Multi-unmanned vehicle anti-deception attack pre-scheduling time control method based on dimension expansion encryption

Abstract

The application relates to the technical field of multi-unmanned vehicle safety control, and particularly discloses a multi-unmanned vehicle anti-deception attack pre-determined time control method based on dimension expansion encryption, which comprises the following steps: an additional data is added to original state information of a sending end unmanned vehicle to generate dimension expansion information, the dimension expansion information is encrypted by using an encryption matrix, and the encrypted dimension expansion information is sent; a receiving end unmanned vehicle acquires receiving information, judges whether a deception attack exists in a communication link according to the receiving information, and executes a data restoration process according to a judgment result to obtain a real state of the sending end unmanned vehicle. By constructing an active defense mechanism based on dimension expansion encryption and a pre-determined time control architecture, the core problem that a multi-unmanned vehicle system is difficult to balance communication safety and control efficiency under a deception attack environment is effectively solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-unmanned vehicle safety control, and particularly relates to a multi-unmanned vehicle anti-deception attack pre-determined time control method based on dimension expansion encryption. BACKGROUND

[0002] The multi-unmanned vehicle system plays a key role in the fields of logistics transportation and disaster rescue due to its high efficiency of cooperative work, and its distributed control seriously depends on the state information interaction between unmanned vehicles through wireless networks. However, the openness of the wireless communication network makes the system not only face the challenge of limited communication bandwidth, but also be easily exposed to malicious network attacks.

[0003] Although the prior art can realize basic formation control in an ideal communication environment, when the communication link is invaded by an external invader, the receiving end often has difficulty in distinguishing between real data and fake data, resulting in the control system calculating based on the wrong state, and further causing the formation to diverge or even the system to collapse, which seriously threatens the safety of the cooperative task.

[0004] Especially, the deception attack with extremely strong concealment, the attacker misleads the control system by tampering with the transmission data, so that the existing defense strategy is trapped in the conflict dilemma of difficult to balance data security and control timeliness. Traditional defense means often focus on passive robustness, that is, tolerate a certain degree of attack interference, but this usually sacrifices the convergence speed of the system, resulting in that the system can only realize gradual convergence and cannot meet the strict requirements of completing the formation control within a predetermined time in time-sensitive tasks.

[0005] Therefore, how to ensure that the data can be accurately restored under attack while forcing the system to stably converge within the preset time is a key technical problem that needs to be solved in the current field. SUMMARY

[0006] The present application aims to at least solve one of the technical problems in the related art. To this end, the purpose of the present application is to propose a multi-unmanned vehicle anti-deception attack pre-determined time control method based on dimension expansion encryption, so as to improve the survival ability and task execution efficiency of the multi-unmanned vehicle system in a complex hostile environment.

[0007] To achieve the above-mentioned purpose, the first aspect of the present application proposes a multi-unmanned vehicle anti-deception attack pre-determined time control method based on dimension expansion encryption, comprising the following steps:

[0008] The sending end unmanned vehicle adds additional data to the original state information to generate dimension expansion information, and encrypts the dimension expansion information using an encryption matrix and sends it;

[0009] The receiving end unmanned vehicle acquires the received information, determines whether there is a spoofing attack on the communication link based on the received information, and performs a data restoration process based on the determination result to obtain the true state of the sending end unmanned vehicle.

[0010] Based on the actual state, a predetermined time control scheme is constructed using a preset time scale function and a backstepping control strategy to drive the unmanned vehicle to complete the formation control within a predetermined time.

[0011] The process of obtaining the true state includes: using the decryption matrix to perform reverse calculation on the received information to obtain an intermediate estimation vector; comparing the consistency of a specific element in the intermediate estimation vector with the additional data; if they are consistent, it is determined that no attack has occurred, and the true state is directly extracted; if they are inconsistent, it is determined that an attack has occurred, the attack parameters are estimated using the intermediate estimation vector, and the true state is reconstructed.

[0012] In some embodiments of the present invention, the transmitting unmanned vehicle adds additional data to the original state information to generate expanded-dimensional information, including:

[0013] Select two distinct scalar data points as additional data.

[0014] The additional data is appended to the end of the vector of the original state information to form an expanded-dimensional information vector.

[0015] Construct an encryption matrix consisting of four sub-matrices, and multiply the encryption matrix by the expanded-dimensional information vector to obtain the encrypted signal to be transmitted.

[0016] In some embodiments of the present invention, the mathematical model for receiving information is expressed as follows:

[0017] The received information is equal to the encrypted information from the sender multiplied by the multiplicative attack parameter matrix, plus the additive attack vector;

[0018] In this context, both the multiplicative attack parameter matrix and the additive attack vector are unknown variables.

[0019] In some embodiments of the present invention, the specific steps for determining whether a communication link is subject to a spoofing attack include:

[0020] The inverse of the encryption matrix is ​​used as the decryption matrix;

[0021] The intermediate estimation vector is obtained by multiplying the decryption matrix by the received information.

[0022] Check the element values ​​in the intermediate estimation vector corresponding to the additional data positions;

[0023] If the value of the element is equal to the preset additional data, it is determined that there is no deception attack;

[0024] If the value of this element is not equal to the preset additional data, it is determined that there is a spoofing attack on the communication link.

[0025] In some embodiments of the present invention, the step of performing a data restoration process based on the judgment result to obtain the true state of the sending-end unmanned vehicle includes:

[0026] When it is determined that there is no deception attack, the element corresponding to the original state information position in the intermediate estimation vector is directly extracted and used as the real state.

[0027] The actual state includes the location and speed information of the unmanned vehicle at the transmitting end.

[0028] In some embodiments of the present invention, the step of performing a data restoration process based on the judgment result to obtain the true state of the sending-end unmanned vehicle further includes:

[0029] When a deception attack is determined to exist, the estimated values ​​of the multiplicative attack parameters and additive attack parameters injected by the attacker are calculated based on the difference between the intermediate estimation vector and the preset additional data.

[0030] Using the estimated value, the decryption matrix, and the received information, attack components are eliminated through algebraic operations to reconstruct the original expanded-dimensional information;

[0031] The true state of the sending unmanned vehicle is extracted from the reconstructed extended-dimensional information.

[0032] In some embodiments of the present invention, the preset time scale function is a dynamic switching function used to achieve predetermined time control, and its definition satisfies:

[0033] When the time is less than the predetermined convergence time, the function value is an exponential function in the form of the reciprocal of the difference between the predetermined convergence time and the current time;

[0034] The function value is constant when the time is greater than or equal to the predetermined convergence time;

[0035] The time scaling function is used to rescale the system's time variables to ensure that the system error converges to zero within a predetermined convergence time.

[0036] In some embodiments of the present invention, the construction of the predetermined timing control scheme involves defining a distributed error, including:

[0037] Define the location distributed error as a weighted sum of the current position of the autonomous vehicle, the positions of neighboring autonomous vehicles, and the expected distance to the neighboring autonomous vehicles;

[0038] Define the speed distributed error as the weighted sum of the current speed of the autonomous vehicle and the speeds of its neighboring autonomous vehicles;

[0039] The location and speed of the neighboring unmanned vehicles are the actual states of the sending unmanned vehicle obtained by the receiving unmanned vehicle after performing the data restoration process.

[0040] In some embodiments of the present invention, the step of constructing a predetermined time control scheme using a preset time scale function and a backstepping control strategy includes:

[0041] A virtual controller is designed based on the aforementioned location distributed error. The virtual controller includes a combination of linear and nonlinear terms of the location error, and introduces relevant variables of the time scale function as gain adjustment.

[0042] Based on the speed distributed error and the virtual controller, an actual control input is designed, which includes a feedback term for the speed error and a compensation term for offsetting the nonlinear dynamics of the system.

[0043] In some embodiments of the present invention, the construction of the predetermined time control scheme further includes a coordinate transformation step:

[0044] Define the first coordinate transformation variable as the position distributed error;

[0045] The second coordinate transformation variable is defined as the difference between the speed of the unmanned vehicle and the speed of the virtual controller;

[0046] By constructing a Lyapunov function from the first and second coordinate transformation variables, and based on the properties of the time scale function, the conclusion that the system convergence time satisfies the preset conditions is derived.

[0047] To achieve the above objectives, a second aspect of the present invention proposes a pre-set time control system for multi-unmanned vehicles based on extended-dimensional encryption to resist spoofing attacks, comprising:

[0048] The dimension expansion encryption module is configured to add unequal extra data to the original state information of the unmanned vehicle at the transmitting end to expand the dimension, and use the encryption matrix to generate encrypted information.

[0049] The attack detection and reconstruction module is configured to process the received information by the receiving unmanned vehicle using a decryption matrix, determine whether a spoofing attack exists by comparing the extra data components in the decrypted data, and estimate the attack parameters based on the judgment result to restore the true state.

[0050] The predetermined time control module is configured to combine the restored real state with a preset time scale function and backstepping control algorithm to calculate the control input, driving the unmanned vehicle to converge to the desired formation within a predetermined time.

[0051] To achieve the above objectives, a third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory. When the computer program is executed by the processor, it implements the above-described method for controlling the predetermined time of anti-spoofing attacks on multi-unmanned vehicles based on extended-dimensional encryption.

[0052] The multi-unmanned vehicle anti-spoofing attack predetermined time control method based on extended-dimensional encryption in this invention effectively solves the core problem of multi-unmanned vehicle systems struggling to balance communication security and control efficiency in a spoofing attack environment by constructing an active defense mechanism and predetermined time control architecture based on extended-dimensional encryption.

[0053] First, by leveraging the extended dimensionality features and the algebraic properties of a specific encryption matrix, the system can not only accurately detect whether a deception attack exists in the communication link, but also use the decryption residual to estimate the attack parameters and restore the true state without loss, achieving a leap from passive defense to active recovery. Second, by combining a dynamic time scale function and a backstepping control strategy, the initial state's limitation on the convergence time is theoretically eliminated, ensuring that the system's tracking error strictly converges to zero within the user-preset physical time, thereby significantly improving the survivability and task execution efficiency of the multi-unmanned vehicle system in complex adversarial environments. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the pre-set time control method for multi-unmanned vehicles based on extended-dimensional encryption to resist deception attacks, provided by the present invention.

[0055] Figure 2 This is a schematic diagram of the communication connection relationship of the multi-unmanned vehicle system in the multi-unmanned vehicle anti-spoofing attack predetermined time control method based on extended-dimensional encryption provided by the present invention;

[0056] Figure 3 This is a schematic diagram of the estimation error after restoring the attack signal using the extended-dimensional encryption strategy in the multi-unmanned vehicle anti-spoofing attack predetermined time control method based on extended-dimensional encryption provided by the present invention;

[0057] Figure 4 This is a schematic diagram of the motion trajectory of multiple unmanned vehicles on a two-dimensional plane in the multi-unmanned vehicle anti-spoofing attack predetermined time control method based on extended-dimensional encryption provided by the present invention;

[0058] Figure 5This is a schematic diagram showing the change of tracking error of each unmanned vehicle in the horizontal (x) and vertical (y) directions over time in the multi-unmanned vehicle anti-spoofing attack predetermined time control method based on extended-dimensional encryption provided by the present invention.

[0059] Figure 6 This is a schematic diagram illustrating the implementation of the multi-unmanned vehicle anti-spoofing attack predetermined time control system based on extended-dimensional encryption provided by the present invention;

[0060] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0061] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0062] The following description, with reference to the accompanying drawings, outlines a method, system, and electronic device for controlling the predetermined time of multi-unmanned vehicles against deception attacks based on extended-dimensional encryption, according to embodiments of the present invention.

[0063] Example 1,

[0064] like Figure 1 As shown, this embodiment details the implementation process of a pre-time control method for multi-unmanned vehicle anti-spoofing attacks based on extended-dimensional encryption. This embodiment focuses on integrating this method into practical engineering application logic, demonstrating its specific operating mechanism in complex electromagnetic environments and network warfare scenarios through in-depth analysis and expansion of key technical features. The method described in this embodiment is applied to a multi-unmanned vehicle system, which consists of several follower unmanned vehicles and at least one leader unmanned vehicle. The unmanned vehicles are connected and interact with each other via a wireless communication network.

[0065] Phase 1: Information expansion and encryption processing at the sending end.

[0066] This phase is primarily executed by the sending unmanned vehicle. In multi-vehicle collaborative formation control, each unmanned vehicle is both a receiver and a sender of information. When a particular unmanned vehicle acts as a sender, its primary task is to ensure the confidentiality and integrity of its own status information during transmission and to provide the receiver with credentials for proactively detecting attacks.

[0067] First, the transmitting unmanned vehicle acquires its own raw state information through its onboard sensor array. This raw state information typically includes the vehicle's position and velocity vectors in the inertial coordinate system. To protect the transmitted data and establish security verification anchors, the transmitting unmanned vehicle does not directly send this raw state information; instead, it performs a dimension expansion operation.

[0068] For example, the sending end adds two unequal data points to the original transmitted state information, expanding the dimension of the original information. The core of this step is to construct an augmented state vector. The sending end processor allocates a new data space in memory, first writing the original position and velocity states, and then selecting two unequal scalar data points as additional data. These two additional data points can be pre-set fixed security codes or dynamic verification codes generated according to a specific pseudo-random algorithm. The key is that these two values ​​must maintain definite mathematical properties and be unequal within the current communication cycle to prevent singularities in subsequent matrix operations. The sending end concatenates these two additional data points to the end of the original state information vector, thus forming the expanded information vector. At this point, the original state vector with dimension n is expanded into an expanded information vector with dimension n+2.

[0069] Furthermore, to defend against eavesdropping and tampering with plaintext data, and to provide a mathematical basis for subsequent parameter estimation, the sending end uses a matrix of appropriate dimensions to encrypt the expanded information. Specifically, this encryption process involves constructing an encryption matrix consisting of four sub-matrices. This encryption matrix is ​​a block matrix whose dimension matches the dimension of the expanded information vector. The four sub-matrices of this encryption matrix must meet strict mathematical constraints during design: the two sub-matrices located on the diagonal must be invertible matrices to ensure that the encryption process is mathematically reversible, thereby guaranteeing that the receiving end can losslessly restore the data in the absence of attacks; while the two sub-matrices located off-diagonally are matrices of appropriate dimensions, used to linearly mix the original state information with the additional data.

[0070] In practical engineering implementation, the sending end performs matrix multiplication by left-multiplying the encryption matrix by the expanded information vector. This operation deeply couples the original physical state information with additional data, generating an encrypted signal to be transmitted. At this point, the output signal no longer has an intuitive physical meaning; any third party intercepting the signal without the corresponding decryption matrix will be unable to distinguish which part is the state information and which part is the verification data. After encryption, the sending end transmits the encrypted information to the network via a wireless communication module.

[0071] Phase Two: Analysis of Communication Transmission and Attack Models.

[0072] It should be noted that during the transmission of signals from the transmitter to the receiver via a wireless channel, the communication link may be subject to malicious spoofing attacks.

[0073] Specifically, the received information acquired by the receiving autonomous vehicle includes a spoofing attack component. Mathematically, the received information is represented as: the received information equals the encrypted information from the sending end multiplied by the multiplicative attack parameter matrix, plus the additive attack vector. Here, two main forms of spoofing attacks are clearly defined: multiplicative attacks and additive attacks. A multiplicative attack involves the attacker injecting an unknown gain matrix into the communication link. This matrix causes scaling or phase rotation of the transmitted signal, simulating signal attenuation or multipath effects, making it highly deceptive. An additive attack, on the other hand, involves the attacker superimposing an unknown bias vector onto the link, attempting to tamper with the fundamental values ​​of the data.

[0074] For example, both the multiplicative attack parameter matrix and the additive attack vector are unknown variables for the receiving end. This means that the receiving end cannot know in advance when or with what strength the attacker will launch an attack. This uncertainty is the fundamental reason why passive defense strategies in the prior art are ineffective. In this embodiment, however, because the sending end performs dimension-expanding encryption in advance, the receiving end has the ability to reverse-engineer these unknown variables from the contaminated data.

[0075] Phase 3: Attack detection, parameter estimation, and state restoration at the receiving end.

[0076] After receiving the information, the receiving autonomous vehicle enters the core processing flow. The receiving processor first determines whether there is a spoofing attack on the communication link based on the received information, and then executes a data restoration process based on the determination result to obtain the true state of the sending autonomous vehicle.

[0077] Specifically, the steps for determining whether a communication link is susceptible to a spoofing attack are as follows: The receiving end pre-stores a decryption key corresponding to the encryption matrix of the sending end, i.e., using the inverse of the encryption matrix as the decryption matrix. The receiving end processor calculates the product of the decryption matrix and the received information. Mathematically, this step is equivalent to attempting to peel off the encryption layer. If the communication link is not interfered with, this operation should perfectly restore the expanded-dimensional information vector, and the calculated result is called the intermediate estimated vector.

[0078] It's also important to note that the physical meaning of the intermediate estimation vector lies in its representation of the reconstructed data from the receiver's perspective. To verify the data's authenticity, the receiver examines the element values ​​in the intermediate estimation vector corresponding to the positions of the additional data. Since the transmitter places the additional data at the end of the vector during dimensionality expansion, the receiver only needs to read the values ​​of the last two dimensions of the intermediate estimation vector and compare them with the preset additional data.

[0079] For example, if the element value is equal to the preset additional data, it is determined that no spoofing attack exists. This indicates that the signal has not been tampered with during transmission, the multiplicative attack parameter matrix is ​​an identity matrix, and the additive attack vector is a zero vector. In this case, the receiving end executes the following process: when it is determined that no spoofing attack exists, the element corresponding to the original state information position in the intermediate estimation vector is directly extracted as the true state. The true state includes the position and speed information of the sending end's unmanned vehicle. This processing method ensures computational efficiency in a safe environment and avoids unnecessary complex calculations.

[0080] Optionally, if the element value is not equal to the preset additional data, a spoofing attack is determined to exist on the communication link. In this case, the terminal data of the intermediate estimation vector is no longer the preset additional data, but distorted data carrying attack characteristics. The receiving end then activates an adaptive restoration mechanism: when a spoofing attack is determined to exist, the system no longer simply discards the data, but uses the bad data to derive bad parameters.

[0081] Specifically, the receiving end constructs a system of algebraic equations about the attack parameters based on the difference between the intermediate estimated vector and the preset additional data. Since the two additional data points added by the sending end are unequal, this provides the necessary rank condition for solving the system of equations, enabling the receiving end to calculate the estimated values ​​of the multiplicative and additive attack parameters injected by the attacker. This process embodies the defensive strategy of turning passive defense into active defense, deducing the full picture of the attack by analyzing the impact of the attack on the additional data.

[0082] For example, after obtaining the estimated attack parameters, the receiver uses the estimated values, the decryption matrix, and the received information to eliminate the attack components through algebraic operations. The processor constructs a reconstruction formula, subtracting the estimated value of the additive attack from the contaminated received information, then left-multiplying by the inverse matrix of the estimated value of the multiplicative attack matrix, and finally combining it with the decryption matrix to peel away the interference layer by layer and reconstruct the original expanded-dimensional information. Ultimately, the receiver extracts the true state of the sending end's unmanned vehicle from the reconstructed expanded-dimensional information. Through this series of rigorous mathematical operations, no matter how the attacker changes the attack parameters, as long as their attack model conforms to the linear transformation law, the system can recover the true position and velocity information without loss.

[0083] Phase 4: Construction of the scheduled time control scheme.

[0084] After successfully obtaining the real-time status of each neighboring autonomous vehicle, the system enters the control decision-making phase. The core objective of control is to construct a predetermined time control scheme based on the real-time status, using a preset time scale function and a backstepping control strategy, to drive the autonomous vehicles to complete platooning control within a predetermined time.

[0085] First, to achieve strict constraints on the system convergence time, this embodiment introduces a time-scale transformation technique. Specifically, the preset time-scale function is a dynamically switching function used to achieve predetermined time control. This function is defined to satisfy specific piecewise characteristics: when the time is less than the predetermined convergence time, the function value is an exponential function of the reciprocal of the difference between the predetermined convergence time and the current time. This design causes the function value to approach infinity as the time approaches the predetermined convergence time, thereby generating a large gain in the control law and forcing the system error to converge quickly. When the time is greater than or equal to the predetermined convergence time, the function value is a constant to maintain system stability and avoid gain overflow. The time-scale function is used to rescale the system's time variable to ensure that the system error converges to zero within the predetermined convergence time. This method fundamentally solves the problem that traditional asymptotic convergence control cannot meet the time-sensitive requirements of tasks.

[0086] Secondly, in order to quantify the control objective, this embodiment defines the system's error variables. The construction of the predetermined time control scheme involves defining distributed errors, including defining position distributed errors and velocity distributed errors.

[0087] Specifically, the position-distributed error is defined as the weighted sum of the current position of the autonomous vehicle (V2V), the positions of neighboring V2Vs, and the expected distance to neighboring V2Vs. This error variable incorporates the geometric configuration requirements and topological connectivity of the formation. Simultaneously, a velocity-distributed error is defined as the weighted sum of the current speed of the V2V and the speeds of neighboring V2Vs, aiming to achieve speed consistency. It is also important to note that the neighboring V2V positions and speeds used in calculating these errors are the actual states of the sending V2V obtained by the receiving V2V after performing the data restoration process. This again emphasizes the fundamental role of the preceding attack detection and restoration steps for control accuracy; only errors calculated based on cleaned, real data can guide the system to generate correct control responses.

[0088] For example, after obtaining the error definition and time scale function, this embodiment designs the controller. A predetermined time control scheme is constructed using the preset time scale function and backstepping control strategy. Backstepping control is a recursive design method suitable for handling nonlinear systems with strict feedback.

[0089] Specifically, the design process consists of two steps:

[0090] The first step is to design a virtual controller based on the distributed position error. Since the position state is not a direct control input channel, the system first treats velocity as a virtual control quantity for position. The virtual controller includes a combination of linear and nonlinear terms of the position error, and introduces the relevant variables of the time scale function as gain adjustment. The purpose of introducing the time scale function here is to dynamically map the error, which was originally defined in an infinite time domain, to a finite time domain. Through the stretching of the time scale, the virtual control law can drive the position error to converge to zero when the physical time reaches the predetermined moment.

[0091] The second step involves designing the actual control input based on the speed distributed error and the virtual controller. The actual control input is a command, such as thrust or torque, that directly acts on the autonomous vehicle's power system. This actual control input includes a feedback term for the speed error and a compensation term to counteract the system's nonlinear dynamics. In this step, the controller not only eliminates the speed error but also compensates for the dynamics of the virtual controller from the first step, ensuring the stability of the entire closed-loop system.

[0092] In the fifth stage, in order to theoretically and rigorously guarantee the stability and convergence of the control system, this embodiment also includes coordinate transformation and stability proof logic.

[0093] The predetermined time control scheme further includes a coordinate transformation step: defining the first coordinate transformation variable as the position distributed error; and defining the second coordinate transformation variable as the difference between the unmanned vehicle speed and the virtual controller. This coordinate transformation transforms the original system's tracking control problem into a stabilization problem in a new coordinate system.

[0094] Specifically, a Lyapunov function is constructed using the first and second coordinate transformation variables. Based on the properties of the time-scale function, the conclusion that the system convergence time satisfies the preset conditions is derived. The Lyapunov function is typically constructed as a positive definite function of the system energy. By differentiating this function and combining it with the designed control law, it can be proven that its derivative is negative definite in the new time domain after the time-scale transformation. This implies that the system energy will monotonically decrease over time until it reaches an equilibrium point. Due to the introduction of the time-scale function, this decay process is extremely compressed on the physical time axis, thereby ensuring that the system state can strictly converge to the desired formation before the user-set predetermined time.

[0095] In summary, this embodiment fully demonstrates the entire process from low-level signal processing to high-level control decision-making. The extended-dimensional encryption at the transmitting end lays the foundation for system security, while the attack detection and reconstruction at the receiving end eliminates interference from network attacks through ingenious algebraic operations. The backstepping controller based on a time-scale function ensures efficient task completion. This entire set of technical solutions works seamlessly together to form a multi-unmanned vehicle control system capable of resisting complex deception attacks and achieving precise formation within a predetermined time, fully demonstrating the dual advantages of this invention in terms of security and timeliness.

[0096] Example 2:

[0097] This embodiment provides a pre-time control method for multi-unmanned vehicles (UAVs) against deception attacks based on extended-dimensional encryption. The method demonstrates its application scenarios through specific mathematical formulas and calculation processes. This method is mainly applied to multi-UAV platooning control systems, aiming to solve data security and platooning convergence problems under network attacks. Specifically, it includes the following:

[0098] Step S1: The sending end adds two unequal data to the original transmitted status information to expand the dimension of the original information, and at the same time uses a matrix of appropriate dimension to encrypt the expanded information.

[0099] In this embodiment, consider a... A multi-vehicle system consisting of one follower autonomous vehicle and one leader autonomous vehicle. First, establish the first... The dynamic model of an autonomous vehicle has the following physical dynamic equations:

[0100] (1)

[0101] In the formula, represent The derivative, and Representing position and velocity respectively, where and This represents the lateral and longitudinal position of the autonomous vehicle, where and Represents the lateral and longitudinal speeds of the autonomous vehicle. It is the inertia matrix. It is the vector of the Coriolis force and the centripetal force. It is the gravity vector. It is a control input. and This represents the inputs of the autonomous vehicle in the lateral and longitudinal directions.

[0102] To facilitate subsequent control operations, the above system will be converted into:

[0103] (2)

[0104] For ease of description later, let .

[0105] In autonomous vehicle system communication, data transmission and reception are often accomplished via wireless networks. During this process, malicious external attackers may inject carefully prepared data into real information, causing neighboring autonomous vehicles to receive incorrect information and resulting in serious accidents. This situation will significantly impact the stability of the multi-vehicle system, preventing it from completing its intended tasks. Consider the following attack methods:

[0106] (3)

[0107] In the formula, The receiver, the autonomous vehicle, can receive and utilize the information. It is a message sent by the autonomous vehicle. , , and These represent the multiplicative and additive aspects of a deception attack, respectively.

[0108] To ensure that the receiver autonomous vehicle can successfully obtain the true state of the sender autonomous vehicle, the sending end is first processed by adding two additional signals:

[0109] (4)

[0110] In the formula, For the status information of autonomous vehicles, This indicates additional data where two elements are not equal.

[0111] Based on the principles of deception attacks, it is known that external attackers have obtained the information transmitted between autonomous vehicles. This means that there is a potential risk that the information of the autonomous vehicles is transparent to attackers, who may use this information to carry out other attacks.

[0112] Therefore, encrypting the transmitted data is essential, and the encryption process is designed as follows:

[0113] (5)

[0114] In the formula, It is an encryption matrix composed of four smaller matrices. and It is an invertible matrix. and It is a matrix of appropriate dimensions.

[0115] When selecting a matrix, the following conditions should be met:

[0116] (6)

[0117] In the formula, , , , and It is a column vector consisting of 6 and 2 dimensions, all of which are 1.

[0118] After processing the initial signal in the above two steps, a dimension-expanded and encrypted signal is obtained. driverless cars It will send this message to its neighboring driverless cars.

[0119] Step S2: The receiver uses the received information to determine if an attack exists, and takes appropriate action based on different situations to obtain the true state, including:

[0120] Because communication between autonomous vehicles is susceptible to spoofing attacks, autonomous vehicles Message sent driverless car When receiving:

[0121] (7)

[0122] In the formula, , ,make

[0123] (8)

[0124] It should be noted that, although and It contains information added by attackers, but they are already known to them;

[0125] According to mathematical knowledge, the following equations are true;

[0126] (9)

[0127] According to formula (4), It is reversible; the above equation can be rewritten as:

[0128] (10)

[0129] Based on seeking and The results can be divided into two cases;

[0130] Scenario 1: =1 and =0, indicating that the communication link has not been attacked; Case 2: When ≠1 and When the value is not equal to 0, it indicates that an attack exists on the communication link.

[0131] When there is no attack, i.e. =1 and =0, the true state of the system can be obtained using the decryption matrix:

[0132] (11)

[0133] Based on the above equation, the calculated value derived solely from existing data can be obtained. With driverless cars actual state The consistency demonstrates that the designed dimensional expansion encryption strategy has successfully recovered the autonomous vehicle. The transmitted data.

[0134] When an attack occurs, i.e. ≠1 and ≠0 indicates that the encrypted information has been leaked and cannot be directly used for controller design; this is due to parameters injected by the attacker. and The estimation has been successful. Subsequent steps will utilize these identified attack parameters, combined with the decryption matrix and the received contaminated data, to reconstruct the original information, namely:

[0135] (12)

[0136] It can be observed that the values ​​of neighboring agents obtained through the designed algorithm match their true states; therefore, we can conclude that the proposed dimension-extended cryptographic security strategy remains effective even under attack conditions.

[0137] Step S3: Based on the actual state obtained in Step S2, construct a control scheme to obtain the predetermined time using backstepping control, including:

[0138] A communication topology diagram is used to represent the communication between autonomous vehicles. If the autonomous vehicle Can receive driverless cars Information and communication weight ,otherwise If you can receive messages from the leader, then ,otherwise .

[0139] To achieve predetermined time control, this invention designs a time scale function. The details are as follows:

[0140] (13)

[0141] In the formula, , .

[0142] Considering the distributed error of the location of multiple autonomous vehicles, the following is:

[0143] (14)

[0144] In the formula, Indicates driverless car Distributed errors in location, Indicates the location of the driverless car. Indicates driverless car Location, Indicates driverless car (The position of leader) Indicates driverless car and The previous expected position Indicates driverless car and The expected position between them.

[0145] Considering the distributed error of the speed of multiple autonomous vehicles, the following is also considered:

[0146] (15)

[0147] In the formula, Indicates driverless car Distributed error of speed Indicates the speed of the driverless car. Indicates driverless car speed, Indicates driverless car (The leader's) speed.

[0148] Define the following coordinate transformation:

[0149] (16)

[0150] In the formula, It is a virtual controller; next, we will use backstepping design to derive the controller form, as follows:

[0151] Step 1: Consider the following Lyapunov candidate function:

[0152] (17)

[0153] right Differentiation yields:

[0154] (18)

[0155] Design a virtual controller as follows:

[0156] (19)

[0157] In the formula, , , All are positive odd numbers;

[0158] Substituting equation (18) into equation (17), we get:

[0159] (20)

[0160] Step 2: Consider the following Lyapunov candidate function:

[0161] ;(twenty one)

[0162] Differentiating the above equation, we get:

[0163] ;(twenty two)

[0164] Design controller as follows:

[0165] ;(twenty three)

[0166] in Substituting equation (22) into equation (21), we get:

[0167] ;(twenty four)

[0168] Step S4: Using the Lyapunov stability principle, prove that the control scheme designed in step S3 is theoretically effective:

[0169] The Lyapunov function is chosen as follows:

[0170] (25)

[0171] Based on mathematical knowledge and in conjunction with equation (22), for:

[0172] (26)

[0173] in, .

[0174] when Using mathematical knowledge, equation (24) can be solved as follows: (27)

[0175] Select parameters The time should meet ,Right now ,in satisfy , It is a small constant.

[0176] when hour, ;

[0177] when Using mathematical knowledge, we can find that:

[0178] (28)

[0179] By solving equation (26), we can obtain the result from... The transition time to zero is:

[0180] (29)

[0181] In summary, the system convergence time satisfies .

[0182] Step S5: Verify the feasibility of the control scheme designed in step S3 through numerical model simulation.

[0183] The designed controller ensures that the multi-agent system converges within a predetermined time, and that no outliers occur during the control process. To verify the effectiveness of the above control scheme, this embodiment constructs a multi-autonomous vehicle simulation model containing one leader and four followers. Figure 2 In this topology, node 0 represents the leader, nodes 1-4 represent followers, and directed edges represent unidirectional information transmission. This topology indicates that the system operates in a distributed communication environment.

[0184] In the simulation experiment of this invention, consider an autonomous vehicle system containing four followers and one leader, whose topology is shown in Figure 2.

[0185] Specifically, driverless cars ( At that time, as a follower The dynamic equation for the leader (at the time) is: (30)

[0186] make The parameters of the autonomous vehicles selected in the simulation are consistent, that is... ;

[0187] The initial position and speed of the driverless car are:

[0188] , , , , ;

[0189] The expected distance between the follower autonomous vehicle and the leader autonomous vehicle is , , , ;

[0190] Control parameters , , , , .

[0191] In the simulation, the multiplicative attack considered in the deception attack is as follows: Additive attack is The final simulation results are as follows: Figures 3-5 As shown.

[0192] in, Figure 3 This paper demonstrates the estimation error after reconstructing the attack signal using the dimension-extended encryption strategy proposed in this invention. The horizontal axis represents time, and the vertical axis represents the magnitude of the error. From... Figure 3 As can be seen, the estimation error converges rapidly to the order of magnitude within an extremely short time (approximately 0.5 seconds). The level of this demonstrates that the strategy can restore the true state with almost no loss.

[0193] Figure 4 The diagram illustrates the movement trajectories of multiple autonomous vehicles on a two-dimensional plane. Dashed boxes connect the positions of each vehicle at the same moment, clearly showing that the four follower vehicles successfully formed the expected rectangular formation around the leader and maintained the formation as the leader moved, thus smoothly completing the formation task.

[0194] like Figure 5 The figure shows the variation of tracking errors of each unmanned vehicle in the horizontal (x) and vertical (y) directions over time. As can be seen from the figure, all error curves converged to the zero mark before the predetermined time (approximately 4 seconds) and remained stable thereafter. This demonstrates that the error converged to zero in a very short time, indicating that formation was complete at this point. This intuitively verifies the effectiveness of the predetermined time control algorithm in this embodiment, achieving the control effect of reaching the target within a specified time.

[0195] Example 3:

[0196] like Figure 6As shown, this embodiment discloses a pre-set time control system for multi-unmanned vehicle anti-spoofing attacks based on extended-dimensional encryption. This system is a physical embodiment of the aforementioned method embodiments at the hardware and software logic levels, aiming to solve the technical problems in existing technologies where multi-unmanned vehicle systems are highly vulnerable to spoofing attacks in open wireless network environments, and where traditional defense strategies struggle to balance data security and control convergence speed. This system can be deployed in the onboard control computer of each follower and leader unmanned vehicle in a multi-unmanned vehicle platoon. Through modular collaborative work, it achieves a closed-loop process from proactive defense, attack detection, data cleaning to precise pre-set time control.

[0197] Specifically, the multi-unmanned vehicle anti-spoofing attack scheduled time control system based on extended-dimensional encryption mainly includes three core functional modules: extended-dimensional encryption module, attack detection and reconstruction module, and scheduled time control module. Each module operates collaboratively under the scheduling of the onboard processor, and its specific structure and functional logic are described below.

[0198] First is the dimension-expansion encryption module. This module is configured to add unequal extra data to the original state information of the transmitting autonomous vehicle to expand its dimensions, and uses an encryption matrix to generate encrypted information. In practical applications, this module first collects the autonomous vehicle's real-time state data, including position and speed information, through the vehicle's sensor interface. Addressing the issue mentioned in the background technology where attackers use transparent information for targeted injection, the dimension-expansion encryption module implements a proactive defense strategy. It does not directly send the original state data, but instead generates two scalars with unequal values ​​as extra data. These two extra data, equivalent to digital watermarks or verification keys, are appended to the end of the original state vector, thus mapping the low-dimensional original state space to a high-dimensional expanded state space. Subsequently, the module calls the encryption matrix stored in a secure storage area to perform a linear transformation on the expanded information vector. This encryption matrix is ​​specially designed, consisting of four sub-matrices and satisfying a specific rank condition to ensure the reversibility and security of the encryption process. After this step, the original physical state is hidden in a confused algebraic structure, generating an encrypted signal to be transmitted, which is then broadcast to neighboring autonomous vehicles via a wireless communication unit.

[0199] Secondly, there is the attack detection and reconstruction module. This module is configured to process the received information using a decryption matrix at the receiving end of the autonomous vehicle. By comparing the extra data components in the decrypted data, it determines whether a spoofing attack exists and estimates the attack parameters to reconstruct the true state based on the judgment result. When the communication antenna of an autonomous vehicle receives a signal sent by a neighboring autonomous vehicle, the signal may have been tampered with by a malicious attacker along the transmission path, superimposed with an unknown multiplicative attack parameter matrix or additive attack vector. The attack detection and reconstruction module first uses a preset decryption matrix, i.e., the inverse matrix of the aforementioned encryption matrix, to reverse-engineer the received mixed signal to obtain an intermediate estimation vector. The core function of this module is to execute consistency verification logic. It extracts the values ​​corresponding to the extra data positions in the intermediate estimation vector and compares them one by one with the locally pre-stored standard extra data. If the values ​​are completely consistent, the communication link is determined to be secure, and the state component in the intermediate estimation vector is directly extracted as the true state. If the values ​​deviate, the system is determined to have suffered a spoofing attack. At this point, instead of simply discarding data as in traditional techniques, the module uses the bias values ​​to construct observation equations and reverse-engineer estimates of the multiplicative parameters and additive biases injected by the attacker. Subsequently, the module uses these estimates to perform algebraic compensation and correction on the contaminated received signal, eliminating the attack components and thus reconstructing the true position and velocity of the transmitting unmanned vehicle without loss. This mechanism effectively solves the problem mentioned in the background technology that passive defense strategies cannot cope with strong deception attacks, achieving high-precision data restoration.

[0200] Finally, there is the pre-defined time control module. This module is configured to combine the reconstructed real state with a preset time scale function and backstepping control algorithm to calculate the control input, driving the autonomous vehicles to converge to the desired formation within a predetermined time. This module is the decision-making center of the system; it receives the purified neighbor state information from the attack detection and reconstruction module and calculates the distributed cooperative error based on its own state. Addressing the issue mentioned in the background technology that asymptotic convergence control cannot meet the time-sensitive task requirements, this module introduces a time scale function with dynamic switching characteristics. This function can nonlinearly rescale the system's time variables, mapping the infinite convergence time to a finite physical time domain. Based on this time scale, the module internally runs a backstepping control algorithm, progressively designing the virtual control law and the actual control law.

[0201] During the calculation process, the pre-set time control module compensates for the system's nonlinear dynamics and coupling terms in real time, ultimately generating actual control inputs containing thrust and torque commands. These commands are sent to the unmanned vehicle's underlying actuators, such as motor drivers and servos, to drive the unmanned vehicle to adjust its flight attitude and trajectory. Through the adjustment of this module, regardless of the unmanned vehicle's initial position or whether the communication link has been attacked, the formation tracking error of the multi-unmanned vehicle system is forcibly constrained to converge strictly to zero within a user-preset time and remains stable thereafter.

[0202] In summary, the system disclosed in this embodiment, through deep integration of hardware and software, constructs a data security barrier using an extended-dimensional encryption module, achieves accurate perception of attack behavior and data regeneration using an attack detection and reconstruction module, and ensures the timely completion of control tasks using a predetermined time control module. These three modules are interconnected, jointly achieving the beneficial technical effect of ensuring the safe, stable, and rapid formation of multiple unmanned vehicle systems in complex adversarial environments.

[0203] Example 4:

[0204] Corresponding to the above embodiments, the present invention also proposes an electronic device.

[0205] like Figure 7 The diagram shows a structural schematic of an electronic device according to the present invention. The electronic device 100 includes a processor 101 and a memory 103. The processor 101 and the memory 103 are connected, for example, via a bus 102. Optionally, the electronic device 100 may further include a transceiver 104. It should be noted that in practical applications, the transceiver 104 is not limited to one unit, and the structure of this electronic device 100 does not constitute a limitation on the embodiments of the present invention.

[0206] Processor 101 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in connection with this disclosure. Processor 101 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0207] Bus 102 may include a pathway for transmitting information between the aforementioned components. Bus 102 may be a PCI bus or an EISA bus, etc. Bus 102 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0208] The memory 103 stores a computer program corresponding to the pre-time control method for multi-unmanned vehicles based on extended-dimensional encryption to resist deception attacks according to the above embodiments of the present invention. This computer program is executed by the processor 101. The processor 101 executes the computer program stored in the memory 103 to implement the content shown in the aforementioned method embodiments.

[0209] Among them, electronic devices 100 include, but are not limited to: mobile terminals such as laptops and PADs (tablet computers) and fixed terminals such as desktop computers. Figure 7 The electronic device 100 shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0210] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A multi-unmanned vehicle anti-deception attack scheduled time control method based on dimension expansion encryption, characterized in that, The method comprises the following steps: The sending end unmanned vehicle adds additional data to the original state information to generate extended dimension information, wherein two scalar data that are not equal to each other are selected as the additional data; the additional data is spliced to the end of the vector of the original state information to form an extended dimension information vector; the extended dimension information is encrypted by using an encryption matrix and is sent, wherein the encryption matrix composed of four sub-matrices is constructed, and the extended dimension information vector is multiplied by the encryption matrix to obtain an encrypted signal to be sent; The receiving end unmanned vehicle acquires received information, and a mathematical model of the received information is represented as: received information = encrypted information of the sending end * multiplicative attack parameter matrix + additive attack vector; wherein the multiplicative attack parameter matrix and the additive attack vector are unknown variables; whether a fraud attack exists in the communication link is determined according to the received information, and a data restoration process is performed according to the determination result to obtain the real state of the sending end unmanned vehicle; Based on the real state, a preset time scale function and a backstepping control strategy are used to construct a predetermined time control scheme to drive the unmanned vehicle to complete the formation control within a predetermined time; The process of obtaining the real state comprises: An intermediate estimation vector is obtained by performing reverse calculation on the received information by using a decryption matrix, which specifically comprises: using the inverse matrix of the encryption matrix as the decryption matrix; calculating the product of the decryption matrix and the received information to obtain the intermediate estimation vector; The consistency of a specific element in the intermediate estimation vector with the additional data is compared, which specifically comprises: checking the element value in the intermediate estimation vector corresponding to the position of the additional data; If the consistency is correct, it is determined that no attack has occurred, and the real state is directly extracted, which specifically comprises: if the element value is equal to the preset additional data, it is determined that no fraud attack exists; when it is determined that no fraud attack exists, the element corresponding to the position of the original state information in the intermediate estimation vector is directly extracted as the real state; the real state comprises position information and speed information of the sending end unmanned vehicle; If the consistency is incorrect, it is determined that an attack has occurred, attack parameters are estimated by using the intermediate estimation vector, and the real state is reconstructed, wherein the attack parameters are the multiplicative attack parameter matrix and the additive attack vector, which specifically comprises: if the element value is not equal to the preset additional data, it is determined that a fraud attack exists on the communication link; when it is determined that a fraud attack exists, the estimated values of the multiplicative attack parameter and the additive attack parameter injected by the attacker are calculated based on the difference between the intermediate estimation vector and the preset additional data; by using the estimated values, the decryption matrix and the received information, the attack components are eliminated through algebraic operation, and the original extended dimension information is reconstructed; the real state of the sending end unmanned vehicle is extracted from the reconstructed extended dimension information.

2. The method of claim 1, wherein, The preset time scale function is a dynamic switching function used to realize predetermined time control, and the definition satisfies: When the time is less than the predetermined convergence time, the function value is an exponential function of the reciprocal of the difference between the predetermined convergence time and the current time. The function value is constant when time is greater than or equal to the predetermined convergence time; The time scale function is used to rescale the time variable of the system to ensure that the system error converges to zero within the predetermined convergence time.

3. The method of claim 1, wherein, The construction of the predetermined time control scheme involves defining a distributed error, including: Defining a position distributed error, which is a weighted sum of the current position of the unmanned vehicle, the positions of neighboring unmanned vehicles, and the desired distances from the neighboring unmanned vehicles; Defining a speed distributed error, which is a weighted sum of the current speed of the unmanned vehicle and the speeds of neighboring unmanned vehicles; Wherein, the positions of neighboring unmanned vehicles and the speeds of neighboring unmanned vehicles are the real states of the sending end unmanned vehicle obtained after the receiving end unmanned vehicle executes the data restoration process.

4. The method of claim 3, wherein, The construction of the predetermined time control scheme using the preset time scale function and the backstepping control strategy includes: Designing a virtual controller based on the position distributed error, which contains a combination of linear and nonlinear terms of the position error and introduces relevant variables of the time scale function as gain adjustment; Designing an actual control input based on the speed distributed error and the virtual controller, which contains a feedback term of the speed error and a compensation term to offset the nonlinear dynamics of the system.

5. The method of claim 4, wherein, The construction of the predetermined time control scheme also includes a coordinate transformation step: Defining a first coordinate transformation variable as the position distributed error; Defining a second coordinate transformation variable as the difference between the speed of the unmanned vehicle and the virtual controller; By constructing a Lyapunov function for the first and second coordinate transformation variables, and according to the properties of the time scale function, it is concluded that the convergence time of the system satisfies the preset conditions.

Images