AI special-purpose internet of things card secure encryption communication method for multi-type robots

Abstract

The application discloses an AI special-purpose Internet of Things card security encryption communication method for multiple types of robots, and belongs to the technical field of secure communication; and is used for solving the technical problem that radio frequency tags are vulnerable to malicious attacks in the prior art, threatening device identity authentication and data integrity; the fingerprint is fused with real-time electromagnetic environment characteristics, and even if an attacker intercepts historical fingerprints, the attacker cannot pass the verification under different electromagnetic environments; the fingerprint is updated in real time based on the motion state and environmental changes, a multi-node alliance chain verification mechanism is implemented, and the global adaptability of the fingerprint fusion is effectively improved while protecting the node privacy through the optimization of the weight matrix by federated learning; the generated final dynamic radio frequency fingerprint is used as the initial value of a chaotic system, a frequency hopping sequence generator is driven to dynamically adjust the communication frequency, meanwhile, the environmental electromagnetic disturbance characteristics are embedded into the data frame as encryption factors to construct a three-in-one encryption transmission link, and deep fusion encryption of the physical layer and the data link layer is realized.

Description

Technical Field

[0001] This invention relates to the field of secure communication technology, specifically to a secure encrypted communication method for AI-dedicated IoT cards for various types of robots. Background Technology

[0002] AI-dedicated IoT SIM cards for secure and encrypted communication for various types of robots refer to a technical system designed to meet the data security needs of various intelligent robots (such as lawnmowers, express sorting robots, and humanoid service robots) in complex application scenarios. This system integrates advanced encryption technology, identity authentication mechanisms, and dedicated communication channels to ensure the confidentiality, integrity, and reliability of data transmission between robots and the cloud, user terminals, and systems. Its core objective is to prevent data from being eavesdropped on, tampered with, or illegally controlled.

[0003] Existing hardware security and encryption modules have defects, making RFID tags vulnerable to malicious attacks, including tampering, forgery, or signal hijacking, which threaten device authentication and data integrity. Existing technical solutions cannot address these defects through targeted supervision and handling, and cannot guarantee the communication security and anti-attack capabilities of robot IoT cards in complex electromagnetic environments. Summary of the Invention

[0004] The purpose of this invention is to provide a secure encrypted communication method for AI-dedicated IoT cards for various types of robots, which solves the technical problem that radio frequency tags are vulnerable to malicious attacks in existing solutions, threatening device identity authentication and data integrity.

[0005] The objective of this invention can be achieved through the following technical solutions: A secure encrypted communication method for AI-dedicated IoT cards for various types of robots includes: Step 1: Collect the nonlinear distortion characteristics of the IoT card RF chip through ultra-wideband spectrum scanning, combine it with the Doppler frequency shift parameters under the robot's motion state, generate dynamic RF fingerprint using generative adversarial network, and anchor and fuse the fingerprint with the real-time environmental electromagnetic spectrum characteristics. Step 2: Use the generated final dynamic RF fingerprint as the initial value of the chaotic system to drive the frequency hopping sequence generator to dynamically adjust the communication frequency. At the same time, embed the environmental electromagnetic disturbance characteristics as encryption factors into the data frame to build a three-in-one encrypted transmission link. Step 3: Based on the encrypted communication link status, predict the attacker's strategy using the Stackelberg game model, and output the optimal combination of interference power and frequency hopping rate in real time. When signal hijacking is detected, trigger the radio frequency fingerprint regeneration mechanism and cut off the abnormal link.

[0006] Preferably, the nonlinear distortion characteristics of the IoT card RF chip are collected, and the nonlinear distortion characteristics are normalized and then spliced ​​into a 128-dimensional feature vector.

[0007] Preferably, an improved deep convolutional generative adversarial network is constructed, which takes the feature vector and the processed Doppler frequency shift parameter vector as input to generate a physically unclonable dynamic radio frequency fingerprint.

[0008] Preferably, when anchoring and fusing fingerprints with real-time environmental electromagnetic spectrum features, the real-time environmental electromagnetic spectrum is collected through the auxiliary receiving antenna of the IoT card, environmental features are extracted and an environmental vector is constructed; the fingerprint vector and the environmental vector are weighted and fused, and the final dynamic radio frequency fingerprint after fusion is hashed using the SHA-3-512 hash function to generate a digest, which is then uploaded to three distributed verification nodes to verify its validity.

[0009] Preferably, the generated final dynamic radio frequency fingerprint is used as the initial value of the chaotic system to drive the improved Lorentz chaotic map to generate a pseudo-random sequence; the system iterates according to a preset iteration period to generate a chaotic sequence; a frequency hopping pattern is generated based on the chaotic sequence to dynamically adjust the communication frequency and achieve physical layer anti-interception.

[0010] Preferably, when real-time acquisition of environmental electromagnetic disturbance characteristics, generation of dynamic encryption factors, and embedding of data frames to achieve disturbance encryption coordination, radio frequency signals are acquired through the auxiliary antenna of the IoT card, disturbance characteristics such as disturbance amplitude, frequency offset, and phase jitter are extracted, and after normalizing all disturbance characteristics, a 16-bit encryption factor is generated by weighted summation.

[0011] Preferably, a dedicated data frame structure containing encryption factors is designed, and the encryption factors are embedded in the physical layer data unit to achieve three-in-one encryption.

[0012] Preferably, when predicting the attacker's strategy using the Stackelberg game model, the defender's strategy is set to (P,R), where P is the jamming power emitted by the defender and R is the frequency hopping rate; the attacker maximizes... Solving for the optimal interference power , Utility function for attackers: ; V represents the weight of the data intercepted by the attacker. Where Q is the attacker's interference power, and Q is the signal interception probability; For attacker power cost coefficient; Optimal interference power Substitute into the defender utility function Solve for maximization of : ; k is the communication success rate weighting coefficient; k is the attacker interference efficiency coefficient. The power cost coefficient for the defender; The optimal defense strategy is output by numerically solving the problem using the gradient ascent method. ; This is the optimal frequency hopping rate.

[0013] Preferably, based on the solved optimal defense strategy It dynamically adjusts communication link parameters and monitors link status in real time to detect signal hijacking behavior.

[0014] Preferably, the signal is determined based on multi-dimensional hijacking detection indicators. When any hijacking detection indicator is triggered, the radio frequency fingerprint regeneration response, abnormal link disconnection and reconstruction response are executed immediately.

[0015] Compared to existing solutions, the beneficial effects achieved by this invention are: This invention fuses fingerprints with real-time electromagnetic environment features. Even if an attacker intercepts historical fingerprints, they will be unable to pass verification under different electromagnetic environments, effectively resisting signal hijacking and replay attacks. Real-time fingerprint updates based on motion state and environmental changes solve the problem of poor adaptability of traditional static fingerprints in robot movement scenarios, ensuring the continuity of identity authentication in high-speed motion or complex electromagnetic environments, and effectively improving dynamic adaptive robustness. By implementing a multi-node consortium blockchain verification mechanism, the risk of single-point failure can be avoided. Federated learning optimizes the weight matrix, protecting node privacy while effectively improving the global adaptability of fingerprint fusion.

[0016] This invention uses the generated final dynamic radio frequency fingerprint as the initial value of the chaotic system to drive the frequency hopping sequence generator to dynamically adjust the communication frequency. At the same time, it embeds the environmental electromagnetic disturbance characteristics as encryption factors into the data frame to construct a three-in-one encrypted transmission link, realizing deep integration encryption of the physical layer and the data link layer. This can ensure the communication security and anti-attack capability of the robot's IoT card in complex electromagnetic environments.

[0017] This invention predicts the attacker's strategy based on the encrypted communication link status using a Stackelberg game model, and outputs the optimal combination of interference power and frequency hopping rate in real time. When signal hijacking is detected, it triggers an RF fingerprint regeneration mechanism and cuts off the abnormal link, achieving full-link anti-hijacking protection of prediction, adjustment, and response. It solves the problem that traditional static encrypted communication is easily hijacked by signals, and provides dynamic security for IoT card communication of various types of robots. Attached Figure Description

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart of a secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, this invention is a secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots, comprising: Step 1: Acquire the nonlinear distortion characteristics of the IoT card's RF chip through ultra-wideband spectrum scanning, combine it with the Doppler frequency shift parameters under robot motion conditions, generate a dynamic RF fingerprint using a generative adversarial network, and then anchor and fuse the fingerprint with the real-time environmental electromagnetic spectrum characteristics; the specific steps include: When collecting the nonlinear distortion characteristics of the IoT card RF chip through ultra-wideband spectrum scanning, the built-in ultra-wideband RF module of the IoT card is controlled to perform spectrum scanning with a bandwidth of 500MHz and a sampling interval of 1MHz, covering the output signal of the IoT card RF chip at different transmit powers; the different transmit powers are specifically -40dBm to 20dBm, with a step size of 5dBm. The acquired signal was subjected to wavelet transform using the db4 wavelet basis and a decomposition layer of 5, and nonlinear distortion features including third-order intermodulation distortion (IMD3), harmonic power ratio (HPR), and noise power spectral density (PSD) were extracted. Among them, third-order intermodulation distortion IMD3: calculates the power ratio of the fundamental frequency point at frequency 2f1-f2 to the fundamental power, in dBc; Harmonic Power Ratio (HPR): Calculates the power ratio of the second harmonic 2f0 to the fundamental frequency f0, in dB. Noise Power Spectral Density (PSD): The average noise power over a 1MHz bandwidth, expressed in dBm / Hz. The nonlinear distortion features are normalized and then concatenated into a 128-dimensional feature vector. ; When estimating the Doppler frequency shift parameter in real time for the robot's motion state, motion data is collected based on the IMU sensor on the robot. Combined with the time difference of arrival of the wireless signal, the real-time motion velocity v of the robot relative to the communication base station is estimated by extended Kalman filtering. The Doppler frequency shift parameter is then calculated according to the following formula: ;in, For Doppler frequency shift; The angle between the direction of motion and the direction of signal propagation; The carrier frequency is the center frequency of the IoT card's operating frequency band; c is the speed of light, with a value of 3 × 10⁻⁶. 8 ; This represents the real-time motion velocity amplitude. , These represent the real-time motion velocity components of the robot along the x-axis, y-axis, and z-axis in the geodetic coordinate system. The calculated Doppler frequency shift Integrating with real-time motion velocity and direction information, a time-stamped Doppler frequency shift parameter vector is generated. t is a time point; When generating dynamic RF fingerprints using generative adversarial networks (GANs), an improved deep convolutional GAN ​​is constructed to integrate the feature vectors. and Doppler frequency shift parameter vector As input, a physically unclonable dynamic radio frequency fingerprint is generated; The deep convolutional generative adversarial network structure includes a generator and a discriminator. The generator uses a 4-layer convolutional neural network, specifically Conv2D, BatchNorm, and LeakyReLU. The input dimension is 133, and the input is a 128-dimensional feature vector. With 5-dimensional The sum is obtained, and a 1024-bit binary fingerprint vector is output. ; The discriminator, constrained by the Physically No-Clone Function (PUF) principle, verifies the fingerprint vector through a 3-layer convolutional network. The loss function is designed to be tied to the hardware characteristics of the RF chip as follows: ;in, This is the loss weighting coefficient, with a default value of 0.7; Losses are classified based on whether the fingerprints are genuine or forged; The loss is the matching loss between fingerprints and hardware features; During training, the generator parameters were optimized through 5000 rounds of adversarial training to ensure the fingerprint vector... It meets the physical non-cloning requirement that the fingerprint similarity of the same chip is >95% and the fingerprint similarity of different chips is <5%. It should be noted that by generating fingerprints through hardware and dynamic two-factor methods using nonlinear distortion of the radio frequency chip and motion Doppler frequency shift, attackers cannot copy or tamper with hardware-level features through software. This can eliminate the risk of static label forgery from the source and improve the physical layer's anti-forgery capability.

[0022] When fingerprints are anchored and fused with real-time environmental electromagnetic spectrum features, the real-time environmental electromagnetic spectrum is collected through the auxiliary receiving antenna of the IoT card, environmental features are extracted, and an environmental vector is constructed. ; Among them, environmental characteristics include 5G NR band interference power spectral density (PSD), industrial electromagnetic noise entropy (H), and multipath fading coefficient (a). 5G NR band (3.5GHz) interference power spectral density (PSD): resolution 1MHz, unit dBm / Hz; Industrial electromagnetic noise entropy value H: calculated based on Shannon entropy, with a value range of 0-1; Multipath fading coefficient a: estimated based on the Rayleigh fading model, with a value range of 2-6, corresponding to urban and suburban scenarios respectively; fingerprint vector Weighted fusion with the environment vector ENV(t): ;in, The final dynamic radio frequency fingerprint after fusing environmental features at time t is the core credential used for robot identity authentication; To achieve non-linear coupling between fingerprint and environmental features through bitwise XOR operation; The dynamic weight matrix is ​​32×32, with elements ranging from 0.1 to 0.9. It dynamically adjusts with environmental complexity; stronger electromagnetic interference results in higher weights for environmental features. The dynamic weight matrix is ​​generated through federated learning and updated by encrypted gradient aggregation from five or more robot nodes. Local nodes calculate gradients based on historical fingerprint verification pass rates and environmental stability. The global weight matrix is ​​updated after encrypted aggregation, with the aggregated weights positively correlated with the node's historical fingerprint credibility. The credibility threshold is set to 0.85. The implementation of the dynamic weight matrix follows existing conventional techniques; specific implementation steps are not detailed here. It should be noted that by fusing fingerprints with real-time electromagnetic environment characteristics, attackers will be unable to pass verification even if they intercept historical fingerprints, under different electromagnetic environments, which can effectively resist signal hijacking and replay attacks.

[0023] The final dynamic radio frequency fingerprint after fusion A digest is generated using the SHA-3-512 hash function and uploaded to three distributed verification nodes. The nodes verify the validity using the following rules: If the cosine similarity between the fingerprint digest and the locally stored historical fingerprint database is ≥0.85, and the deviation between the environment vector ENV(t) and the node-aware environment features is ≤±10%, then the verification is considered successful. In all other cases, the verification is deemed unsuccessful. If the verification pass rate is ≥90%, the fingerprint is confirmed to be valid; otherwise, the regeneration process is triggered. In addition, set dynamic update trigger conditions: The cosine similarity between the environment vector ENV(t) and the environment vector ENV(t-1) at the previous time step is <0.7; In addition, the robot's real-time speed relative to the communication base station is v > 5 m / s.

[0024] It should be noted that updating fingerprints in real time based on motion state and environmental changes can solve the problem of poor adaptability of traditional static fingerprints in robot mobile scenarios, ensure the continuity of identity authentication in high-speed motion or complex electromagnetic environments, and effectively improve dynamic adaptive robustness.

[0025] By implementing a multi-node consortium blockchain verification mechanism, the risk of single point of failure can be avoided. Federated learning optimizes the weight matrix, which can effectively improve the global adaptability of fingerprint fusion while protecting node privacy.

[0026] Step 2: Use the generated final dynamic RF fingerprint as the initial value of the chaotic system to drive the frequency hopping sequence generator to dynamically adjust the communication frequency. Simultaneously, embed environmental electromagnetic disturbance characteristics as encryption factors into the data frame to construct a three-in-one encrypted transmission link. Specific steps include: The generated final dynamic radio frequency fingerprint As the initial value for the chaotic system, the improved Lorentz chaotic map is used to generate a pseudo-random sequence. The iterative formula for the chaotic system is: ;in, Let be the state value of the chaotic system at time t, and the initial value. hash() is a hash function, which hashes the radio frequency fingerprint and then takes the modulo to ensure that the initial value is in the range [0,1). For system parameters, by The hash value is dynamically modulated. The value ranges from 3.57 to 4.0 to ensure chaotic characteristics; This is the disturbance feedback factor, with a default value of 0.15; For environmental disturbance characteristic components, from the environmental vector Extract the noise entropy value H(t) and normalize it to [0,1]. Iterates every 10ms to generate a chaotic sequence {x(t)}, which serves as the driving source for subsequent frequency hopping sequences; Frequency hopping patterns are generated based on chaotic sequences {x(t)} to dynamically adjust communication frequencies, achieving physical layer anti-interception by pre-setting a set of dedicated frequency bands for industrial IoT. N is the number of frequency hopping channels, which is 64. The frequency band range is 433-434MHz, which is an unlicensed ISM band. The channel spacing is 25kHz to avoid frequency band overlap interference. It should be explained that frequency hopping patterns refer to a sequence in which the carrier frequency periodically jumps between multiple frequency bands according to a preset rule during communication. Its core is to achieve physical layer anti-interception and anti-interference through pseudo-random changes in frequency. In chaotic frequency hopping systems, the frequency hopping pattern is driven by a chaotic sequence, which has the characteristics of initial value sensitivity and long-term unpredictability, and can effectively resist the risk of cracking traditional pseudo-random sequences. The generation of frequency hopping patterns is an existing conventional technical solution, and the specific implementation steps will not be elaborated here. The output x(t) of the chaotic system is mapped to a frequency index using the following formula: Where k(t) is the frequency index at time t, corresponding to the frequency f(t) = F[k(t)]; F is a preset set of dedicated frequency bands for industrial IoT. The number of channels in the preset frequency set; When controlling the frequency hopping rate through the frequency hopping interval, the frequency hopping interval is used. Dynamically adjusted based on the intensity of environmental disturbance: Where PSD(t) is the interference power spectral density at time t, with a value ranging from [-120, -60]. The larger the value, the stronger the interference; when the interference is strong, the frequency hopping interval should be shortened. Frequency hopping interval The unit is ms; It should be noted that the chaotic sequence based on dynamic radio frequency fingerprint initialization is non-periodic and sensitive to initial conditions. Attackers cannot predict the frequency hopping pattern, which can significantly reduce the probability of signal interception.

[0027] Real-time acquisition of environmental electromagnetic disturbance characteristics, generation of dynamic encryption factors, and embedding of data frames to achieve disturbance encryption coordination, while acquiring radio frequency signals through the auxiliary antenna of the IoT card to extract the disturbance amplitude. Frequency offset Phase jitter The disturbance characteristics; among them, Disturbance amplitude Peak-to-peak value of ambient noise, in mV, ranging from 0 to 500 mV; Frequency offset The instantaneous drift of the carrier frequency, in Hz, ranging from -100 to 100 Hz; Phase jitter This represents the random fluctuation of the signal phase, measured in rad, ranging from -0.1 to 0.1. After normalizing all perturbation features, a 16-bit encryption factor is generated through weighted summation. : ; where 65535 is the maximum value of a 16-bit unsigned integer; It should be noted that by embedding real-time electromagnetic disturbance characteristics into the data frame, the encryption factor is strongly bound to the physical environment. Even if the frequency hopping sequence is leaked, attackers will be unable to decrypt the data because they cannot reproduce the same environmental disturbance, thus achieving enhanced encryption based on environmental disturbance.

[0028] Design a dedicated data frame structure that includes encryption factors, and store the encryption factors... Embedding physical layer data units enables fingerprint-frequency hopping-disturbance three-in-one encryption; The data frame format includes the fields SYNC (synchronization header), K (encryption factor), DATA (payload), and CRC (checksum), with corresponding lengths of 4, 2, 1-255, and 2, respectively. Encryption factor When physical layer embedding is achieved through orthogonal amplitude modulation constellation diagram offset, the transmitting end performs 16-QAM modulation on the payload DATA field, and the constellation point coordinates (I,Q) are determined according to the encryption factor. Calculate offset , The adjusted constellation points are Wherein, the constellation point coordinates (I,Q) are the signal point coordinates that map 4 bits of binary data onto a two-dimensional plane in 16-QAM modulation; I is the in-phase component, which is the real part; Q is the quadrature component, which is the imaginary part; I and Q together constitute a complex signal representation. The receiving end extracts the constellation point offset and uses it to deduce the encryption factor. Used for decryption and frequency hopping synchronization verification; In addition, the receiving end uses locally generated chaotic sequences and environmental characteristics to achieve frequency hopping synchronization and encryption factor decryption, thereby verifying link security. During frequency hopping synchronization, the receiver uses the same final dynamic radio frequency fingerprint as the transmitter. Initialize the chaotic system and generate a local chaotic sequence. The frequency hopping pattern of the received signal is compared with the frequency hopping pattern. When three consecutive frequency hopping points match, the synchronization is confirmed to be successful. It needs to be explained that a frequency hopping point refers to the specific carrier frequency used by the communication link at a certain moment or within a certain frequency hopping period in frequency hopping communication; it is the basic building block of the frequency hopping pattern. Frequency hopping synchronization refers to the receiving end aligning its locally generated frequency hopping sequence with the transmitting end's frequency hopping pattern to ensure correct reception of data at each frequency hopping point.

[0029] During encryption factor verification, the receiving end extracts the encryption factor from the data frame. The perturbation encryption factor is calculated based on the environmental perturbation characteristics collected locally in real time. If compared, If the condition is met, it is determined that the device has not been hijacked; otherwise, a link reset is triggered. It should be noted that by using a three-pronged approach of unique RF fingerprinting, dynamic and unpredictable chaotic frequency hopping, and scene-related environmental disturbances to form a multi-layered security barrier, the problem of traditional static encryption being easily tampered with and forged can be solved. This approach is suitable for the high-security communication needs of industrial robots in multi-interference scenarios.

[0030] In this embodiment of the invention, the generated final dynamic radio frequency fingerprint is used as the initial value of the chaotic system to drive the frequency hopping sequence generator to dynamically adjust the communication frequency. At the same time, the environmental electromagnetic disturbance characteristics are embedded as encryption factors into the data frame to construct a three-in-one encrypted transmission link. This achieves deep integration encryption of the physical layer and the data link layer, which can ensure the communication security and anti-attack capability of the robot IoT card in complex electromagnetic environments.

[0031] Step 3: Based on the encrypted communication link status, predict the attacker's strategy using a Stackelberg game model, and output the optimal combination of interference power and frequency hopping rate in real time. When signal hijacking is detected, trigger the RF fingerprint regeneration mechanism and disconnect the abnormal link. Specific steps include: When constructing the Stackelberg game model and initializing its parameters, the encryption link state and encryption factor verification results are used. Construct a Stackelberg game model, with the robot as the defender, the attacker as the follower, and the strategy space and payoff function defined for both sides as follows: Regarding the policy space: Defender strategy is Where P is the jamming power emitted by the defender, and R is the frequency hopping rate; The attacker's strategy is ,in, The attacker's interference power is represented by Q, which is the signal interception probability, ranging from 0 to 1 and determined by the interception capability of the attacker's device. Payment function definition: The defender's utility function is: ;in, is the communication success rate weighting coefficient, with a value range of 0.8-0.9; k is the attacker interference efficiency coefficient, with a value range of 0.1-0.3; The power cost coefficient for the defender ranges from 0.05 to 0.1 and can be set according to the robot's battery capacity. The attacker's utility function is: Where V is the weight of the data intercepted by the attacker, with a value ranging from 0.5 to 1.0; The attacker's power cost coefficient, with a value ranging from 0.02 to 0.05, can be obtained based on the energy consumption model of typical attack devices; When predicting the attacker's strategy using the Stackelberg game model, the defender's strategy is set to (P,R). The attacker aims to maximize... Solving for the optimal interference power : ; ; Optimal interference power Substitute into the defender utility function Solve for maximization of : ; The optimal defense strategy is output by numerically solving the problem using the gradient ascent method. The iteration step size is 0.1, and the convergence condition is... ; The optimal frequency hopping rate; It should be noted that by using the Stackelberg game model to predict the attacker's strategy in real time and outputting the optimal combination of interference power and frequency hopping rate, the defense strategy can be dynamically adjusted according to the attacker's behavior, which can effectively improve the success rate of anti-hijacking compared to static encryption schemes. Furthermore, the parameters of the Stackelberg game model directly utilize the chaotic frequency hopping link states obtained in the preprocessing, such as the frequency hopping rate R and the encryption factor verification results. This ensures that strategy adjustments are based on the real-time communication environment.

[0032] Based on the optimal defense strategy found It dynamically adjusts communication link parameters and monitors link status in real time to detect signal hijacking behavior, including interference power adjustment and frequency hopping rate adjustment; When adjusting the interference power, the transmit power is adjusted from the current P to [a higher value] via the power amplifier of the IoT card. ; When adjusting the frequency hopping rate, update the iteration period of the chaotic frequency hopping sequence generator so that the frequency hopping rate switches from R to... ;For example, =30 hops / second, then the frequency hopping interval ; The signal is determined based on multi-dimensional hijacking detection indicators, which include the number of encryption factor verification failures, frequency hopping synchronization mismatch rate, and constellation diagram offset anomaly. Among them, the number of failed encryption factor verifications was 3 consecutive times. ; Frequency hopping synchronization mismatch rate: The number of frequency hopping point matching errors is greater than 3 times within 1 second; among which, a frequency error greater than 5kHz is judged as a matching error; Constellation chart offset anomaly: QAM constellation point offset ; It should be noted that integrating multi-dimensional hijacking detection indicators such as encryption factor verification, frequency hopping synchronization mismatch, and constellation graph offset can effectively reduce the false detection rate and avoid link interruption caused by misjudgment of a single indicator.

[0033] When any hijacking detection indicator is triggered, the radio frequency fingerprint regeneration response, abnormal link disconnection and reconstruction response are executed immediately. Specifically, when executing the RF fingerprint regeneration response, the dynamic RF fingerprint generation process in step 1 is called and re-executed to generate a new final dynamic RF fingerprint. It is verified through distributed nodes and becomes effective after the verification pass rate is ≥90%. When executing the abnormal link disconnection and reconstruction response, the radio frequency switch of the control IoT card is turned off, the shutdown time is <10ms, the abnormal frequency band signal transmission is stopped, and the current communication link is disconnected at the hardware level. And, using the regenerated final dynamic radio frequency fingerprint The chaotic frequency hopping system in step 2 is reinitialized, the initial value x(0) is updated, a new frequency hopping sequence is generated, and the system is switched to the set of backup frequency bands to avoid continuous interference from the original frequency bands and realize the reconstruction of the encrypted link based on the new fingerprint.

[0034] It should be noted that by executing RF fingerprint regeneration response, abnormal link cutoff and reconstruction response, a closed-loop defense mechanism of detection, regeneration and reconstruction is formed, which fundamentally undermines the attacker's continuous interception attempts and ensures that the robot can quickly restore secure communication after being hijacked. It is suitable for high real-time control scenarios of industrial robots.

[0035] In this embodiment of the invention, the above steps achieve end-to-end anti-hijacking protection of prediction, adjustment and response, solve the problem that traditional static encrypted communication is easily hijacked by signals, and provide dynamic security for IoT card communication of various types of robots.

[0036] In the several embodiments provided by this invention, it should be understood that the disclosed system can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for example, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0037] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0038] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in the form of hardware plus software functional modules.

[0039] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the essential characteristics of the present invention.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots, characterized in that, include: Step 1: Collect the nonlinear distortion characteristics of the IoT card RF chip through ultra-wideband spectrum scanning, combine it with the Doppler frequency shift parameters under the robot's motion state, generate a dynamic RF fingerprint using a generative adversarial network, and then anchor and fuse the dynamic RF fingerprint with the real-time environmental electromagnetic spectrum characteristics to obtain the final dynamic RF fingerprint. Specifically, when anchoring and fusing fingerprints with real-time environmental electromagnetic spectrum features, the real-time environmental electromagnetic spectrum is collected through the auxiliary receiving antenna of the IoT card, environmental features are extracted and an environmental vector is constructed; the fingerprint vector and the environmental vector are weighted and fused, and the final dynamic radio frequency fingerprint after fusion is hashed using the SHA-3-512 hash function to generate a digest, which is then uploaded to three distributed verification nodes to verify its validity; Step 2: Use the final dynamic RF fingerprint as the initial value of the chaotic system to drive the frequency hopping sequence generator to dynamically adjust the communication frequency. At the same time, embed the environmental electromagnetic disturbance characteristics as encryption factors into the data frame to build an encrypted transmission link. Step 3: Based on the encrypted communication link status, predict the attacker's strategy using the Stackelberg game model, and output the optimal combination of interference power and frequency hopping rate in real time. When signal hijacking is detected, trigger the radio frequency fingerprint regeneration mechanism and cut off the abnormal link.

2. The secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 1, characterized in that, The nonlinear distortion characteristics of the IoT card RF chip are collected, and the nonlinear distortion characteristics are normalized and then concatenated into a 128-dimensional feature vector.

3. The secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 2, characterized in that, An improved deep convolutional generative adversarial network is constructed, which takes the feature vector and the processed Doppler frequency shift parameter vector as input to generate physically unclonable dynamic radio frequency fingerprints; the deep convolutional generative adversarial network structure includes a generator and a discriminator. The generator uses a 4-layer convolutional neural network; The discriminator, constrained by the Physically Unclonable Function (PUF) principle, verifies the binding relationship between the fingerprint vector and the hardware characteristics of the RF chip through a 3-layer convolutional network.

4. The secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 1, characterized in that, The final dynamic radio frequency fingerprint As the initial value for the chaotic system, the improved Lorentz chaotic map is used to generate a pseudo-random sequence. The iterative formula for the chaotic system is: ;in, Let be the state value of the chaotic system at time t, and the initial value. hash() is a hash function, which hashes the radio frequency fingerprint and then takes the modulo to ensure that the initial value is in the range [0,1). For system parameters, by The hash value is dynamically modulated. The value range is 3.57-4.0; For disturbance feedback factors; For environmental disturbance characteristic components, from the environmental vector Extract the noise entropy value H(t) and normalize it to [0,1]. Iterate according to the preset iteration period to generate a chaotic sequence. Generate a frequency hopping pattern based on the chaotic sequence, dynamically adjust the communication frequency, and realize physical layer anti-interception.

5. A secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 4, characterized in that, Real-time acquisition of environmental electromagnetic disturbance characteristics, generation of dynamic encryption factors, and embedding of data frames to achieve disturbance encryption collaboration. The auxiliary antenna of the IoT card is used to acquire radio frequency signals, extract disturbance characteristics such as disturbance amplitude, frequency offset, and phase jitter, normalize all disturbance characteristics, and generate a 16-bit encryption factor by weighted summation.

6. A secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 5, characterized in that, Design a dedicated data frame structure that includes encryption factors, embedding the encryption factors into physical layer data units to achieve encryption.

7. A secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 6, characterized in that, When predicting the attacker's strategy using the Stackelberg game model, the defender's strategy is set to (P,R). The attacker aims to maximize... Solving for the optimal interference power , Utility function for attackers: ； ； V represents the weight of the data intercepted by the attacker; P represents the jamming power emitted by the defender; and R represents the frequency hopping rate. Where Q is the attacker's interference power, and Q is the signal interception probability; Optimal interference power Substitute into the defender utility function Solve for maximization of : ; k is the communication success rate weighting coefficient; k is the attacker interference efficiency coefficient. The power cost coefficient for the defender; For attacker power cost coefficient; The optimal defense strategy is output by numerically solving the problem using the gradient ascent method. ; To achieve the optimal transmission power, This is the optimal frequency hopping rate.

8. A secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 7, characterized in that, Based on the optimal defense strategy found It dynamically adjusts communication link parameters and monitors link status in real time to detect signal hijacking behavior.

9. A secure encrypted communication method for AI-dedicated IoT cards for multiple types of robots according to claim 8, characterized in that, The system determines whether a signal has been hijacked based on multi-dimensional hijacking detection indicators. When any hijacking detection indicator is triggered, the system immediately executes the radio frequency fingerprint regeneration response, abnormal link disconnection and reconstruction response. The multi-dimensional hijacking detection indicators include the number of encryption factor verification failures, frequency hopping synchronization mismatch rate and constellation diagram offset anomaly.

Images