Unmanned aerial vehicle active protection method and system aiming at GPS attack

Abstract

The invention provides an unmanned aerial vehicle active protection method and system for a GPS attack, and belongs to the field of unmanned aerial vehicle protection, and the method comprises the steps: obtaining GPS measurement data received at a current t moment according to the position of an unmanned aerial vehicle; performing consistency detection on the GPS measurement data at the current t moment, if the detection result is that the data is abnormal, inputting the GPS measurement data into a pre-trained time sequence neural network for prediction, outputting a GPS measurement value, and making a difference between the trajectory information in the GPS measurement value and the trajectory information in the GPS measurement data to obtain a trajectory prediction error value; and inputting the trajectory prediction error value into an unmanned aerial vehicle navigation system to realize GPS attack resolution. According to the method, the GPS attack is eliminated, the effectiveness and robustness of the whole attack protection are greatly improved, and the real-time detection and elimination capability of the unmanned aerial vehicle navigation system on the GPS attack is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of drone protection, and in particular relates to a method and system for actively protecting drones against GPS attacks. Background Art

[0002] With the development of low-altitude economy, drones are increasingly used in modern society, covering many fields from agricultural monitoring, logistics and transportation to military reconnaissance. However, when performing tasks, especially during flight navigation, drones rely heavily on the precise location information provided by GPS. Although GPS technology is widely used and reliable, its vulnerability is gradually emerging, especially in the face of malicious attacks such as GPS spoofing attacks and jamming attacks.

[0003] In response to the above problems, the following methods are usually used in the prior art to respond to GPS spoofing attacks and interference attacks on drones. 1. Signal detection and filtering technology: Detect and filter abnormal signals by analyzing the characteristics of the received GPS signals. This method relies on complex signal processing algorithms, and under high-intensity attacks, detection accuracy and real-time performance are difficult to guarantee. 2. Dual antenna system: Use two sets of antennas to receive GPS signals, and identify spoofing signals by comparing the differences between the two sets of signals. This method has a certain increase in equipment cost and weight, and performs poorly in multipath interference environments. 3. Ground-based augmentation system GBAS: Enhance positioning accuracy and anti-interference capabilities through differential GPS signals provided by ground reference stations. However, GBAS requires additional infrastructure support and has limited scope of application.

[0004] In summary, although the existing technology has enhanced the protection ability of drones against GPS attacks to some extent, in practical applications, especially in the complex environment of drone flight, the existing methods generally have the problem of too high protection costs, and the effectiveness and robustness are still insufficient. Summary of the invention

[0005] In order to solve the above problems, the present invention provides a method and system for actively protecting a drone against GPS attacks.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] An active protection method for drones against GPS attacks comprises the following steps:

[0008] Obtain the GPS measurement data received at the current time t according to the position of the drone;

[0009] The GPS measurement data at the current time t is checked for consistency. If the test result is that the data is abnormal, the GPS measurement data is input into the pre-trained time series neural network for prediction, and the GPS measurement value is output. The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value; wherein, the pre-trained time series neural network establishes a mapping relationship between the historical flight trajectory and the future flight trajectory by learning the historical trajectory information of the UAV;

[0010] The trajectory prediction error value is input into the UAV navigation system to eliminate GPS attacks.

[0011] Preferably, when the detection result of the GPS measurement data is that the data is normal, the GPS measurement data is directly input into the UAV navigation system as measurement information.

[0012] Preferably, the consistency check of the GPS measurement data at the current time t is specifically performed as follows:

[0013] Extract the drone location information from the GPS measurement data, and compare the drone location information with the preset value. If the comparison result shows no position deviation, the test result is judged to be normal data; if the comparison result shows position deviation, the test result is judged to be abnormal data.

[0014] Preferably, the temporal neural network is a GRU network.

[0015] Preferably, the training of the pre-trained temporal neural network includes:

[0016] Get the historical trajectory feature x(t) of the drone, x(t) = {lon, lat, hig, v}; where lon is the longitude value, lat is the latitude value, hig is the altitude value, and v is the speed value of the drone;

[0017] The flight trajectory data of the UAV at n consecutive moments [x(t-n+1),...,x(t)] As the input data of the timing network, it predicts n steps backward, that is, it predicts the trajectory data of n moments in the future. [x(t+1),...,x(t+n)] As the output, the time series neural network is trained to obtain a pre-trained time series neural network. The expression of the pre-trained time series neural network is:

[0018] {x(t+1),...,x(t+n)}=f[x(t-n+1),...,x(t)];

[0019] In the formula, x(t) represents the location information of the UAV at time t, that is, the historical trajectory characteristics.

[0020] Preferably, the trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain a trajectory prediction error value, specifically:

[0021] The trajectory information in the GPS measurement value and the trajectory information in the GPS measurement data are input into the pre-trained BP neural network to obtain the trajectory prediction error value.

[0022] Preferably, the training of the pre-trained BP neural network includes:

[0023] The posture quaternion of n consecutive moments of historical trajectory data is selected as the input of the BP neural network, and the preset filtering algorithm is used to predict the posture quaternion Y = [y(t), ..., y(t+n)] of n consecutive moments to obtain Y' = [y'(t), ...y'(t+n)], and the trajectory prediction error value M = Y'-Y is selected as the output of the model to train the BP neural network, and the mapping relationship between the posture data and the filtering error is obtained to form a pre-trained BP neural network.

[0024] The present invention also provides a method system for actively protecting a drone from GPS attacks, comprising:

[0025] The data acquisition module is used to obtain the GPS measurement data received at the current time t according to the position of the drone;

[0026] The data processing module is used to perform consistency detection on the GPS measurement data at the current time t. If the detection result is that the data is abnormal, the GPS measurement data is input into the pre-trained time series neural network for prediction, and the GPS measurement value is output. The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value; wherein, the pre-trained time series neural network establishes a mapping relationship between the historical flight trajectory and the future flight trajectory by learning the historical trajectory information of the UAV;

[0027] The attack mitigation module is used to input the trajectory prediction error value into the UAV navigation system to achieve the mitigation of GPS attacks.

[0028] The present invention also provides a computer device, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement any one of the steps in the method for active protection of a drone against GPS attacks.

[0029] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is loaded by a processor, it can execute any one of the steps in the method for actively protecting a drone against GPS attacks.

[0030] The active protection method for unmanned aerial vehicles against GPS attacks provided by the present invention has the following beneficial effects:

[0031] In the case of abnormal detection results of GPS measurement data, the present invention predicts GPS measurement data through a pre-trained time series neural network. The network establishes a mapping relationship between historical flight trajectories and future flight trajectories by learning the historical trajectory information of the drone, and can predict the GPS data of the next moment based on the current GPS measurement data, without relying on complex signal processing algorithms, and significantly improves the detection effect with a small number of parameters. At the same time, no other hardware is required, reducing the overall protection cost. At the same time, the trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain a trajectory prediction error value to reduce the uncertainty of the entire algorithm, and finally the trajectory prediction error value is input into the drone navigation system to achieve the elimination of GPS attacks, greatly improving the effectiveness and robustness of attack protection, and improving the real-time detection and elimination capabilities of the drone navigation system for GPS attacks. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiment of the present invention and its design scheme, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 A flowchart of a method for actively protecting a drone against GPS attacks provided in Example 1 of the present invention;

[0034] Figure 2 A flowchart of a method for actively protecting a drone against GPS attacks provided in Example 2 of the present invention;

[0035] Figure 3 This is a principle block diagram of the UAV active protection system against GPS attacks provided by the present invention. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.

[0037] Example 1

[0038] The present invention provides a method for actively protecting drones from GPS attacks, specifically: Figure 1 As shown, the following steps are included:

[0039] S1. Obtain the GPS measurement data received at the current time t according to the position of the drone.

[0040] S2. Perform consistency check on the GPS measurement data at the current time t. If the test result shows that the data is normal, the GPS measurement data is directly input into the UAV navigation system as measurement information. If the test result shows that the data is abnormal, the GPS measurement data is input into the pre-trained time series neural network for prediction, and the GPS measurement value is output. The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value. Among them, the pre-trained time series neural network establishes a mapping relationship between historical flight trajectories and future flight trajectories by learning the historical trajectory information of the UAV.

[0041] Specifically, the consistency check of the GPS measurement data at the current time t is:

[0042] Extract the drone location information from the GPS measurement data, and compare the drone location information with the preset value. If the comparison result shows no position deviation, the test result is judged to be normal data; if the comparison result shows position deviation, the test result is judged to be abnormal data.

[0043] S3. Input the trajectory prediction error value into the UAV navigation system to eliminate the GPS attack.

[0044] Example 2

[0045] When a drone is attacked, the sensor data will fluctuate significantly, so the sensor data can be directly selected as the feature quantity to form GPS measurement data. After preliminary data preprocessing, the drone data features are preliminarily selected through joint hypothesis testing, and the baseline model is used to detect different types of attacks to select appropriate data fields, and explore how to detect the most effective sensor features for attacks in the shortest time, how to determine the minimum number of required features, and how to ensure the applicability of the used detection for different types of attack detection. In addition, in view of the problem of limited computing resources for small and medium-sized commercial and civilian drones, the eigenvalues ​​are further screened according to the physical meaning of the eigenvalues. When the performance of the baseline model is sharply reduced, the present invention combines GRU (gated recurrent unit) and BP (back propagation) neural networks to improve the anti-GPS attack capability of the drone navigation system, and significantly improves the detection effect with a small number of parameters, such as Figure 2 As shown, a general processing flow for low-cost detection of drone attacks is formed.

[0046] In terms of protection, since GPS data is often used as the data input of the UAV navigation system to perform multi-sensor fusion with the IMU and calculate the optimal estimate of the position information, the typical IMU / GNSS UAV navigation system based on the extended Kalman filter algorithm (EKF) in inertial navigation is taken as the research object. Based on artificial intelligence technology, this paper studies the rapid response and dynamic defense technology of the UAV navigation system under GPS spoofing attacks.

[0047] Based on this, the active protection method for drones against GPS attacks provided in this embodiment is as follows: Figure 2 As shown, the following steps are included:

[0048] Step 1: Select the GRU network as the time series neural network, train the time series neural network, and obtain the trajectory prediction model. The specific training process is as follows:

[0049] Obtain the historical trajectory feature x(t) of the drone, x(t) = {lon, lat, hig, v}; where lon is the longitude value, lat is the latitude value, hig is the altitude value, and v is the speed value of the drone.

[0050] The flight trajectory data of the drone at n consecutive moments [x(t-n+1),...,x(t)] is used as the input data of the time series network, and the trajectory data [x(t-n+1),...,x(t)] of the next n moments is predicted n steps backward, that is, the trajectory data [x(t+1),...,x(t+n)] is used as the output to train the time series neural network, where n is the step size of the model input layer. The trajectory prediction model is obtained, that is, the pre-trained time series neural network in Example 1, and the expression of the trajectory prediction model is:

[0051] {x(t+1),...,x(t+n)}=f[x(t-n+1),...,x(t)];

[0052] In the formula, x(t) represents the location information of the UAV at time t, that is, the historical trajectory characteristics.

[0053] At the same time, the BP neural network is trained to obtain a pre-trained BP neural network. The specific training process is:

[0054] The posture quaternion of n consecutive moments of historical trajectory data is selected as the input of the BP neural network, and the preset filtering algorithm is used to predict the posture quaternion Y = [y(t), ..., y(t+n)] of n consecutive moments to obtain Y' = [y'(t), ...y'(t+n)], and the trajectory prediction error value M = Y'-Y is selected as the output of the model to train the BP neural network, and the mapping relationship between the posture data and the filtering error is obtained to form a pre-trained BP neural network.

[0055] The navigation information of the UAV system usually consists of a 24-dimensional vector Y, where each dimension of the Y vector means:

[0056] [0...3]: attitude quaternion q0, q1, q2, q3, where q0 is the scalar part, reflecting the size of the rotation angle, and q1, q2, q3 are the vector parts, together indicating the direction of the rotation axis.

[0057] [4...6]: Speed ​​in NED coordinate system (m / s).

[0058] [7...9]: Position in NED coordinate system (m).

[0059] [10...12]: IMU angle increment deviation in XYZ directions (rad).

[0060] [13...15]: IMU velocity increment deviation in XYZ directions (m / s).

[0061] [16...18]:Earth magnetic field in NED coordinate system (gauss).

[0062] [19...21]: Magnetic field of the machine body in XYZ directions (gauss).

[0063] [22...23]:wind speed NE (m / s).

[0064] Since the size of the model error has a significant relationship with the attitude of the drone, for the BP neural network, the attitude quaternion [q0, q1, q2, q3] of the trajectory data at n consecutive moments is selected as the input of the model, and the navigation information Y = [y(t),..., y(t+n)] of the trajectory data at n consecutive moments is predicted using the preset filtering algorithm to obtain Y' = [y'(t),...y'(t+n)], and the trajectory prediction error value M = Y'-Y is selected as the output of the model to train the BP neural network, and the mapping relationship between the attitude data and the filtering error is obtained to form a pre-trained BP neural network. Use the Sigmoid function to map it to (0, 1), and then perform a joint analysis with the data with the original attribute value between [0, 1]. The definition of the Sigmoid function is as follows:

[0065]

[0066] In the formula, x ij represents the i-th value of the predicted output at time j, Represents an exponential with base e.

[0067] The pre-trained BP neural network model can be expressed as:

[0068]

[0069] In the formula, w jk represents the weight of the kth data of the trajectory feature at the jth moment, Y' jk Represents the predicted output of the kth component of the trajectory feature at the jth moment, n represents the moment, j is the moment index, and k represents the kth component of the trajectory feature.

[0070] Step 2: Obtain the GPS measurement data received at the current time t according to the position of the drone.

[0071] Step 3: Perform consistency check on the GPS measurement data at the current time t. If the test result is that the data is normal, the GPS measurement data is directly input into the UAV navigation system as measurement information; if the test result is that the data is abnormal, the GPS measurement data is input into the pre-trained time series neural network for prediction, and the GPS measurement value is output. The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value; Among them, the core of the dynamic defense technology consists of two time series neural networks that reduce GPS deception and BP neural networks that reduce model uncertainty. Among them, the time series neural network establishes the mapping relationship between the historical flight trajectory and the future flight trajectory by learning the historical trajectory information of the UAV, and the BP neural network is used to learn the model error in the filtering algorithm to reduce the uncertainty of the model.

[0072] Specifically, the consistency check of the GPS measurement data at the current time t is:

[0073] Extract the drone location information from the GPS measurement data, and compare the drone location information with the preset value. If the comparison result shows no position deviation, the test result is judged to be normal data; if the comparison result shows position deviation, the test result is judged to be abnormal data.

[0074] The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value, which is specifically:

[0075] The trajectory information in the GPS measurement value and the trajectory information in the GPS measurement data are input into the pre-trained BP neural network to obtain the trajectory prediction error value.

[0076] S3. Input the trajectory prediction error value into the UAV navigation system IMU / GNSS to eliminate the GPS attack.

[0077] This embodiment improves the real-time and accuracy of detecting and eliminating GPS attacks by introducing GRU and BP neural networks, forms a fast and effective protection framework against GPS attacks, overcomes the defects of the prior art, and provides an efficient and robust drone navigation security solution.

[0078] Based on the same inventive concept, the present invention also provides a method system for actively protecting drones from GPS attacks, such as Figure 3 As shown, it includes a data acquisition module 101, a data processing module 102 and an attack mitigation module 103.

[0079] Among them, the data acquisition module 101 is used to obtain the GPS measurement data received at the current time t according to the position of the drone; the data processing module 102 is used to perform consistency detection on the GPS measurement data at the current time t. If the detection result is that the data is abnormal, the GPS measurement data is input into the pre-trained time series neural network for prediction, and the GPS measurement value is output. The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value; wherein, the pre-trained time series neural network establishes a mapping relationship between the historical flight trajectory and the future flight trajectory by learning the historical trajectory information of the drone; the attack mitigation module 103 is used to input the trajectory prediction error value into the drone navigation system to achieve the mitigation of GPS attacks.

[0080] Each module in the above-mentioned active protection method for drones against GPS attacks can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.

[0081] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps in the embodiment of the method for actively protecting a drone from GPS attacks. The specific implementation method can be found in the method embodiment, which will not be described in detail here.

[0082] Furthermore, the present invention also provides a non-temporary computer-readable storage medium containing instructions, and a computer program is stored on the storage medium. For example, a memory containing instructions, the above instructions can be executed by a processor of a computer device to complete the above method. For example, a non-temporary computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device. When the computer program is executed by the processor, the steps in the embodiment of the active protection method for drones against GPS attacks can be implemented. The specific implementation method can be found in the method embodiment, which will not be repeated here.

[0083] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0084] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0085] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0086] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0087] It should be pointed out that the specific implementation methods described above can enable those skilled in the art to understand the invention more comprehensively, but do not limit the invention in any way. Therefore, although the invention has been described in detail in this specification and embodiments, those skilled in the art should understand that the invention can still be modified or replaced by equivalents; and all technical solutions and improvements that do not deviate from the spirit and scope of the invention are included in the protection scope of the patent for the invention. Any figure mark in the claims should not be regarded as limiting the claims involved. Any simple change or equivalent replacement of the technical solution that can be obviously obtained by any technician familiar with the field within the technical scope disclosed in the present invention belongs to the protection scope of the present invention.

Claims

1. A method for actively protecting drones from GPS attacks, characterized in that: The following steps are involved: Obtain the GPS measurement data received at the current time t according to the position of the drone; The GPS measurement data at the current time t is checked for consistency. If the test result is that the data is abnormal, the GPS measurement data is input into the pre-trained time series neural network for prediction, and the GPS measurement value is output. The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value; wherein, the pre-trained time series neural network establishes a mapping relationship between the historical flight trajectory and the future flight trajectory by learning the historical trajectory information of the UAV; The trajectory prediction error value is input into the UAV navigation system to eliminate GPS attacks.

2. The method for actively protecting drones from GPS attacks according to claim 1, characterized in that: It also includes inputting the GPS measurement data directly into the UAV navigation system as measurement information when the detection result of the GPS measurement data is that the data is normal.

3. The method for actively protecting drones from GPS attacks according to claim 2, characterized in that: The consistency check of the GPS measurement data at the current time t is specifically as follows: Extract the drone location information from the GPS measurement data, and compare the drone location information with the preset value. If the comparison result shows no position deviation, the test result is judged to be normal data; if the comparison result shows position deviation, the test result is judged to be abnormal data.

4. The method for actively protecting drones from GPS attacks according to claim 1, characterized in that: The temporal neural network is a GRU network.

5. The method for actively protecting drones from GPS attacks according to claim 4, characterized in that: The training of the pre-trained temporal neural network includes: Get the historical trajectory feature x(t) of the drone, x(t) = {lon, lat, hig, v}; where lon is the longitude value, lat is the latitude value, hig is the altitude value, and v is the speed value of the drone; The flight trajectory data of the UAV at n consecutive moments [x(t-n+1),...,x(t)] is used as the input data of the time series network, and the time series neural network is trained by predicting n steps backward, that is, predicting the trajectory data of the next n moments [x(t+1),...,x(t+n)] as the output to obtain the pre-trained time series neural network. The expression of the pre-trained time series neural network is: {x(t+1),...,x(t+n)}=f[x(t-n+1),...,x(t)]; In the formula, x(t) represents the location information of the UAV at time t, that is, the historical trajectory characteristics.

6. The method for actively protecting drones from GPS attacks according to claim 1, characterized in that: The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value, which is specifically: The trajectory information in the GPS measurement value and the trajectory information in the GPS measurement data are input into the pre-trained BP neural network to obtain the trajectory prediction error value.

7. The method for actively protecting drones from GPS attacks according to claim 6, characterized in that: The training of the pre-trained BP neural network includes: The posture quaternion of n consecutive moments of historical trajectory data is selected as the input of the BP neural network, and the preset filtering algorithm is used to predict the posture quaternion Y = [y(t), ..., y(t+n)] of n consecutive moments to obtain Y' = [y'(t), ...y'(t+n)], and the trajectory prediction error value M = Y'-Y is selected as the output of the model to train the BP neural network, and the mapping relationship between the posture data and the filtering error is obtained to form a pre-trained BP neural network.

8. An active protection method system for drones against GPS attacks, characterized in that: include: The data acquisition module is used to obtain the GPS measurement data received at the current time t according to the position of the drone; The data processing module is used to perform consistency detection on the GPS measurement data at the current time t. If the detection result is that the data is abnormal, the GPS measurement data is input into the pre-trained time series neural network for prediction, and the GPS measurement value is output. The trajectory information in the GPS measurement value is subtracted from the trajectory information in the GPS measurement data to obtain the trajectory prediction error value; wherein, the pre-trained time series neural network establishes a mapping relationship between the historical flight trajectory and the future flight trajectory by learning the historical trajectory information of the UAV; The attack mitigation module is used to input the trajectory prediction error value into the UAV navigation system to achieve the mitigation of GPS attacks.

9. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is loaded into a processor, the computer program can execute the steps of the method according to any one of claims 1 to 7.

Images