Unmanned aerial vehicle trusted navigation method based on anomaly detection and multi-source fusion under sensor attack

Abstract

The invention discloses an unmanned aerial vehicle trusted navigation method based on anomaly detection and multi-source fusion under sensor attack, and aims to solve the problem that the navigation precision of an unmanned aerial vehicle navigation system is reduced due to sensor attack and anomaly interference in a complex environment. Establishing kinematics and dynamics models of the unmanned aerial vehicle in an Euler angle description mode; analyzing the flight state of the unmanned aerial vehicle and a sensor measurement model, and establishing an unmanned aerial vehicle state estimation and measurement equation; analyzing sensor attack and abnormal interference conditions encountered in the navigation process of the unmanned aerial vehicle, and establishing an attack and interference mathematical model; an unmanned aerial vehicle sensor attack and interference detection framework is designed for demands, and a multi-source fusion strategy is implemented based on model prediction. The anti-interference capability of the unmanned aerial vehicle in a complex environment is effectively improved, the safety, reliability and continuity of the navigation system of the unmanned aerial vehicle are ensured, and the method is suitable for design, optimization and practical application scenes of the navigation system of the unmanned aerial vehicle. The method belongs to the field of aircraft credible multi-source navigation.

Description

Technical Field

[0001] The present invention belongs to the technical field of aircraft trusted multi-source navigation, and particularly relates to a method for trusted navigation of an unmanned aerial vehicle (UAV) for anomaly detection and multi-source fusion under sensor attacks. Background Art

[0002] In recent years, with the wide application of UAV technology, the trustworthiness and anti-interference ability of UAV navigation systems have become key research directions. Especially in complex environments, since the navigation system is vulnerable to threats from sensor attacks (such as signal spoofing, replay attacks, electromagnetic interference, etc.), the robustness and adaptability of traditional navigation methods are insufficient. Under these attacks, sensors may provide abnormal data, affecting the stability and accuracy of the navigation system, and even leading to navigation failure or mission failure of the UAV.

[0003] In existing research, multi-source sensor data fusion technology has been proven to effectively enhance the navigation ability of UAVs. By combining multiple data sources (such as economically universal sensors like GPS, inertial measurement units, magnetometers, etc.), multi-source fusion technology can utilize the advantages of each sensor to reduce the impact of single sensor failure on the navigation system. However, most current navigation methods rely on simple trustworthiness determination mechanisms when facing abnormal sensor data and cannot identify and isolate attacked data in a timely manner in a dynamic environment. In addition, existing methods mostly rely on data-driven models when solving the problem of multi-sensor anomaly detection, often requiring a large number of attack samples for training, and the effect on unknown types of attacks is limited. Summary of the Invention

[0004] The present invention aims to solve the problem of the decline in navigation accuracy of UAV navigation systems in complex environments due to sensor attacks and abnormal interference, and further proposes a method for trusted navigation of UAVs for anomaly detection and multi-source fusion under sensor attacks.

[0005] The technical solution adopted by the present invention to solve the above problems is as follows:

[0006] The method for trusted navigation of UAVs for anomaly detection, isolation and multi-source fusion under sensor attacks and interference includes the following steps:

[0007] Step 1: For the problem of trusted navigation of UAVs, establish a kinematic and dynamic model of the UAV in the form of Euler angle description;

[0008] Step 2: Analyze the flight state of the UAV and the sensor measurement model, and establish a state estimation and measurement equation of the UAV;

[0009] Step 3: Analyze the sensor attacks and abnormal interference situations encountered in the UAV navigation process, and establish an attack and interference mathematical model;

[0010] Step 4: Design a UAV sensor attack and interference detection framework according to the requirements, and implement a multi-source fusion strategy based on model prediction.

[0011] Furthermore, the UAV kinematic and dynamic models described in Step 1 include the following:

[0012]

[0013] where Fx, Fy, and Fz are the resultant forces generated by the engines and aerodynamic forces in the body coordinate system; g is the acceleration due to gravity; u, v, and w are the three-axis velocity vectors in the body coordinate system; is the three-axis acceleration vector in the body coordinate system; p, q, and r are the three-axis angular velocities in the body coordinate system;

[0014] The kinematic equations are as follows:

[0015]

[0016] where φ, θ, and ψ are the Euler angles, is the attitude angle change rate in the north-east-down ground coordinate system.

[0017] Furthermore, the sensor measurement model described in Step 2 includes the following:

[0018]

[0019] where, and are the position measurement and linear velocity measurement respectively, e P and e V represent Gaussian white noise vectors with covariance matrices Q P and Q V respectively;

[0020] The UAV heading angle is obtained through the magnetometer:

[0021]

[0022] where e ψ represents a Gaussian white noise vector with covariance matrix Q ψ respectively.

[0023] Furthermore, in Step 3, the UAV may encounter abnormal situations such as sensor attacks and interferences during flight, which may affect the navigation and positioning outputs discussed above; to solve this problem, first determine the main types of UAV sensor attacks and interferences, and conduct mathematical modeling;

[0024] GPS jamming attacks render the navigation system completely unable to determine its position; its mathematical model is as follows:

[0025] R(t) = 0 (complete signal interruption) (5) where R(t) is the received GPS signal;

[0026] Magnetic interference attacks can cause the magnetometer to output an incorrect heading angle; its mathematical model is as follows:

[0027] In an ideal situation, the geomagnetic signal is M = [Mx, My, Mz] T ; after adding the interference signal ΔM, the output signal is:

[0028] M' = M + ΔM (6)

[0029] where ΔM represents the influence of external interference, which may lead to heading angle errors;

[0030] Spoofing attacks induce the UAV navigation system to make incorrect decisions by forging signals or modifying real signals,

[0031] GPS spoofing attacks can cause the UAV to calculate incorrect position or speed information; its mathematical model is as follows:

[0032] Assume the real GPS signal is S(t) and the forged signal is S'(t), then the received signal R(t) is:

[0033] R(t) = S'(t) (7)

[0034] The forged signal S'(t) can be achieved by injecting false position offset ΔP and speed offset ΔV:

[0035] P'(t) = P(t) + ΔP (8)

[0036] V'(t) = V(t) + ΔV (9)

[0037] where P(t) and V(t) are the real position and speed respectively.

[0038] Furthermore, in step 4, the anomaly detection of UAV sensor attacks and interference is divided into four cases. By combining the detection results of GPS and magnetometer, different types of sensor attacks are classified and identified, and the identified results are used as the basis for multi-source fusion.

[0039] The beneficial effects of the present invention are:

[0040] By introducing a dynamic anomaly detection and isolation mechanism, the present invention can monitor the consistency between sensor data and multi-source navigation information in real time, enabling timely identification of potential attacks. At the same time, by dynamically adjusting data weights through multi-source fusion technology, the full utilization of trustworthy data is achieved, thereby ensuring the robustness and accuracy of the UAV navigation system in complex environments.

[0041] Aiming at the trustworthy navigation of UAVs in a sensor attack environment, the present invention designs an innovative method that integrates sensor anomaly detection and multi-source fusion to enhance the reliability and anti-interference ability of the UAV navigation system. This can not only solve the security problems in complex environments but also provide theoretical and engineering support for the realization of intelligent and trustworthy UAV navigation. Brief Description of the Drawings

[0042] Figure 1 It is a flowchart of the trustworthy navigation method for UAVs with anomaly detection, isolation, and multi-source fusion under sensor attacks and interference of the present invention. Detailed Embodiments

[0043] As Figure 1 shown, for the trustworthy navigation method of UAVs with anomaly detection and multi-source fusion under sensor attacks in this embodiment, the specific operation steps are as follows:

[0044] S1: First, construct the linear motion dynamics equation of the UAV as follows:

[0045]

[0046] where F is the resultant external force acting on the aircraft, m is the weight of the UAV, and V is the body velocity obtained in the ground coordinate system.

[0047] Next, the angular motion dynamics equation can be expressed as follows:

[0048]

[0049] where M is the resultant external moment acting on the aircraft (only affected by gravity and aerodynamic forces), and L is the angular momentum.

[0050] The force equations of the UAV are as follows:

[0051]

[0052] where Fx, Fy, and Fz are the resultant forces generated by the engines and aerodynamic forces in the body coordinate system; g is the acceleration due to gravity; u, v, and w are the three-axis velocity vectors in the body coordinate system; is the three-axis acceleration vector in the body coordinate system; p, q, and r are the three-axis angular velocities in the body coordinate system.

[0053] The kinematic equations are as follows:

[0054]

[0055] Where φ, θ, ψ are Euler angles, is the rate of change of the attitude angle in the north-east-earth ground coordinate system.

[0056] S2: Analyze the flight state of the UAV and the sensor measurement model, and establish the UAV state estimation and measurement equations; The sensors used in this invention include GPS, IMU, and magnetometer. First, the UAV obtains its position measurement P~ and linear velocity measurement

[0057]

[0058] Where, and are the position measurement and linear velocity measurement respectively, e P and e V represent Gaussian white noise vectors with covariance matrices Q P and Q V respectively.

[0059] The heading angle of the UAV is obtained through the magnetometer:

[0060]

[0061] Where, e ψ represents a Gaussian white noise vector with covariance matrix Q ψ respectively.

[0062] The inertial measurement unit (IMU) sensor mainly obtains the acceleration and angular velocity of the UAV through the accelerometer and gyroscope, and can further obtain the velocity and position information through calculation. First, the measurement model of the IMU accelerometer is as follows:

[0063]

[0064] Where, is the UAV acceleration measurement value, b a is the fixed offset of the IMU accelerometer, e a represents zero-mean Gaussian white noise.

[0065] Then, the measurement model of the IMU gyroscope is given:

[0066]

[0067] Where, is the UAV gyroscope measurement value, is the fixed bias noise of the gyroscope,

[0068] which represents the zero-mean Gaussian white noise of the gyroscope.

[0069] S3: Analyze the sensor attacks and abnormal interferences encountered during the UAV navigation process, and establish mathematical models for attacks and interferences.

[0070] During the flight of the UAV, it may encounter abnormal situations such as sensor attacks and interferences, which may affect the navigation and positioning outputs discussed above. To solve this problem, we first determined the main types of UAV sensor attacks and interferences and carried out mathematical modeling.

[0071] Attacks on the sensor layer are divided into two categories according to the attack methods: jamming attacks and spoofing attacks. Jamming attacks introduce external interference or interruption to the sensor signal, making it ineffective or outputting incorrect data. Such attacks include the following: GPS interruption, magnetic interference.

[0072] Characteristics of GPS interruption attack: The GPS signal is transmitted to the UAV navigation system through satellites, and is vulnerable to electromagnetic interference or signal power suppression, or the GPS signal is blocked from reaching the UAV through a shielding device, causing the navigation system to completely lose its positioning ability. Its mathematical modeling is as follows:

[0073] R(t) = 0 (signal completely interrupted) (18) where R(t) is the received GPS signal.

[0074] Characteristics of magnetic interference attack: The magnetometer is sensitive to the geomagnetic field and is easily affected by strong electromagnetic interference or nearby magnetic objects. The interference will cause the magnetometer to output an incorrect heading angle. Its mathematical modeling is as follows:

[0075] The geomagnetic signal in the ideal case is M = [Mx, My, Mz] T . After adding the interference signal △M, the output signal is:

[0076] M' = M + △M (19)

[0077] where ΔM represents the influence of external interference, which may cause heading angle errors.

[0078] Spoofing attacks induce the UAV navigation system to make incorrect decisions by forging signals or modifying real signals.

[0079] Characteristics of GPS spoofing attack: The attacker sends forged GPS signals, causing the UAV to calculate incorrect position or speed information. Its mathematical modeling is as follows:

[0080] Assume that the true GPS signal is S(t) and the forged signal is S'(t). Then the received signal R(t) is as follows:

[0081] R(t) = S'(t) (20)

[0082] The forged signal S'(t) can be achieved by injecting false position offset △P and velocity offset △V:

[0083] P'(t) = P(t) + △P (21)

[0084] V'(t) = V(t) + △V (22)

[0085] Where P(t) and V(t) are the true position and velocity respectively.

[0086] S4: Divide the anomaly detection of UAV sensor attacks and interferences into four cases as shown in Table 1:

[0087] Table 1 Proposed detection and fusion scheme

[0088] Serial number GPS Magnetometer Detection result 1 No alarm No alarm No abnormality 2 Alarm No alarm GPS attack 3 No alarm Alarm Magnetic interference attack 4 Alarm Alarm GPS attack and magnetic interference attack

[0089] By combining the detection results of GPS and magnetometer, different types of sensor attacks can be classified and identified. The identified results are dynamically multi-source fused through the EKF algorithm to obtain reliable UAV navigation. This method can provide a basis for the anomaly detection framework and multi-source data fusion strategy, thereby enhancing the robustness and reliability of the navigation system.

[0090] The above steps can complete the reliable navigation of the UAV under sensor attacks with anomaly detection and multi-source fusion.

[0091] The above is only a preferred embodiment of the present invention and does not impose any form of limitation on the present invention. Although the present invention has been disclosed above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art, without departing from the scope of the technical solution of the present invention, can make some changes or modifications to the above-disclosed technical content to obtain equivalent embodiments with equivalent changes. However, as long as it does not depart from the content of the technical solution of the present invention and is based on the technical essence of the present invention, any simple modification, equivalent replacement, and improvement made to the above embodiments still fall within the protection scope of the technical solution of the present invention.

Claims

1. A UAV trusted navigation method based on anomaly detection, isolation and multi-source fusion under sensor attacks and interference, characterized in that: The method comprises the following steps: Step 1: Aiming at the UAV reliable navigation problem, establish the UAV kinematic and dynamic models using Euler angle description; Step 2: Analyze the UAV flight state and sensor measurement model, and establish the UAV state estimation and measurement equations; Step 3: Analyze the sensor attacks and abnormal interference encountered during the UAV navigation process, and establish the attack and interference mathematical model; Step 4: Design a drone sensor attack and jamming detection framework based on demand, and implement a multi-source fusion strategy based on model prediction.

2. The unmanned aerial vehicle trusted navigation method for anomaly detection, isolation and multi-source fusion under sensor attack and interference according to claim 1 is characterized in that: The kinematic and dynamic models of the drone described in step 1 include the following: Among them, Fx, Fy, Fz are the combined forces generated by the power and aerodynamic force of the aircraft system; g is the acceleration of gravity; u, v, w are the three-axis velocity vectors in the aircraft coordinate system; is the three-axis acceleration vector in the body coordinate system; p, q, r are the three-axis angular velocities in the body coordinate system; The kinematic equations are as follows: Among them, φ, θ, ψ are Euler angles, is the attitude angle change rate in the north-east coordinate system.

3. The unmanned aerial vehicle trusted navigation method for anomaly detection, isolation and multi-source fusion under sensor attack and interference according to claim 1 is characterized in that: The sensor measurement model described in step 2 includes the following: in, and They are position measurement and linear velocity measurement, e P and e V Respectively represent the covariance matrix Q P and Q V Gaussian white noise vector; Drone heading angle Obtained by magnetometer: Among them, e ψ Represents the covariance matrix Q ψ Gaussian white noise vector.

4. The unmanned aerial vehicle trusted navigation method for anomaly detection, isolation and multi-source fusion under sensor attack and interference according to claim 1 is characterized in that: In step 3, the drone may encounter abnormal conditions such as sensor attacks and interference during flight, which may affect the navigation and positioning outputs discussed above; determine the main types of drone sensor attacks and interference and perform mathematical modeling; GPS interruption attack causes the navigation system to completely lose its positioning capability; its mathematical modeling is as follows: R(t)=0 (5) Where, R(t) is the received GPS signal; Magnetic interference attack will cause the magnetometer to output an incorrect heading angle; its mathematical modeling is as follows: The ideal geomagnetic signal is M = [Mx, My, Mz] T ; After adding the interference signal △M, the output signal is: M'=M+△M (6) Among them, ΔM represents the influence of external interference, which may cause heading angle error; Spoofing attacks induce drone navigation systems to make wrong decisions by forging signals or modifying real signals. GPS spoofing attacks can cause drones to calculate incorrect position or speed information. Its mathematical modeling is as follows: Assuming the real GPS signal is S(t) and the fake signal is S'(t), the received signal R(t) is: R(t)=S'(t) (7) The spurious signal S'(t) can be achieved by injecting a false position offset △P and velocity offset △V: P'(t)=P(t)+△P (8) V'(t)=V(t)+△V (9) Among them, P(t) and V(t) are the real position and velocity respectively.

5. The unmanned aerial vehicle trusted navigation method for anomaly detection, isolation and multi-source fusion under sensor attack and interference according to claim 1 is characterized in that: In step 4, the anomaly detection of drone sensor attacks and interference is divided into four cases. By combining the GPS and magnetometer detection results, different types of sensor attacks are classified and identified, and the identified results are used as the basis for multi-source fusion.

Images