anti-deception methods only detect GNSS position drift or trajectory anomalies, without introducing time consistency verification; 2. Although visual SLAM has independent positioning capability, if the UAV mapping environment is tampered with, it may still cause recognition errors; 3. Most navigation and positioning schemes rely on electromagnetic signal features for anomaly detection, but it is difficult to play a role in the face of position simulators and directional antennas, and lack a mechanism to resist false signals; Therefore, the prior art is difficult to build a comprehensive and highly reliable UAV anti-deception system, and there is an urgent need for a multi-dimensional verification method that integrates local time sources, GNSS space-time features, and visual SLAM autonomous positioning capabilities to improve the navigation safety and control reliability of the UAV in high-risk environments. SUMMARY

[0003] According to the embodiments of the present application, a cross-verification UAV anti-deception method, device and medium are provided, which solve the problem that the time dimension verification, visual SLAM and GNSS are not effectively integrated in the prior art anti-deception scheme.

[0004] According to the embodiments of the present application, a cross-verification UAV anti-deception method is provided, which comprises: obtaining a local timestamp T1 based on a high-precision local crystal oscillator integrated in the UAV, obtaining a standard time T2 provided by a GNSS navigation system, and calculating a time deviation ΔT of the local timestamp T1 and the standard time T2; acquire an autonomous positioning result P1 of the UAV through the visual SLAM module, acquire a real-time positioning result P2 of the GNSS system, and calculate a spatial deviation ΔP of the positioning results P1 and P2; compare the time deviation ΔT with a preset time deviation threshold T thr compare the spatial deviation ΔP with a preset spatial deviation threshold P thr compare, and determine whether the UAV is decoyed according to a comparison result; trigger a preset security response mechanism according to a determination result.

[0005] According to an embodiment of the present application, an electronic device is provided, comprising: a processor; and a memory arranged to store computer executable instructions that, when executed, cause the processor to perform the steps of the cross-verified UAV decoy prevention method as described above.

[0006] According to an embodiment of the present application, a storage medium is provided for storing computer executable instructions that, when executed, implement the steps of the cross-verified UAV decoy prevention method as described above.

[0007] By adopting the embodiment of the present application, a complete technical system of a fusion time and space two-dimensional judgment mechanism is proposed, which overcomes the limitations brought by the current single judgment mode which only relies on GNSS position abnormality detection or signal feature analysis. First, the present application constructs an independent time reference system by integrating a local high-precision crystal oscillator at the UAV end, which can verify the credibility of GNSS time in real time, fundamentally improving the recognition ability of time synchronization type decoy attack. Second, the visual SLAM module is introduced as a non-GNSS dependent autonomous positioning means, and through the coordinate comparison with the GNSS position, the spatial dimension abnormality detection is realized, which is especially suitable for the stable flight demand in complex border environment and enemy electromagnetic interference scene. In addition, the present application constructs a standardized cross-determination logic, allowing time abnormality and space abnormality as parallel input to trigger the decoy recognition mechanism. Finally, the system also designs a security response strategy based on the determination level, from early warning, flight restriction, return to self-destruction mechanism when necessary, which has a complete security side. Therefore, the present application is superior to the prior art in decoy recognition ability, security response strategy and other aspects, and has significant innovation and practical value. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the one or more embodiments of the present specification or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present specification, and other drawings can be obtained by those skilled in the art without creative labor.

[0009] Figure 1 Flow chart of the cross-validated UAV anti-decoying method of the embodiment of the present application; Figure 2 One specific implementation flow of the cross-validated UAV anti-decoying method of the embodiment of the present application. DETAILED DESCRIPTION

[0010] In order to make the person skilled in the art better understand the technical solutions in the one or more embodiments of the present specification, the technical solutions in the one or more embodiments of the present specification will be clearly and completely described in the following with reference to the drawings in the one or more embodiments of the present specification. Obviously, the described embodiments are only some embodiments of the present specification, not all. Based on the one or more embodiments of the present specification, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present document.

[0011] Method embodiment According to the embodiment of the present application, a cross-validated UAV anti-decoying method is provided, Figure 1 The flow chart of the cross-validated UAV anti-decoying method of the embodiment of the present application is according to Figure 1 As shown in the figure, the cross-validated UAV anti-decoying method of the embodiment of the present application specifically includes: S1, obtaining a local timestamp T1 based on a high-precision local crystal oscillator integrated inside the UAV, obtaining a standard time T2 provided by a GNSS navigation system, and calculating the time deviation AT of the local timestamp T1 and the standard time T2; As shown in the figure, the cross-validated UAV anti-decoying method of the embodiment of the present application specifically includes: Figure 2 As shown in the figure, one specific implementation flow of the embodiment of the present application is shown. The high-precision local crystal oscillator integrated inside the UAV is used as an independent time reference source to obtain the local timestamp T1 in real time, and compare it with the standard time T2 provided by the GNSS navigation system. The time deviation threshold of the two is set to T thr, if the difference AT between GNSS time and local crystal oscillator time exceeds the threshold value, it is judged that the GNSS time is abnormal. Such deviation may be caused by the slight error in time synchronization of the fake GNSS signal. Through this mechanism, even if the GNSS position information still remains reasonable on the surface, the potential deception behavior can be identified in time in the time dimension, which supplements the traditional method of relying only on position information to judge whether it is deceived. Using the local high-precision crystal oscillator time as an independent reference, compare it with the GNSS system time, when the deviation between the two exceeds the set threshold T thr , it is determined that the GNSS time is abnormal, which constitutes the time dimension identification ability of the deception signal.

[0012] S2, obtain the autonomous positioning result P1 of the unmanned aerial vehicle through the visual SLAM module, and obtain the real-time positioning result P2 of the GNSS system, and calculate the spatial deviation AP of the positioning results P1 and P2; The visual SLAM module constructs a local map and completes autonomous positioning through camera images and inertial measurement unit information, and does not rely on external electromagnetic signals; wherein a dynamic feature point screening algorithm is adopted in the local map construction process to eliminate suspected artificial interference feature points.

[0013] The positioning result P1 obtained by the unmanned aerial vehicle through the visual SLAM module is compared with the real-time positioning result P2 of the GNSS system, and the spatial coordinate error AP between the two is calculated to evaluate the position credibility. The visual SLAM constructs a local map and completes its own positioning through camera images and inertial navigation information, and does not rely on any external electromagnetic signal, and has strong anti-interference ability. The system presets a spatial deviation tolerance threshold P thr , when the deviation AP between GNSS positioning and SLAM positioning exceeds the threshold value, it is considered that the GNSS position may be tampered with or deceived.

[0014] The autonomous positioning result is obtained through the visual SLAM module, and the real-time coordinate error is calculated with the GNSS positioning, if the error exceeds the threshold P thr , it can be judged that the GNSS positioning is not reliable, and the anti-deception judgment is enhanced from the spatial dimension.

[0015] S3, compare the time deviation AT with the preset time deviation threshold T thr , compare the spatial deviation AP with the preset spatial deviation threshold P thr , compare the spatial deviation AP with the preset spatial deviation threshold P The judgment of whether the unmanned aerial vehicle is deceived specifically includes: If AT T thr , and AP P thr, determine as trusted navigation, otherwise determine as spoofed, wherein when ΔT ≥ T thr or ΔP ≥ P thr , further combine the magnitude of ΔT exceeding T thr , the magnitude of ΔP exceeding P thr , and the duration of both, to classify the spoofing level.

[0016] In one specific implementation, the specific level classification is as follows: Low-level spoofing: single-dimension anomaly, i.e. only ΔT exceeds or only ΔP exceeds, and the exceeding magnitude < 50% threshold, and the duration < 5 seconds; Medium-level spoofing: single-dimension anomaly and the exceeding magnitude ≥ 50% threshold, or the duration ≥ 5 seconds; High-level spoofing: double-dimension anomaly, i.e. both ΔT and ΔP exceed, or either dimension anomaly exceeding magnitude ≥ 100% threshold.

[0017] After determining as spoofed, the system returns the flight data and positioning results to the ground control center through an encrypted channel, and performs logging. The returned data includes ΔT, ΔP, GNSS raw observation values, SLAM key frame images, and local crystal oscillator timestamps at the time of anomaly, uses AES-256 encryption algorithm to ensure transmission safety, and the log file is named in the format of "anomaly type + timestamp + device ID" and stored in read-only memory (ROM).

[0018] Specifically, the UAV spoofing determination is based on Table 1: Table 1 Cross-validation logic table

[0019] The time deviation threshold T thr and the spatial deviation threshold P thr are adjusted according to the task type of the UAV and the flight airspace situation, and a dynamic threshold model constructed from historical flight data is introduced in the adjustment process, so that the threshold is self-adaptively corrected according to the complexity of the flight environment. The dynamic threshold model uses a random forest algorithm, the input features include flight height, terrain complexity, GNSS signal-to-noise ratio (SNR), and historical spoofing event occurrence rate, and the output is the real-time corrected T thr and P thr , for example, in a high-rise dense area, Pthr can be temporarily relaxed by 20% to reduce the false judgment caused by SLAM positioning error.

[0020] S3 builds a cross-validation mechanism running in parallel with time deviation and spatial deviation, respectively based on the difference ΔT between GNSS time and local crystal oscillator time, and the spatial difference ΔP between GNSS and visual SLAM positioning results. This mechanism avoids false judgments of single-dimension criteria, and improves the overall recognition accuracy and anti-interference robustness of the system.

[0021] S4, triggering a preset security response mechanism according to the determination result.

[0022] The security response mechanism includes at least one of the following: hovering, returning, limiting flight radius, locking control, power-off, and self-destruction, and the security response mechanism is adjusted according to the deception level. After recognizing the deception, the system will trigger the corresponding security response mechanism according to the preset strategy level. The specific response strategy is as follows: low-level deception: execute "hovering + warning", the UAV keeps hovering at the current position, sends an audible and visual alarm to the ground station through a 4G / 5G link, and starts SLAM map reconstruction to verify the environment authenticity; medium-level deception: execute "return + lock control", the UAV immediately switches to SLAM autonomous navigation mode, returns to the nearest safe take-off point along the original flight route, and prohibits receiving flight route modification instructions from the ground station during the return; high-level deception: execute "self-destruction + data backhaul", the self-destruction process includes: (1) cutting off the power system to make the UAV powerless gliding; (2) destroying the EEPROM chip storing sensitive data; (3) ejecting an emergency recovery capsule carrying a positioning beacon for subsequent debris recovery.

[0023] For medium and low-level deception, such as tolerable spatial deviation but abnormal time, the system will preferentially adopt conservative strategies such as returning, hovering in place, or limiting flight radius, and send warning information to the ground station in real time; for high-level deception, such as double abnormality or obvious rapid deviation trend, the self-destruction mechanism can be triggered, including disconnecting the power system, cutting off the power supply, and shedding key components, to ensure that the UAV is not induced to fly out of the safe area or captured by the enemy. According to the task type, flight airspace conditions, etc., the response level and execution action can be flexibly adjusted to maximize safety while improving system availability.

[0024] The method further includes: when comparing and determining the time deviation ΔT and the spatial deviation ΔP, the change rates of ΔT and ΔP are monitored, the change amount ΔT' of ΔT and the change amount ΔP' of ΔP per unit time are calculated, and when ΔT' or ΔP' exceeds the preset change rate threshold, the potential deception risk is determined and the warning mechanism is triggered even if ΔT does not reach T thr or ΔP does not reach P thr , to identify rapid changing deception behavior.

[0025] The method further includes a multi-source data fusion verification step: in addition to the visual SLAM module and the GNSS system, barometer height data and geomagnetic sensor data are introduced as auxiliary verification sources, and when ΔT or ΔP exceeds the threshold, secondary verification is performed in combination with the auxiliary verification source data to improve the accuracy of deception determination.

[0026] To explain the specific application of the embodiments of the application more specifically, three scenarios are used for specific description below: Scenario one: GNSS time overrun triggers return On a UAV performing a patrol task in a high-risk border area, a GNSS receiving module, a local crystal oscillator module and a visual SLAM positioning module are integrated. During flight, the system periodically collects local crystal oscillator time T1 and GNSS timestamp T2 for comparison, and the set time deviation threshold T thr is 10 milliseconds; at the same time, the positioning result P1 is obtained through SLAM mapping and positioning by the camera and IMU, and the position deviation is calculated with the GNSS position information P2, and the set spatial deviation threshold P thr is 5 meters. When the UAV flies to a specific location, the system detects that the GNSS time and the local crystal oscillator time have a deviation of 17 milliseconds, which exceeds the threshold T thr , although the SLAM and GNSS positions do not exceed the spatial deviation threshold P thr , the system still determines that the current GNSS signal has spoofing risk, triggers the return mode, and reports the related information to the ground control center, realizing the safety control closed loop.

[0027] Scenario two: spatial deviation overrun triggers emergency landing On a UAV performing a path tracking task on the edge of a combat no-fly zone, the system continuously enables GNSS positioning and SLAM positioning modules for comparison. After flying to a specific area, the system detects that the spatial deviation ΔP between the SLAM positioning result and the GNSS positioning result reaches 12 meters, which exceeds the preset spatial deviation threshold Pthr=5 meters; at the same time, the GNSS system time and the local crystal oscillator time deviation is only 3 milliseconds, which does not exceed the time threshold Tthr=10 milliseconds. Although the GNSS time is normal, the system determines that the UAV is spoofed according to the spatial anomaly, immediately triggers the safety response mechanism, and controls it to leave the current route and perform emergency landing operation, and restricts the communication link, only leaving the abnormal log back and position reporting functions. The disposal process can effectively guarantee the flight control safety, while retaining the flight data for post-analysis and safety evaluation.

[0028] Scenario three: dual anomaly recognition, triggering self-destruction program The UAV performing a special monitoring task in a high-alert level border area is equipped with the system of the application, and continuously runs the GNSS positioning, visual SLAM positioning and time comparison modules. During the task flight, the system detects two anomalies: one is that the GNSS system time and the local crystal oscillator time deviation reaches 26 milliseconds, which exceeds the preset time deviation threshold T thr= 10 milliseconds; secondly, the spatial deviation between the GNSS positioning result and the SLAM positioning result reaches 9.8 meters, far exceeding the spatial deviation threshold P thr = 5 meters. The system determines that the current GNSS signal is suspected to encounter a high-level spoofing attack, there is a double tampering behavior of time synchronization and position simulation, and it is extremely likely to be guided to the enemy control area. According to the preset high-level response strategy, the system immediately triggers the self-destruction program, interrupts the power system, cuts off the flight control connection, and ejects the key hardware modules to prevent the aircraft from crossing the border or being acquired by the enemy. Before self-destruction, the system returns the last set of flight data and positioning results to the ground control center through an encrypted channel for battlefield situation analysis and subsequent trace evidence. This scenario reflects the rapid response and task safety closed loop capability of the present application under high-risk conditions.

[0029] By adopting the embodiment of the present application, the following beneficial effects are achieved: The present application proposes a complete technical system that integrates time and space dual-dimension judgment mechanism in the field of unmanned aerial vehicle anti-spoofing, overcoming the limitations brought by the current single judgment method relying only on GNSS position anomaly detection or signal feature analysis. First, the present application constructs an independent time reference system by integrating a local high-precision crystal oscillator at the unmanned aerial vehicle end, which can verify the credibility of GNSS time in real time, fundamentally improving the recognition ability of time synchronization type spoofing attack. Second, the visual SLAM module is introduced as a non-GNSS dependent autonomous positioning means, and through coordinate comparison with GNSS position, spatial dimension anomaly detection is realized, which is particularly suitable for stable flight requirements in complex border environment and enemy electromagnetic interference scenarios. In addition, the present application constructs a standardized cross-judgment logic, allowing time anomaly and space anomaly as parallel inputs to trigger the spoofing recognition mechanism. Finally, the system also designs a safety response strategy based on judgment level, from early warning, flight restriction, return to self-destruction mechanism when necessary, with complete safety strategy. Therefore, the present application is superior to the prior art in terms of spoofing recognition ability, safety response strategy, etc., and has significant innovation and practical value.

[0030] Device embodiment one According to the embodiment of the present application, an electronic device is provided, comprising: a processor; and a memory arranged to store computer executable instructions that, when executed, cause the processor to perform the steps of the above method embodiments.

[0031] Device embodiment two According to the embodiment of the present application, a storage medium is provided for storing computer executable instructions that, when executed, implement the steps of the above method embodiments.

[0032] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A cross-validated drone anti-decoy method, characterized in that The method comprises: acquiring a local timestamp T1 based on a high-precision local crystal oscillator integrated inside the UAV, acquiring a standard time T2 provided by a GNSS navigation system, and calculating a time deviation ΔT between the local timestamp T1 and the standard time T2; acquiring a self-positioning result P1 of the UAV through a visual SLAM module, simultaneously acquiring a real-time positioning result P2 of the GNSS system, and calculating a spatial deviation ΔP between the positioning results P1 and P2; comparing the time deviation AT with a preset time deviation threshold T thr In comparison, the spatial deviation AP is compared with a preset spatial deviation threshold P thr In comparison, whether the UAV is decoyed is determined according to the comparison result. triggering a preset safety response mechanism according to the determination result.

2. The method of claim 1, wherein, The determination of whether the UAV is decoyed specifically comprises: if ΔT T thr and ΔP P thr , then the navigation is trusted, otherwise it is spoofed, wherein when ΔT≥T thr or ΔP≥P thr , further combining the magnitude of ΔT exceeding T thr , the magnitude of ΔP exceeding P thr , and the duration of both, the spoofing level is classified.

3. The method of claim 2, wherein, The safety response mechanism comprises at least one of the following: hovering, returning, limiting a flight radius, locking control, power-off, and self-destruction, and the safety response mechanism is adjusted according to a decoy level.

4. The method of claim 1, wherein, The time deviation threshold T thr And the space deviation threshold P thr According to the task type of the unmanned aerial vehicle and the flight airspace situation, and a dynamic threshold model constructed by introducing historical flight data in the adjustment process, the threshold is self-adaptively corrected according to the complexity of the flight environment.

5. The method of claim 1, wherein, The visual SLAM module constructs a local map and completes self-positioning through camera images and inertial measurement unit information, and does not rely on external electromagnetic signals; wherein a dynamic feature point screening algorithm is used in the local map construction process to eliminate suspected artificially laid interference feature points.

6. The method of claim 1, wherein, After determining that the UAV is decoyed, the system returns flight data and positioning results to a ground control center through an encrypted channel, and performs log recording.

7. The method of claim 1, wherein, The method further comprises: when the comparison and determination of the time deviation ΔT and the space deviation ΔP are performed, the change rates of the two are introduced, the change amount ΔT' of ΔT and the change amount ΔP' of ΔP in a unit of time are calculated, and when ΔT' or ΔP' exceeds a preset change rate threshold, it is determined that there is a potential spoofing risk and a warning mechanism is triggered to identify spoofing behavior, even if ΔT does not reach T thr or ΔP does not reach P thr .

8. The method of claim 1, wherein, The method further comprises a multi-source data fusion verification step: in addition to the visual SLAM module and the GNSS system, barometer height data and geomagnetic sensor data are introduced as auxiliary verification sources, when ΔT or ΔP exceeds a threshold value, secondary verification is performed in combination with the data of the auxiliary verification sources, so as to improve the accuracy of decoy determination. 9.An electronic device comprising: a processor; and a memory arranged to store computer-executable instructions that, when executed, cause the processor to perform the steps of the cross-verified UAV anti-decoy method of any one of claims 1-8. 10.A storage medium for storing computer-executable instructions that, when executed, implement the steps of the cross-verified UAV anti-decoy method of any one of claims 1-8.

Images