Anti-spoofing-attack unmanned system state observer and observation method thereof

Abstract

The invention provides an anti-spoofing-attack unmanned system state observer and an observation method thereof, and relates to the technical field of intelligent unmanned systems. The unmanned system state observer comprises a data processing module, a spoofing attack detection module and a state estimation module which are connected in sequence. The data processing module obtains a code symbol estimation value from the decoder, calculates a differential signal estimation value of measurement output, then calculates a measurement output estimation value, and sends the measurement output estimation value to the spoofing attack detection module; the spoofing attack detection module judges whether a network spoofing attack occurs or not, performs data fusion based on a judgment result to obtain a corresponding differential signal estimation value, and sends the differential signal estimation value to the state estimation module; and the state estimation module calculates the state estimation value of the unmanned system based on the differential signal estimation value, and sends the state estimation value to a controller of the unmanned system, so that the state of the unmanned system is observed, the influence of cheating attacks is reduced or eliminated, the unmanned system is ensured to effectively control a controlled machine body, and the network security of the unmanned system is improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent unmanned systems technology, and in particular to an unmanned system state observer and its observation method that are resistant to deception attacks. Background Technology

[0002] Unmanned systems consist of platforms, mission payloads, control systems, and information and communication systems. They are comprehensive systems that integrate a series of advanced scientific and technological innovations, such as automation control technology, information and communication technology, sensor technology, and embedded technology. Their scope is quite broad, including multiple fields such as drones, robots, and autonomous driving systems.

[0003] A drone is an unmanned aerial vehicle that is radio-controlled or flies autonomously and can perform a variety of missions. A drone system mainly includes the drone airframe, flight control system, data communication network, ground remote control system, and power system. The task of the drone flight control system is to maintain the stability of the drone's flight attitude and trajectory when it is interfered with during flight, and to control the drone's flight attitude and trajectory according to the transmission commands issued by the ground remote control system.

[0004] A robot is an intelligent machine capable of semi-autonomous or fully autonomous operation, able to perform tasks such as work or movement through programming and automatic control. A robot system is a whole comprised of the robot, the work object, and the environment, including four main parts: mechanical system, drive system, control system, and perception system. A robot is an automated machine possessing some intelligent capabilities similar to humans or other living organisms, such as perception, planning, movement, and coordination; it is a highly flexible automated machine.

[0005] The driverless system is an automated control system for urban rail transit based on communication technology. Vehicles equipped with driverless systems can operate under a communication-based control system, including operations such as vehicle wake-up, station preparation, entry into the main line service, main line vehicle operation, vehicle washing, and vehicle hibernation. The starting, traction, cruising, coasting, and braking of driverless vehicles, as well as the opening and closing of doors and platform screen doors, can all be fully automated in an unmanned state.

[0006] In recent years, cyberattacks have posed an increasingly significant threat to unmanned systems, manifesting in multiple ways, including loss of control on the battlefield, data breaches, and system paralysis. Deception attacks, a type of cyberattack, utilize false information or disguise to mislead unmanned systems for illicit gains or malicious purposes. For example, cyber attackers can hijack drones, seizing control of their control systems and causing them to attack friendly targets; additionally, unencrypted stored data can be injected with false content, leading to errors in drone flight decisions.

[0007] Unmanned systems rely on state observation to achieve environmental perception and dynamic decision-making. For example, military drones use state observers to locate themselves and plan their routes in real time in high-risk environments to complete penetration missions. Therefore, developing effective state observers and observation methods for unmanned systems to eliminate or reduce the impact of deception attacks is a very meaningful research topic. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide an unmanned system state observer and its observation method that are resistant to deception attacks, which is mainly applicable to the state observation of an unmanned system, and is particularly suitable for the state observation of an unmanned system when the communication network is subjected to deception attacks.

[0009] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0010] On the one hand, the present invention provides a state observer for unmanned systems resistant to deception attacks, comprising a data processing module, a deception attack detection module, and a state estimation module connected in sequence;

[0011] The unmanned system comprises, in sequence, a controlled body, a state sensor, a state encoder, a transmitter, a data communication network, a receiver, a decoder, an unmanned system state observer, a controller, and an actuator; let h represent the data acquisition period of the unmanned system's state sensor, then the k-th sampling time t is expressed as: t = k·h; let X(k) represent the state value of the unmanned system, Y(k) represent the measured output value of the unmanned system, V(k) represent the measured noise value of the unmanned system, U(k) represent the control input value of the unmanned system, and W(k) represent the disturbance input value of the unmanned system; the discrete state equation of the unmanned system is expressed as:

[0012] X(k+1) = Ψ(X(k), U(k), W(k));

[0013] Y(k) = Φ(X(k), V(k));

[0014] Where Ψ(·) represents the functional relationship between X(k), U(k), and W(k), and Φ(·) represents the functional relationship between X(k), V(k), and Y(k); Ψ(·) and Φ(·) are determined by the specific unmanned system structure and motion characteristics.

[0015] The unmanned system obtains its measured output value Y(k) through state sensor detection and sends it to the state encoder. The state encoder uses predictive coding technology to encode the measured output value Y(k), that is, it encodes the differential signal Z(k) of the measured output value Y(k) instead of directly encoding the measured output value Y(k); let X... eY(k) represents the state prediction value of the unmanned system; the differential signal Z(k) of the measured output value Y(k) is calculated using the following formula:

[0016] Z(k) = Y(k) - Φ(X) e (k));

[0017] The differential signal Z(k) is quantized to obtain the quantized value Z of the differential signal. q (k), then encode to obtain the corresponding code symbol C(k), and the transmitter converts the code symbol C(k) into a transmission signal S(k) suitable for data communication network transmission, and then transmits the transmission signal S(k);

[0018] After receiving the signal about the code symbol C(k) through the data communication network, the receiver estimates the transmitted signal S(k) to obtain the estimated value of the transmitted signal S. a (k) will estimate the transmitted signal S a (k) is sent to the decoder; the decoder then uses the obtained transmitted signal estimate S. a Decode (k) to obtain the code symbol estimate C of code symbol C(k). a (k), then the code symbol estimate C a (k) Data processing module sent to the unmanned system state observer;

[0019] For unmanned systems, when the transmitted signal S(k) is transmitted in the data communication network, a network attacker can also simultaneously receive the transmitted signal S(k), estimate the transmitted signal S(k), obtain the estimated value of the transmitted signal, and decode the estimated value of the transmitted signal to obtain the code symbol estimate C(k). e (k); Based on code symbol estimation C e (k), the network attacker obtains the differential signal estimate Z. e (k); In order to disrupt the stability of the unmanned system and avoid being detected by the unmanned system, the network attacker carries out a deception attack; let B a (k)=0 indicates that the network attacker is carrying out a deception attack; conversely, let B a (k)=1 indicates that the network attacker did not launch a deception attack;

[0020] The data processing module obtains the previous moment's state estimate X from the state estimation module. a (k-1) and the state estimation error covariance value P(k-1), from which the code symbol estimate C is obtained from the decoder. a (k), calculate the differential signal estimate Z of the measured output value Y(k). a (k), based on the differential signal estimate Z a (k) Further calculate the estimated value of the measurement output Ya (k), for the obtained state estimate X a (k-1), State estimation error covariance P(k-1), Differential signal estimate Z a (k) and the estimated value of the measurement output Y a (k) is cached and finally provided to the deception attack detection module;

[0021] The deception attack detection module obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a Given (k-n+1) and the state estimation error covariances P(k-1), P(k-2), ..., P(kn) at n times, data analysis is used to determine whether a network spoofing attack has occurred. Based on the determination results, data fusion is performed to update the differential signal estimate Z at the current time. a (k), to obtain the corresponding differential signal estimate Z. r (k), then estimate the differential signal Z r (k) is sent to the state estimation module;

[0022] The state estimation module obtains the differential signal estimation value Z from the data processing module. r (k) is used to calculate the state estimate X of the unmanned system. a (k), and the state estimate X a (k) is sent to the controller of the unmanned system. The controller of the unmanned system further calculates the control input value U(k) of the unmanned system and sends it to the actuator of the unmanned system. The actuator of the unmanned system obtains the control input value U(k) from the controller and performs operation control on the controlled body to move according to the predetermined trajectory.

[0023] Furthermore, the data processing module includes a differential signal estimation submodule, an output signal estimation submodule, and a data buffer submodule connected in sequence;

[0024] The differential signal estimation submodule obtains the code symbol estimate C from the decoder. a (k) is decoded, and the differential signal estimate Z is calculated. a (k); During decoding, the differential signal estimation submodule uses a decoding table consistent with the encoder, and determines the symbol estimate C based on the decoding table. a (k) is transformed into the differential signal estimate Z corresponding to the differential signal Z(k). a(k); then the obtained differential signal estimate Z a (k) is sent to the data buffer submodule for storage, and also to the output signal estimation submodule;

[0025] The output signal estimation submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) obtains the previous time-step state estimate X from the data cache submodule. a (k-1), generate the estimated measurement output value Y a (k); based on the obtained state estimate X from the previous time step. a (k-1), the output signal estimation submodule calculates the state prediction value X according to the following formula. e (k):

[0026] X e (k)=Ψ(X a (k-1),U(k-1));

[0027] U(k-1) is calculated using the following formula:

[0028] U(k-1)=K(X) a (k-1));

[0029] Where K(·) represents U(k-1) and X a The functional relationship between (k-1) is determined by the specific unmanned system control algorithm;

[0030] Based on the calculated state prediction value X e (k), the output signal estimation submodule calculates the estimated measurement output value Y corresponding to the measured output value Y(k) according to the following formula. a (k):

[0031] Y a (k)=Φ(X e (k))+Z a (k);

[0032] Then, the calculated measurement output estimate Y a (k) is sent to the data cache submodule;

[0033] The data caching submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) Cache the state estimate X from the state estimation module at the previous time step. a (k-1) and the state estimation error covariance value P(k-1) are cached, and the state prediction value X is stored in the cache. e(k) Provided to the output signal estimation submodule, and the estimated measurement output value Y is obtained from the output signal estimation submodule. a (k) is buffered, and then the differential signal estimates Z at n time points in the buffer are obtained. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a (k-n+1), and the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points are sent to the deception attack detection module. The value of n is determined according to the frequency of network deception attacks.

[0034] Furthermore, the deception attack detection module includes a differential signal monitoring submodule, an output signal monitoring submodule, a state estimation error monitoring submodule, and a data fusion submodule connected in sequence.

[0035] The differential signal monitoring submodule obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Differential signal detector: Let Φ z Z represents the estimated value of the differential signal. a The covariance of (k) is defined by the differential signal detection variable H. z (k) is:

[0036] ;

[0037] in, express transpose;

[0038] For the differential signal monitoring submodule, the differential signal detection threshold is set to A. z Set the differential signal judgment result B z (k)=0 indicates that a deception attack has occurred, B z (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. z (k); when H z (k) z At that time, set B z (k)=0; when H z (k)≥A z At that time, set B​z (k)=1; then determine the result B of the differential signal. z (k) and the differential signal estimate Z a (k) is sent to the data fusion submodule;

[0039] The output signal monitoring submodule obtains the estimated output values ​​Y at n time points from the data processing module. a (k), Y a (k-1), ..., Y a (k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Measurement output detector: Let Φ y Indicates the estimated value Y of the measurement output. a The covariance of (k) is defined by the output signal detection variable H. y (k) is:

[0040] ;

[0041] in, express transpose;

[0042] For the output signal monitoring submodule, the output signal detection threshold is set to A. y Set the measurement output judgment result B y (k)=0 indicates that a deception attack has occurred, B y (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. y (k); when H y (k) y At that time, set B y (k)=0; when H y (k)≥A y At that time, set B y (k)=1; then measure the output and judge the result B. y (k) and the estimated value of the measurement output Y a (k) is sent to the data fusion submodule;

[0043] The state estimation error monitoring submodule obtains the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points from the data processing module, performs data analysis, and provides a judgment on whether a network attacker has launched a deception attack; it uses an interval X... 2 State estimation error covariance detector: Define the state estimation error covariance detection variable H p (k) is:

[0044] ;​

[0045] For the state estimation error monitoring submodule, the state estimation error covariance detection threshold is set to A. p Set the state estimation error covariance judgment result B p (k)=0 indicates that a deception attack has occurred, B p (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. p (k); when H p (k) p At that time, set B p (k)=0; when H p (k)≥A p At that time, set B p (k)=1; then determine the state estimation error covariance result B. p (k) is sent to the data fusion submodule;

[0046] The data fusion submodule receives the differential signal judgment result B from the differential signal monitoring submodule. z (k) and the differential signal estimate Z a (k), receives the measurement output judgment result B from the output signal monitoring submodule. y (k) Receives the state estimation error covariance judgment result B from the state estimation error monitoring submodule. p (k) is used to calculate the differential signal estimate Z through data fusion. r (k);

[0047] Let B o (k) represents the result of the data fusion submodule's judgment on whether a network attacker has launched a deception attack, B o (k) Calculate using the following formula:

[0048] B o (k)=B z (k)·B y (k)·B p (k);

[0049] When B o When (k)=0, the data fusion submodule determines that a network attacker may have launched a spoofing attack, and the data transmitted through the data communication network is no longer reliable, i.e., the differential signal estimate Z provided by the decoder is no longer reliable. a (k) has been tampered with by a network attacker and is no longer reliable; therefore, the differential signal estimation value Z is set. r (k)=0; when B o When (k)=1, the data fusion submodule determines that the network attacker has not launched a deception attack, and the data transmitted through the data communication network is reliable, i.e., the differential signal estimate Z provided by the decoder is reliable.​a (k) was not tampered with by a network attacker, therefore the differential signal estimate Z is set. r (k)=Z a (k);

[0050] The data fusion submodule estimates the differential signal value Z. r (k) is sent to the state estimation module.

[0051] Furthermore, in the state estimation module, let Y... r (k) represents the estimated measurement output value, which is calculated using the following formula: Y = (k - k) / (k - k) r (k):

[0052] Y r (k)=Φ(X e (k))+Z r (k);

[0053] The state equations for the state observer are established as shown in the following formula:

[0054] X a (k+1)=Ψ(X a (k),U(k))+G(Y r (k)-Φ(X a (k)));

[0055] Where G represents the observer gain;

[0056] Solve the above state equations and calculate the state estimate X. a (k); then, the state estimation module obtains the state estimate X. a (k) is sent to the controller of the unmanned system;

[0057] The controller of the unmanned system obtains a state estimate X from the state estimation module. a (k), calculate the control input value U(k) of the unmanned system, as follows:

[0058] U(k) = K(X) a (k));

[0059] Then, the controller of the unmanned system sends the control input value U(k) to the actuator of the unmanned system.

[0060] On the other hand, the present invention also provides a method for state observation of unmanned systems resistant to deception attacks, which is implemented using the aforementioned unmanned system state observer resistant to deception attacks, and includes the following steps:

[0061] Step 1: The data processing module obtains the state estimate X from the state estimation module at the previous moment. a(k-1) and the state estimation error covariance value P(k-1), from which the code symbol estimate C is obtained from the decoder. a (k), calculate the differential signal estimate Z of the measured output value Y(k). a (k), based on the differential signal estimate Z a (k) Further calculate the estimated value of the measurement output Y a (k), for the obtained state estimate X a (k-1), State estimation error covariance P(k-1), Differential signal estimate Z a (k) and the estimated value of the measurement output Y a (k) is cached and finally provided to the deception attack detection module;

[0062] Step 2: The deception attack detection module obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a Given (k-n+1) and the state estimation error covariances P(k-1), P(k-2), ..., P(kn) at n times, data analysis is used to determine whether a network spoofing attack has occurred. Based on the determination results, data fusion is performed to update the differential signal estimate Z at the current time. a (k), to obtain the corresponding differential signal estimate Z. r (k), then estimate the differential signal Z r (k) is sent to the state estimation module;

[0063] Step 3: The state estimation module obtains the differential signal estimate Z from the data processing module. r (k) is used to calculate the state estimate X of the unmanned system. a (k), and the state estimate X a (k) is sent to the controller of the unmanned system, thereby enabling state observation of the unmanned system.

[0064] Furthermore, the specific method of step 1 is as follows:

[0065] Step 1.1: The differential signal estimation submodule obtains the code symbol estimate C from the decoder. a (k) is decoded, and the differential signal estimate Z is calculated. a (k); then the obtained differential signal estimate Z a (k) is sent to the data buffer submodule for storage, and also to the output signal estimation submodule;

[0066] Step 1.2: The output signal estimation submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) obtains the previous time-step state estimate X from the data cache submodule. a (k-1), generate the estimated measurement output value Y a (k); based on the obtained state estimate X from the previous time step. a (k-1), the output signal estimation submodule calculates the state prediction value X according to the following formula. e (k):

[0067] X e (k)=Ψ(X a (k-1),U(k-1));

[0068] U(k-1) is calculated using the following formula:

[0069] U(k-1)=K(X) a (k-1));

[0070] Where K(·) represents U(k-1) and X a The functional relationship between (k-1) is determined by the specific unmanned system control algorithm;

[0071] Based on the calculated state prediction value X e (k), the output signal estimation submodule calculates the estimated measurement output value Y corresponding to the measured output value Y(k) according to the following formula. a (k):

[0072] Y a (k)=Φ(X e (k))+ Z a (k);

[0073] Then, the calculated measurement output estimate Y a (k) is sent to the data cache submodule;

[0074] Step 1.3: The data caching submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) Cache the state estimate X from the state estimation module at the previous time step. a (k-1) and the state estimation error covariance value P(k-1) are cached, and the state prediction value X is stored in the cache. e (k) Provided to the output signal estimation submodule, and the estimated measurement output value Y is obtained from the output signal estimation submodule. a (k) is buffered, and then the differential signal estimates Z at n time points in the buffer are obtained. a (k), Za (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a (k-n+1), and the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points are sent to the deception attack detection module.

[0075] Furthermore, the specific method for step 2 is as follows:

[0076] Step 2.1: The differential signal monitoring submodule obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Differential signal detector: Let Φ z Z represents the estimated value of the differential signal. a The covariance of (k) is defined by the differential signal detection variable H. z (k) is:

[0077] ;

[0078] For the differential signal monitoring submodule, the differential signal detection threshold is set to A. z Set the differential signal judgment result B z (k)=0 indicates that a deception attack has occurred, B z (k)=1 indicates that a deception attack did not occur;

[0079] Calculate H using the formula above. z (k); when H z (k) z At that time, set B z (k)=0; when H z (k)≥A z At that time, set B z (k)=1; then determine the result B of the differential signal. z (k) and the differential signal estimate Z a (k) is sent to the data fusion submodule;

[0080] Step 2.2: The output signal monitoring submodule obtains the estimated output values ​​Y at n time points from the data processing module. a (k), Y a (k-1), ..., Y a ​(k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Measurement output detector: Let Φ y Indicates the estimated value Y of the measurement output. a The covariance of (k) is defined by the output signal detection variable H. y (k) is:

[0081] ;

[0082] For the output signal monitoring submodule, the output signal detection threshold is set to A. y Set the measurement output judgment result B y (k)=0 indicates that a deception attack has occurred, B y (k)=1 indicates that a deception attack did not occur;

[0083] Calculate H using the formula above. y (k); when H y (k) y At that time, set B y (k)=0; when H y (k)≥A y At that time, set B y (k)=1; then measure the output and judge the result B. y (k) and the estimated value of the measurement output Y a (k) is sent to the data fusion submodule;

[0084] Step 2.3: The state estimation error monitoring submodule obtains the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points from the data processing module, performs data analysis, and provides a judgment on whether the network attacker has launched a deception attack; using an interval X... 2 State estimation error covariance detector: Define the state estimation error covariance detection variable H p (k) is:

[0085] ;

[0086] For the state estimation error monitoring submodule, the state estimation error covariance detection threshold is set to A. p Set the state estimation error covariance judgment result B p (k)=0 indicates that a deception attack has occurred, B p (k)=1 indicates that a deception attack did not occur;

[0087] Calculate H using the formula above. p (k); when H p (k) p ​​At that time, set B p (k)=0; when H p (k)≥A p At that time, set B p (k)=1; then determine the state estimation error covariance result B. p (k) is sent to the data fusion submodule;

[0088] Step 2.4: The data fusion submodule receives the differential signal judgment result B from the differential signal monitoring submodule. z (k) and the differential signal estimate Z a (k), receives the measurement output judgment result B from the output signal monitoring submodule. y (k) Receives the state estimation error covariance judgment result B from the state estimation error monitoring submodule. p (k) is used to calculate the differential signal estimate Z through data fusion. r (k); Let B o (k) represents the result of the data fusion submodule's judgment on whether a network attacker has launched a deception attack, calculated using the following formula:

[0089] B o (k)=B z (k)·B y (k)·B p (k);

[0090] When B o When (k)=0, the data fusion submodule determines that a network attacker may have launched a spoofing attack, and the data transmitted through the data communication network is no longer reliable, i.e., the differential signal estimate Z provided by the decoder is no longer reliable. a (k) has been tampered with by a network attacker and is no longer reliable; therefore, the differential signal estimation value Z is set. r (k)=0; when B o When (k)=1, the data fusion submodule determines that the network attacker has not launched a deception attack, and the data transmitted through the data communication network is reliable, i.e., the differential signal estimate Z provided by the decoder is reliable. a (k) was not tampered with by a network attacker, therefore the differential signal estimate Z is set. r (k)=Z a (k);

[0091] The data fusion submodule estimates the differential signal value Z. r (k) is sent to the state estimation module.

[0092] The beneficial effects of adopting the above technical solution are as follows: The unmanned system state observer and its observation method resistant to deception attacks provided by this invention address the problem of network attackers launching deception attacks and tampering with data transmitted through data communication networks, thereby causing distortion of unmanned system state observation data and preventing effective control of the controlled machine. This is achieved by using the interval X... 2 Differential signal detector, interval X 2 Measurement output detector and interval X 2 The state estimation error covariance detector calculates the estimated value Z of the differential signal through data fusion. r (k) Design an observation system to obtain the state estimate X a (k) Reduce or eliminate the impact of deception attacks and ensure effective control of the unmanned system's controlled body. Attached Figure Description

[0093] Figure 1 This is a schematic diagram of the unmanned system structure;

[0094] Figure 2 This is a block diagram of the unmanned system state observer provided in Embodiment 1 of the present invention;

[0095] Figure 3 This is a schematic diagram of the data processing module provided in Embodiment 1 of the present invention;

[0096] Figure 4 This is a schematic diagram of the deception attack detection module provided in Embodiment 1 of the present invention;

[0097] Figure 5 This is a flowchart of the unmanned system state observation method provided in Embodiment 2 of the present invention. Detailed Implementation

[0098] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0099] Unmanned systems, such as Figure 1 As shown, the system includes a controlled machine, state sensors, a state encoder, a transmitter, a data communication network, a receiver, a decoder, an unmanned system state observer, a controller, and actuators. The state sensors detect and obtain the measured output value Y(k) of the unmanned system, which is then sent to the state encoder. The state encoder uses predictive coding technology to encode the measured output value Y(k), that is, it encodes the differential signal Z(k) of the measured output value Y(k) instead of directly encoding the measured output value Y(k). Let X... e Z(k) represents the predicted state value of the unmanned system. The differential signal Z(k) is calculated using the following formula:

[0100] Z(k)=Y(k)-Φ(X e (k));

[0101] Generate a differential signal Z(k) of the measured output value Y(k), and quantize the differential signal Z(k) to obtain the quantized value Zk of the differential signal. q (k), then encode to obtain the corresponding code symbol C(k), and the transmitter converts the code symbol C(k) into a transmission signal S(k) suitable for data communication network transmission, and then transmits the transmission signal S(k).

[0102] After receiving the signal about the code symbol C(k) through the data communication network, the receiver estimates the transmitted signal S(k) to obtain the estimated value of the transmitted signal S. a (k) will estimate the transmitted signal S a (k) is sent to the decoder. The decoder then uses the obtained transmitted signal estimate S. a Decode (k) to obtain the code symbol estimate C of code symbol C(k). a (k), then the code symbol estimate C a (k) Data processing module sent to the unmanned system state observer.

[0103] For unmanned systems, when the transmitted signal S(k) is transmitted in the data communication network, a network attacker can also simultaneously receive the transmitted signal S(k), estimate the transmitted signal S(k), obtain the estimated value of the transmitted signal, and decode it to obtain the code symbol estimate C(k). e (k). Because the decoder of the unmanned system differs from the decoder used by network attackers, therefore C e (k) and C a (k) may be equal or may have some differences. Based on the code symbol estimate C e (k), a cyber attacker can obtain the differential signal estimate Z. e (k). Cyber ​​attackers, in order to disrupt the stability of unmanned systems while avoiding detection by the systems themselves, employ deception attacks. Let B... a (k)=0 indicates that the network attacker is carrying out a deception attack; conversely, let B... a (k)=1 indicates that the attacker did not launch a deception attack. The specific steps an attacker takes to launch a deception attack are as follows:

[0104] (1) By monitoring the data communication network, the transmitted data is stolen and the differential signal estimate Z is intercepted. e (k);

[0105] (2) Estimate the differential signal Z e (k) transforms into Z f(k) employs the following linear attack strategy:

[0106] Z f (k)=Γ(k)Z e (k)+Υ(k);

[0107] Where Γ(k) is a constant matrix, and Υ(k) is a Gaussian random vector with a mean of 0. Appropriate Γ(k) and Υ(k) are chosen such that Z... f (k) and Z e The root mean squares of (k) are equal;

[0108] (3) Z f (k) is encoded to obtain the corresponding code symbol. The code symbol is then converted into a transmission signal suitable for data communication network transmission by the transmitter, and then the transmission signal is transmitted.

[0109] (4) In order to reduce the probability of being detected by unmanned systems and to more effectively improve the attack effect, network attackers adopt a random attack strategy, that is, the probability of a deception attack occurring is P(B). a (k)=0)<1;

[0110] The unmanned system state observer obtains the code symbol estimate C from the decoder. a (k) is used to calculate the state estimate X of the unmanned system. a (k), and the state estimate X a (k) The state estimate is sent to the controller of the unmanned system. The controller obtains the state estimate X from the state estimation module of the unmanned system state observer. a (k), calculate the control input value U(k) of the unmanned system. The control input value U(k) is calculated using the following formula:

[0111] U(k) = K(X) a (k));

[0112] Then, the controller sends the control input value U(k) to the actuator. The actuator receives the control input value U(k) from the controller and performs operational control on the controlled machine, causing it to move along a predetermined trajectory.

[0113] Example 1

[0114] Unmanned system state observers, such as Figure 2 As shown, it includes a data processing module, a spoofing attack detection module, and a state estimation module connected in sequence.

[0115] The main task of the data processing module is to obtain the state estimate X from the state estimation module at the previous moment. a (k-1) and the state estimation error covariance value P(k-1), from which the code symbol estimate C is obtained from the decoder. a(k), calculate the differential signal estimate Z of the measured output value Y(k). a (k), based on the differential signal estimate Z a (k) Further calculate the estimated value of the measurement output Y a (k), for the obtained state estimate X a (k-1), State estimation error covariance P(k-1), Differential signal estimate Z a (k) and the estimated value of the measurement output Y a (k) is cached and finally provided to the deception attack detection module, such as Figure 3 As shown, it includes a differential signal estimation submodule, an output signal estimation submodule, and a data buffer submodule connected in sequence.

[0116] The differential signal estimation submodule obtains the code symbol estimate C from the decoder. a (k) is decoded, and the differential signal estimate Z is calculated. a (k). During decoding, the differential signal estimation submodule uses a decoding table consistent with the encoder, and determines the symbol estimate C based on the decoding table. a (k) is transformed into the differential signal estimate Z corresponding to the differential signal Z(k). a (k). Then, the obtained differential signal estimate Z is... a (k) is sent to the data buffer submodule for storage, and also to the output signal estimation submodule.

[0117] The output signal estimation submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) obtains the previous time-step state estimate X from the data cache submodule. a (k-1), generate the estimated measurement output value Y a (k). Based on the obtained state estimate X from the previous time step. a (k-1), the output signal estimation submodule can calculate the state prediction value X according to the following formula. e (k):

[0118] X e (k)=Ψ(X a (k-1),U(k-1));

[0119] U(k-1) is calculated using the following formula:

[0120] U(k-1)=K(X) a (k-1));

[0121] In the formula above, K(·) represents U(k-1) and X. aThe functional relationship between (k-1) is determined by the specific unmanned system control algorithm. This is based on the calculated state prediction value X. e The output signal estimation submodule can calculate the estimated value Y(k) corresponding to the measured output value Y(k) according to the following formula. a (k):

[0122] Y a (k)=Φ(X e (k))+Z a (k);

[0123] Then, the calculated measurement output estimate Y a (k) is sent to the data cache submodule.

[0124] The data caching submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) Cache the state estimate X from the state estimation module at the previous time step. a (k-1) and the state estimation error covariance value P(k-1) are cached, and the state prediction value X is stored in the cache. e (k) Provided to the output signal estimation submodule, and the estimated measurement output value Y is obtained from the output signal estimation submodule. a (k) is buffered, and then the differential signal estimates Z at n time points in the buffer are obtained. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k) 、Y a (k-1), ..., Y a (k-n+1), and the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points are sent to the deception attack detection module. The value of n can be determined according to the frequency of network deception attacks.

[0125] The deception attack detection module obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y aGiven (k-n+1) and the state estimation error covariances P(k-1), P(k-2), ..., P(kn) at n times, data analysis is used to determine whether a network spoofing attack has occurred. Based on the determination results, data fusion is performed to update the differential signal estimate Z at the current time. a (k), to obtain the corresponding differential signal estimate Z. r (k), then estimate the differential signal Z r (k) Send to the state estimation module, such as Figure 4 As shown, it includes a differential signal monitoring submodule, an output signal monitoring submodule, a state estimation error monitoring submodule, and a data fusion submodule connected in sequence.

[0126] The differential signal monitoring submodule obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1) is used, and data analysis is performed to determine whether a network attacker has launched a deception attack. Unlike existing literature, this invention employs an interval X... 2 Differential signal detector. Let Φ z Z represents the estimated value of the differential signal. a The covariance of (k). Define the differential signal detection variable H. z (k) is:

[0127] ;

[0128] For the differential signal monitoring submodule, the differential signal detection threshold is set to A. z Set the differential signal judgment result B z (k)=0 indicates a deception attack has occurred; conversely, setting the differential signal to determine the result B... z (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. z (k). When H z (k)< A z At that time, set B z (k)=0; conversely, when H z (k)≥ A z At that time, set B z (k)=1. Then determine the result B of the differential signal. z (k) and the differential signal estimate Z a (k) is sent to the data fusion submodule.

[0129] The output signal monitoring submodule obtains the estimated output values ​​Y at n time points from the data processing module. a (k), Y a(k-1), ..., Y a (k-n+1) is used, and data analysis is performed to determine whether a network attacker has launched a deception attack. Unlike existing literature, this invention employs an interval X... 2 Measurement output detector. Let Φ y Indicates the estimated value Y of the measurement output. a The covariance of (k). Define the output signal detection variable H. y (k) is:

[0130] ;

[0131] For the output signal monitoring submodule, the output signal detection threshold is set to A. y Set the measurement output judgment result B y (k)=0 indicates a deception attack has occurred; conversely, setting the measurement output judgment result B... y (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. y (k). When H y (k)< A y At that time, set B y (k)=0; conversely, when H y (k)≥ A y At that time, set B y (k)=1. Then measure and output the judgment result B. y (k) is sent to the data fusion submodule.

[0132] The state estimation error monitoring submodule obtains the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points from the data processing module, performs data analysis, and provides a judgment on whether a network attacker has launched a deception attack. Unlike existing literature, this invention uses an interval X... 2 State estimation error covariance detector. Define the state estimation error covariance detection variable H. p (k) is:

[0133] ;

[0134] For the state estimation error monitoring submodule, the state estimation error covariance detection threshold is set to A. p Set the state estimation error covariance judgment result B p (k)=0 indicates that a deception attack has occurred; conversely, setting the state estimation error covariance to determine the result B. p (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. p (k). When H p (k) <Ap At that time, set B p (k)=0; conversely, when H p (k)≥A p At that time, set B p (k)=1. Then, determine the state estimation error covariance result B. p (k) is sent to the data fusion submodule.

[0135] The data fusion submodule receives the differential signal judgment result B from the differential signal monitoring submodule. z (k) and the differential signal estimate Z a (k), receives the measurement output judgment result B from the output signal monitoring submodule. y (k) Receives the state estimation error covariance judgment result B from the state estimation error monitoring submodule. p (k) is used to calculate the differential signal estimate Z through data fusion. r (k).

[0136] Let B o (k) represents the result of the data fusion submodule's judgment on whether a network attacker has launched a deception attack. B o (k) can be calculated using the following formula:

[0137] B o (k)=B z (k)·B y (k)·B p (k);

[0138] When B o When (k)=0, the data fusion submodule determines that a network attacker may have launched a spoofing attack, and the data transmitted through the data communication network is no longer reliable, that is, the differential signal estimate Z provided by the decoder is lost. a (k) has been tampered with by a network attacker and is no longer reliable; therefore, the differential signal estimation value Z is set. r (k)=0; conversely, when B o When (k)=1, the data fusion submodule determines that the network attacker has not launched a deception attack, and the data transmitted through the data communication network is reliable, that is, the differential signal estimate Z provided by the decoder is reliable. a (k) was not tampered with by a network attacker, therefore the differential signal estimate Z is set. r (k)=Z a (k). Then, the data fusion submodule estimates the differential signal Z. r (k) is sent to the state estimation module.

[0139] The state estimation module obtains the differential signal estimate Z from the data processing module. r(k) is used to calculate the state estimate X of the unmanned system. a (k), and the state estimate X a (k) is sent to the controller of the unmanned system. Let Y... r (k) represents the estimated measurement output value. The state estimation module calculates the estimated measurement output value Y using the following formula. r (k):

[0140] Y r (k)=Φ(X e (k))+Z r (k);

[0141] The state estimation module establishes the state equations for the state observer, as shown in the following formula:

[0142] X a (k+1)=Ψ(X a (k),U(k))+G(Y r (k)-Φ(X a (k)));

[0143] Where G represents the observer gain. Solve the above state equation to calculate the state estimate X. a (k). Then, the state estimation module obtains the state estimate X. a (k) is sent to the controller of the unmanned system.

[0144] Example 2

[0145] A method for state observation of unmanned systems resistant to deception attacks is implemented using the aforementioned unmanned system state observer, such as... Figure 5 As shown, it includes the following steps:

[0146] Step 1: The data processing module obtains the state estimate X from the state estimation module at the previous moment. a (k-1) and the state estimation error covariance value P(k-1), from which the code symbol estimate C is obtained from the decoder. a (k), calculate the differential signal estimate Z of the measured output value Y(k). a (k), based on the differential signal estimate Z a (k) Further calculate the estimated value of the measurement output Y a (k), for the obtained state estimate X a (k-1), State estimation error covariance P(k-1), Differential signal estimate Z a (k) and the estimated value of the measurement output Y a (k) is cached and finally provided to the deception attack detection module. The specific method is as follows:

[0147] Step 1.1: The differential signal estimation submodule obtains the code symbol estimate C from the decoder. a (k) is decoded, and the differential signal estimate Z is calculated. a (k); then the obtained differential signal estimate Z a (k) is sent to the data buffer submodule for storage, and also to the output signal estimation submodule;

[0148] Step 1.2: The output signal estimation submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) obtains the previous time-step state estimate X from the data cache submodule. a (k-1), generate the estimated measurement output value Y a (k), and send it to the data cache submodule;

[0149] Step 1.3: The data caching submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) Cache the state estimate X from the state estimation module at the previous time step. a (k-1) and the state estimation error covariance value P(k-1) are cached, and the state prediction value X is stored in the cache. e (k) Provided to the output signal estimation submodule, and the estimated measurement output value Y is obtained from the output signal estimation submodule. a (k) is buffered, and then the differential signal estimates Z at n time points in the buffer are obtained. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a (k-n+1), and the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points are sent to the deception attack detection module;

[0150] Step 2: The deception attack detection module obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y aGiven (k-n+1) and the state estimation error covariances P(k-1), P(k-2), ..., P(kn) at n times, data analysis is used to determine whether a network spoofing attack has occurred. Based on the determination results, data fusion is performed to update the differential signal estimate Z at the current time. a (k), to obtain the corresponding differential signal estimate Z. r (k), then estimate the differential signal Z r (k) is sent to the state estimation module, specifically as follows:

[0151] Step 2.1: The differential signal monitoring submodule obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), and perform data analysis to determine whether the network attacker has launched a deception attack. Then, the differential signal judgment result B is used. z (k) and the differential signal estimate Z a (k) is sent to the data fusion submodule;

[0152] Step 2.2: The output signal monitoring submodule obtains the estimated output values ​​Y at n time points from the data processing module. a (k), Y a (k-1), ..., Y a (k-n+1), and perform data analysis on it to determine whether the network attacker has launched a deception attack. Then, measure and output the judgment result B. y (k) is sent to the data fusion submodule;

[0153] Step 2.3: The state estimation error monitoring submodule obtains the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points from the data processing module, performs data analysis, and provides a judgment on whether the network attacker has launched a deception attack. Then, the state estimation error covariance judgment result B is calculated. p (k) is sent to the data fusion submodule;

[0154] Step 2.4: The data fusion submodule receives the differential signal judgment result B from the differential signal monitoring submodule. z (k) and the differential signal estimate Z a (k), receives the measurement output judgment result B from the output signal monitoring submodule. y (k) and the estimated value of the measurement output Y a (k) Receives the state estimation error covariance judgment result B from the state estimation error monitoring submodule. p (k) is used to calculate the differential signal estimate Z through data fusion.r (k), and then estimate the differential signal Z. r (k) is sent to the state estimation module.

[0155] Step 3: The state estimation module obtains the differential signal estimate Z from the data processing module. r (k) is used to calculate the state estimate X of the unmanned system. a (k), and the state estimate X a (k) is sent to the controller of the unmanned system. Let Y... r (k) represents the estimated measurement output value. The state estimation module calculates the estimated measurement output value Y using the following formula. r (k):

[0156] Y r (k)=Φ(X e (k))+Z r (k);

[0157] The state estimation module establishes the state equations for the state observer, as shown in the following formula:

[0158] X a (k+1)=Ψ(X a (k),U(k))+G(Y r (k)-Φ(X a (k)));

[0159] Where G represents the observer gain. Solve the above state equation to calculate the state estimate X. a (k). Then, the state estimation module obtains the state estimate X. a (k) is sent to the controller of the unmanned system.

[0160] This invention provides a state observer and observation method for unmanned systems resistant to deception attacks. Addressing the problem of network attackers launching deception attacks and tampering with transmitted data in data communication networks, it employs an interval X... 2 Differential signal detector, interval X 2 Measurement output detector and interval X 2 The state estimation error covariance detector effectively monitors deception attacks launched by network attackers. By using data fusion methods, it calculates differential signal estimates, designs observation systems to obtain state estimates, reduces or eliminates the impact of deception attacks, ensures effective control of the controlled machine by the unmanned system, and improves the network security of the unmanned system.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope defined by the present invention.

Claims

1. A state observer for an unmanned system resistant to deception attacks, characterized in that: It includes a data processing module, a deception attack detection module, and a state estimation module connected in sequence; The unmanned system comprises, in sequence, a controlled body, a state sensor, a state encoder, a transmitter, a data communication network, a receiver, a decoder, an unmanned system state observer, a controller, and an actuator; let h represent the data acquisition period of the unmanned system's state sensor, then the k-th sampling time t is expressed as: t = k·h; let X(k) represent the state value of the unmanned system, Y(k) represent the measured output value of the unmanned system, V(k) represent the measured noise value of the unmanned system, U(k) represent the control input value of the unmanned system, and W(k) represent the disturbance input value of the unmanned system; the discrete state equation of the unmanned system is expressed as: X(k+1) = Ψ(X(k), U(k), W(k)); Y(k) = Φ(X(k), V(k)); Where Ψ(·) represents the functional relationship between X(k), U(k), and W(k), and Φ(·) represents the functional relationship between X(k), V(k), and Y(k); Ψ(·) and Φ(·) are determined by the specific unmanned system structure and motion characteristics. The unmanned system obtains its measured output value Y(k) through state sensor detection and sends it to the state encoder. The state encoder uses predictive coding technology to encode the measured output value Y(k), that is, it encodes the differential signal Z(k) of the measured output value Y(k) instead of directly encoding the measured output value Y(k); let X... e Y(k) represents the predicted state value of the unmanned system; the differential signal Z(k) of the measured output value Y(k) is calculated using the following formula: Z(k)= Y(k)-Φ(X e (k)); The differential signal Z(k) is quantized to obtain the quantized value Z of the differential signal. q (k), then encode to obtain the corresponding code symbol C(k), and convert the code symbol C(k) into a transmission signal S(k) suitable for data communication network transmission through the transmitter, and then transmit the transmission signal S(k); After receiving the signal about the code symbol C(k) through the data communication network, the receiver estimates the transmitted signal S(k) to obtain the estimated value of the transmitted signal S. a (k) will estimate the transmitted signal S a (k) is sent to the decoder; the decoder then uses the obtained transmitted signal estimate S. a Decode (k) to obtain the code symbol estimate C of code symbol C(k). a (k), then the code symbol estimate C a (k) Data processing module sent to the unmanned system state observer; For unmanned systems, when the transmitted signal S(k) is transmitted in the data communication network, a network attacker can also simultaneously receive the transmitted signal S(k), estimate the transmitted signal S(k), obtain the estimated value of the transmitted signal, and decode the estimated value of the transmitted signal to obtain the code symbol estimate C(k). e (k); Based on code symbol estimation C e (k), the network attacker obtains the differential signal estimate Z. e (k); In order to disrupt the stability of the unmanned system and avoid being detected by the unmanned system, the network attacker carries out a deception attack; let B a (k)=0 indicates that the network attacker is carrying out a deception attack; conversely, let B a (k)=1 indicates that the network attacker did not launch a deception attack; The data processing module obtains the previous state estimate value X from the state estimation module. a (k-1) and the state estimation error covariance value P(k-1), from which the code symbol estimate C is obtained from the decoder. a (k), calculate the differential signal estimate Z of the measured output value Y(k). a (k), based on the differential signal estimate Z a (k) Further calculate the estimated value of the measurement output Y a (k), for the obtained state estimate X a (k-1), State estimation error covariance P(k-1), Differential signal estimate Z a (k) and the estimated output value Y a (k) is cached and finally provided to the deception attack detection module; The deception attack detection module obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a Given (k-n+1) and the state estimation error covariances P(k-1), P(k-2), ..., P(kn) at n times, data analysis is used to determine whether a network spoofing attack has occurred. Based on the determination results, data fusion is performed to update the differential signal estimate Z at the current time. a (k), to obtain the corresponding differential signal estimate Z. r (k), then estimate the differential signal Z r (k) is sent to the state estimation module; The state estimation module obtains the differential signal estimation value Z from the data processing module. r (k) is used to calculate the state estimate X of the unmanned system. a (k), and the state estimate X a (k) is sent to the controller of the unmanned system. The controller of the unmanned system further calculates the control input value U(k) of the unmanned system and sends it to the actuator of the unmanned system. The actuator of the unmanned system obtains the control input value U(k) from the controller and performs operation control on the controlled body to move according to the predetermined trajectory.

2. The unmanned system state observer resistant to deception attacks according to claim 1, characterized in that: The data processing module includes a differential signal estimation submodule, an output signal estimation submodule, and a data buffer submodule connected in sequence. The differential signal estimation submodule obtains the code symbol estimate C from the decoder. a (k) is decoded, and the differential signal estimate Z is calculated. a (k); During decoding, the differential signal estimation submodule uses a decoding table consistent with the encoder, and determines the symbol estimate C based on the decoding table. a (k) is transformed into the differential signal estimate Z corresponding to the differential signal Z(k). a (k); then the obtained differential signal estimate Z a (k) is sent to the data buffer submodule for storage, and also to the output signal estimation submodule; The output signal estimation submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) obtains the previous time-step state estimate X from the data cache submodule. a (k-1), generate the estimated measurement output value Y a (k); based on the obtained state estimate X from the previous time step. a (k-1), the output signal estimation submodule calculates the state prediction value X according to the following formula. e (k): X e (k)=Ψ(X a (k-1),U(k-1)); U(k-1) is calculated using the following formula: U(k-1)=K(X a (k-1)); Where K(·) represents U(k-1) and X a The functional relationship between (k-1) is determined by the specific unmanned system control algorithm; Based on the calculated state prediction value X e (k), the output signal estimation submodule calculates the estimated measurement output value Y corresponding to the measured output value Y(k) according to the following formula. a (k): Y a (k)=Φ(X e (k))+Z a (k); Then, the calculated measurement output estimate Y a (k) is sent to the data cache submodule; The data caching submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) Cache the state estimate X from the state estimation module at the previous time step. a (k-1) and the state estimation error covariance value P(k-1) are cached, and the state prediction value X is stored in the cache. e (k) Provided to the output signal estimation submodule, and the estimated measurement output value Y is obtained from the output signal estimation submodule. a (k) is buffered, and then the differential signal estimates Z at n time points in the buffer are obtained. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a (k-n+1), and the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points are sent to the deception attack detection module. The value of n is determined according to the frequency of network deception attacks.

3. The unmanned system state observer resistant to deception attacks according to claim 2, characterized in that: The deception attack detection module includes a differential signal monitoring submodule, an output signal monitoring submodule, a state estimation error monitoring submodule, and a data fusion submodule connected in sequence. The differential signal monitoring submodule obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Differential signal detector: Let Φ z Z represents the estimated value of the differential signal. a The covariance of (k) is defined by the differential signal detection variable H. z (k) is: ； in, express transpose; For the differential signal monitoring submodule, the differential signal detection threshold is set to A. z Set the differential signal judgment result B z (k)=0 indicates that a deception attack has occurred, B z (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. z (k); when H z (k) z At that time, set B z (k)=0; when H z (k)≥A z At that time, set B z (k)=1; then determine the result B of the differential signal. z (k) and the differential signal estimate Z a (k) is sent to the data fusion submodule;​ The output signal monitoring submodule obtains the estimated output values ​​Y at n time points from the data processing module. a (k), Y a (k-1), ..., Y a (k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Measurement output detector: Let Φ y Indicates the estimated value Y of the measurement output. a The covariance of (k) is defined by the output signal detection variable H. y (k) is: ； in, express transpose; For the output signal monitoring submodule, the output signal detection threshold is set to A. y Set the measurement output judgment result B y (k)=0 indicates that a deception attack has occurred, B y (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. y (k); when H y (k) y At that time, set B y (k)=0; when H y (k)≥A y At that time, set B y (k)=1; then measure the output and judge the result B. y (k) and the estimated output value Y a (k) is sent to the data fusion submodule;​ The state estimation error monitoring submodule obtains the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points from the data processing module, performs data analysis, and provides a judgment on whether a network attacker has launched a deception attack; it uses an interval X... 2 State estimation error covariance detector: Define the state estimation error covariance detection variable H p (k) is: ； For the state estimation error monitoring submodule, the state estimation error covariance detection threshold is set to A. p Set the state estimation error covariance judgment result B p (k)=0 indicates that a deception attack has occurred, B p (k)=1 indicates that the deception attack did not occur. Then calculate H according to the formula above. p (k); when H p (k) p At that time, set B p (k)=0; when H p (k)≥A p At that time, set B p (k)=1; then determine the state estimation error covariance result B. p (k) is sent to the data fusion submodule;​ The data fusion submodule receives the differential signal judgment result B from the differential signal monitoring submodule. z (k) and the differential signal estimate Z a (k), receives the measurement output judgment result B from the output signal monitoring submodule. y (k) Receives the state estimation error covariance judgment result B from the state estimation error monitoring submodule. p (k) is used to calculate the differential signal estimate Z through data fusion. r (k); Let B o (k) represents the result of the data fusion submodule's judgment on whether a network attacker has launched a deception attack, B o (k) Calculate using the following formula: B o (k)=B z (k)·B y (k)·B p (k); When B o When (k)=0, the data fusion submodule determines that a network attacker may have launched a spoofing attack, and the data transmitted through the data communication network is no longer reliable, i.e., the differential signal estimate Z provided by the decoder is no longer reliable. a (k) has been tampered with by a network attacker and is no longer reliable; therefore, the differential signal estimation value Z is set. r (k)=0; when B o When (k)=1, the data fusion submodule determines that the network attacker has not launched a deception attack, and the data transmitted through the data communication network is reliable, i.e., the differential signal estimate Z provided by the decoder is reliable. a (k) was not tampered with by a network attacker, therefore the differential signal estimate Z is set. r (k)=Z a (k); The data fusion submodule estimates the differential signal value Z. r (k) is sent to the state estimation module.

4. The unmanned system state observer resistant to deception attacks according to claim 3, characterized in that: In the state estimation module, let Y r (k) represents the estimated measurement output value, which is calculated using the following formula: Y = (k - k) / (k - k) r (k): Y r (k)=Φ(X e (k))+Z r (k); The state equations for the state observer are established as shown in the following formula: X a (k+1)=Ψ(X a (k),U(k))+G(Y r (k)-Φ(X a (k))); Where G represents the observer gain; Solve the above state equations and calculate the state estimate X. a (k); then, the state estimation module obtains the state estimate X. a (k) is sent to the controller of the unmanned system; The controller of the unmanned system obtains a state estimate X from the state estimation module. a (k), calculate the control input value U(k) of the unmanned system, as follows: U(k)=K(X a (k)); Then, the controller of the unmanned system sends the control input value U(k) to the actuator of the unmanned system.

5. A method for state observation of an unmanned system resistant to deception attacks, characterized in that: The unmanned system state observer resistant to deception attacks as described in claim 1 is implemented by including the following steps: Step 1: The data processing module obtains the state estimate X from the state estimation module at the previous moment. a (k-1) and the state estimation error covariance value P(k-1), from which the code symbol estimate C is obtained from the decoder. a (k), calculate the differential signal estimate Z of the measured output value Y(k). a (k), based on the differential signal estimate Z a (k) Further calculate the estimated value of the measurement output Y a (k), for the obtained state estimate X a (k-1), State estimation error covariance P(k-1), Differential signal estimate Z a (k) and the estimated output value Y a (k) is cached and finally provided to the deception attack detection module; Step 2: The deception attack detection module obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a Given (k-n+1) and the state estimation error covariances P(k-1), P(k-2), ..., P(kn) at n times, data analysis is used to determine whether a network spoofing attack has occurred. Based on the determination results, data fusion is performed to update the differential signal estimate Z at the current time. a (k), to obtain the corresponding differential signal estimate Z. r (k), then estimate the differential signal Z r (k) is sent to the state estimation module; Step 3: The state estimation module obtains the differential signal estimate Z from the data processing module. r (k) is used to calculate the state estimate X of the unmanned system. a (k), and the state estimate X a (k) is sent to the controller of the unmanned system, thereby enabling state observation of the unmanned system.

6. The unmanned system state observation method resistant to deception attacks according to claim 5, characterized in that: The specific method for step 1 is as follows: Step 1.1: The differential signal estimation submodule obtains the code symbol estimate C from the decoder. a (k) is decoded, and the differential signal estimate Z is calculated. a (k); then the obtained differential signal estimate Z a (k) is sent to the data buffer submodule for storage, and also to the output signal estimation submodule; Step 1.2: The output signal estimation submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) obtains the previous time-step state estimate X from the data cache submodule. a (k-1), generate the estimated measurement output value Y a (k); based on the obtained state estimate X from the previous time step. a (k-1), the output signal estimation submodule calculates the state prediction value X according to the following formula. e (k): X e (k)=Ψ(X a (k-1),U(k-1)); U(k-1) is calculated using the following formula: U(k-1)=K(X a (k-1)); Where K(·) represents U(k-1) and X a The functional relationship between (k-1) is determined by the specific unmanned system control algorithm; Based on the calculated state prediction value X e (k), the output signal estimation submodule calculates the estimated measurement output value Y corresponding to the measured output value Y(k) according to the following formula. a (k): Y a (k)=Φ(X e (k))+ Z a (k); Then, the calculated measurement output estimate Y a (k) is sent to the data cache submodule; Step 1.3: The data caching submodule obtains the differential signal estimate Z from the differential signal estimation submodule. a (k) Cache the state estimate X from the state estimation module at the previous time step. a (k-1) and the state estimation error covariance value P(k-1) are cached, and the state prediction value X is stored in the cache. e (k) Provided to the output signal estimation submodule, and the estimated measurement output value Y is obtained from the output signal estimation submodule. a (k) is buffered, and then the differential signal estimates Z at n time points in the buffer are obtained. a (k), Z a (k-1), ..., Z a (k-n+1), estimated output values ​​Y at n time points. a (k), Y a (k-1), ..., Y a (k-n+1), and the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points are sent to the deception attack detection module.

7. The unmanned system state observation method resistant to deception attacks according to claim 6, characterized in that: The specific method for step 2 is as follows: Step 2.1: The differential signal monitoring submodule obtains the differential signal estimates Z at n time points from the data processing module. a (k), Z a (k-1), ..., Z a (k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Differential signal detector: Let Φ z Z represents the estimated value of the differential signal. a The covariance of (k) is defined by the differential signal detection variable H. z (k) is: ； For the differential signal monitoring submodule, the differential signal detection threshold is set to A. z Set the differential signal judgment result B z (k)=0 indicates that a deception attack has occurred, B z (k)=1 indicates that a deception attack did not occur; Calculate H using the formula above. z (k); when H z (k) z At that time, set B z (k)=0; when H z (k)≥A z At that time, set B z (k)=1; then determine the result B of the differential signal. z (k) and the differential signal estimate Z a (k) is sent to the data fusion submodule;​ Step 2.2: The output signal monitoring submodule obtains the estimated output values ​​Y at n time points from the data processing module. a (k), Y a (k-1), ..., Y a (k-n+1), and perform data analysis to determine whether a network attacker has launched a deception attack; using an interval X... 2 Measurement output detector: Let Φ y Indicates the estimated value Y of the measurement output. a The covariance of (k) is defined by the output signal detection variable H. y (k) is: ； For the output signal monitoring submodule, the output signal detection threshold is set to A. y Set the measurement output judgment result B y (k)=0 indicates that a deception attack has occurred, B y (k)=1 indicates that a deception attack did not occur; Calculate H using the formula above. y (k); when H y (k) y At that time, set B y (k)=0; when H y (k)≥A y At that time, set B y (k)=1; then measure the output and judge the result B. y (k) and the estimated output value Y a (k) is sent to the data fusion submodule;​ Step 2.3: The state estimation error monitoring submodule obtains the state estimation error covariance values ​​P(k-1), P(k-2), ..., P(kn) at n time points from the data processing module, performs data analysis, and provides a judgment on whether the network attacker has launched a deception attack; using an interval X... 2 State estimation error covariance detector: Define the state estimation error covariance detection variable H p (k) is: ； For the state estimation error monitoring submodule, the state estimation error covariance detection threshold is set to A. p Set the state estimation error covariance judgment result B p (k)=0 indicates that a deception attack has occurred, B p (k)=1 indicates that a deception attack did not occur; Calculate H using the formula above. p (k); when H p (k) p At that time, set B p (k)=0; when H p (k)≥A p At that time, set B p (k)=1; then determine the state estimation error covariance result B. p (k) is sent to the data fusion submodule;​ Step 2.4: The data fusion submodule receives the differential signal judgment result B from the differential signal monitoring submodule. z (k) and the differential signal estimate Z a (k), receives the measurement output judgment result B from the output signal monitoring submodule. y (k) Receives the state estimation error covariance judgment result B from the state estimation error monitoring submodule. p (k) is used to calculate the differential signal estimate Z through data fusion. r (k); Let B o (k) represents the result of the data fusion submodule's judgment on whether a network attacker has launched a deception attack, calculated using the following formula: B o (k)=B z (k)·B y (k)·B p (k); When B o When (k)=0, the data fusion submodule determines that a network attacker may have launched a spoofing attack, and the data transmitted through the data communication network is no longer reliable, i.e., the differential signal estimate Z provided by the decoder is no longer reliable. a (k) has been tampered with by a network attacker and is no longer reliable; therefore, the differential signal estimation value Z is set. r (k)=0; when B o When (k)=1, the data fusion submodule determines that the network attacker has not launched a deception attack, and the data transmitted through the data communication network is reliable, i.e., the differential signal estimate Z provided by the decoder is reliable. a (k) was not tampered with by a network attacker, therefore the differential signal estimate Z is set. r (k)=Z a (k); The data fusion submodule estimates the differential signal value Z. r (k) is sent to the state estimation module.

Images