An ai electronic file management method and system based on a mobile terminal

Abstract

This application discloses an AI-based electronic record management method and system for mobile terminals, relating to the field of dynamic encryption. The method includes: acquiring the wearer's electromyography (EMG) signals and bone conduction acceleration signals, and determining an inherent time delay threshold and a cross-correlation coefficient threshold; acquiring the current EMG signal envelope and the current acceleration signal envelope in real time through an AI module, and calculating the similarity and time difference between the current EMG signal envelope and the current acceleration signal envelope; generating a trigger pulse when the similarity is greater than the cross-correlation coefficient threshold and the time difference is less than or equal to the inherent time delay threshold; acquiring bone conduction micro-vibration waveforms within the time window of trigger pulse generation, and extracting the instantaneous phase value of the bone conduction micro-vibration waveforms; determining the chaotic state value at the current moment, and performing fragmentation processing and instantaneous encryption on the electronic record data stream to be transmitted to obtain processed electronic record data; and sending the processed electronic record data to the receiving end.

Description

Technical Field

[0001] This application relates to the field of dynamic encryption, and more particularly to an AI-based electronic record management method and system based on a mobile terminal. Background Technology

[0002] With the rapid development of mobile internet and smart terminals, electronic records management has become an important part of the informatization construction of government agencies and enterprises. Especially in high-security scenarios such as classified archives, personnel files, and medical records, mobile terminals, due to their portability and flexibility, have gradually become the primary tools for accessing, approving, and transmitting records. However, electronic records management in mobile environments faces more complex security challenges than in traditional fixed network environments. Particularly when performing record operations offline or in weak network conditions, the security and authenticity verification of data transmission become critical issues that urgently need to be addressed.

[0003] Currently, the security of data transmission for electronic archives on mobile devices primarily relies on traditional encryption technologies. Existing technologies typically employ static key encryption or time-synchronized dynamic token mechanisms to protect the data transmission process. Specifically, static key schemes store a preset encryption key in the mobile terminal and use this key to encrypt the archive data during transmission; while dynamic token schemes generate time-sensitive authentication tokens through periodic handshakes between the server and client to verify user identity and the legitimacy of operations. These schemes can provide a certain level of security in stable network environments, but they still have significant limitations in practical applications.

[0004] When reviewing documents offline or in weak network environments, mobile devices cannot maintain real-time communication with the server, thus failing to update keys or verify token validity in a timely manner. In such cases, attackers can exploit signal interception and replay attacks. This involves intercepting encrypted data streams during network communication and resending them to the server later, forging legitimate document operation requests. Because the encrypted data itself is formally valid, the server has difficulty distinguishing between a genuine user's real-time action and an attacker's replay attack. Furthermore, attackers may also use methods such as Bluetooth hijacking to forward user operation data in real-time without the user's knowledge, thereby gaining unauthorized access to the document system.

[0005] The core flaw in existing encryption technologies lies in their reliance on key confidentiality and real-time communication, rather than effectively verifying the current nature and authenticity of operations. Even with high-strength encryption algorithms, attackers can bypass security mechanisms simply by recording and replaying legitimate encrypted data streams, without needing to crack the key. Therefore, in offline or weak network environments, they are unable to effectively defend against attacks such as signal interception and replay, and man-in-the-middle transparent forwarding, posing serious security risks to the mobile management of highly classified electronic archives. Summary of the Invention

[0006] This application provides an AI-based electronic record management method and system based on a mobile terminal, which effectively addresses the security flaws of traditional static keys and periodic token mechanisms when facing signal interception and replay attacks and man-in-the-middle transparent forwarding attacks.

[0007] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: Firstly, an AI-based electronic record management method based on a mobile terminal is provided. This method is applied to a mobile terminal, which includes an AI module. The method includes: The AI ​​module collects the wearer's electromyography (EMG) and bone conduction acceleration signals, and determines the inherent time delay threshold and cross-correlation coefficient threshold between the EMG and acceleration signals. In response to the trigger signal of electronic record operation, the AI ​​module collects the current electromyographic signal envelope and the current acceleration signal envelope in real time, and calculates the similarity and time difference between the current electromyographic signal envelope and the current acceleration signal envelope; A trigger pulse is generated when the similarity is greater than the cross-correlation threshold and the time difference is less than or equal to the inherent time delay threshold. Within the time window of trigger pulse generation, the AI ​​module acquires bone conduction micro-vibration waveforms and extracts the instantaneous phase value of the bone conduction micro-vibration waveforms. The chaotic state value at the current moment is determined based on the instantaneous phase value, and the electronic archive data stream to be transmitted is fragmented and instantaneously encrypted based on the chaotic state value to obtain the processed electronic archive data. The processed electronic archive data is sent to the receiving end, whereby the receiving end decrypts the processed electronic archive data to obtain fragmented data and reassembles the fragmented data to obtain the electronic archive data.

[0008] In one possible implementation of the first aspect, determining the intrinsic time delay threshold between the electromyographic signal and the acceleration signal includes: Collect the start time points of electromyography (EMG) signals and acceleration signals when the user performs standard movements; Calculate the time difference between the start time of the electromyographic signal and the start time of the acceleration signal in multiple standard movements; Perform statistical analysis on the time difference to determine its mean and standard deviation; The effective range of the time delay is determined based on the mean and standard deviation. Set the median value of the effective range as the inherent time delay threshold.

[0009] In another possible implementation of the first aspect, the current electromyographic signal envelope and the current acceleration signal envelope are acquired in real time via an AI module, including: When a user performs a preset action on a mobile terminal, the potential changes of the arm muscles are monitored in real time, and the envelope curve of the potential changes is extracted as the electromyographic signal envelope. Real-time monitoring of hand movement trajectory, extracting the acceleration change envelope of the hand movement trajectory as the acceleration signal envelope.

[0010] In another possible implementation of the first aspect, determining the chaotic state value at the current moment based on the instantaneous phase value includes: The instantaneous phase value is normalized to a preset range to obtain the normalized phase value; The normalized phase value is used as the perturbation parameter of the chaotic equation; Update the system state of the chaotic equation based on the perturbation parameters; The chaotic equations are calculated iteratively to obtain the chaotic state value at the current moment; Among them, the chaos equation is a nonlinear dynamic equation.

[0011] In another possible implementation of the first aspect, the electronic archive data stream to be transmitted is fragmented and instantaneously encrypted based on chaotic state values ​​to obtain processed electronic archive data, including: The electronic archive data stream is cut into multiple data particles according to a preset particle size; Generate a nonlinear mapping matrix based on chaotic state values; Based on the mapping relationship in the nonlinear mapping matrix, assign a target logical channel number to each data particle; Arrange the data particles in random order according to the target logical channel number; Multiple temporary keys are derived from the chaotic state values, and each temporary key corresponds to a data particle. Each data particle is instantaneously symmetrically encrypted using the corresponding temporary key to obtain the processed electronic archive data.

[0012] In another possible implementation of the first aspect, a nonlinear mapping matrix is ​​generated based on the chaotic state values, including: The chaotic state values ​​are used as seed parameters to input the Lorentz equations and the three-dimensional phase space coordinates are initialized. Numerical integration of the Lorentz equations is performed according to a preset iteration step size to generate a chaotic trajectory sequence. Extract the X and Y coordinate components from the chaotic trajectory sequence to construct a two-dimensional chaotic coordinate pair sequence; Normalize the two-dimensional chaotic coordinate pair sequence by performing a modulo operation, and map the coordinate values ​​to the data granular index space; Based on the normalized coordinate pairs, construct a bijective mapping relationship from the source index to the target index; The bijective mapping relationship is filled into the matrix structure to generate a nonlinear mapping matrix, where the dimension of the nonlinear mapping matrix is ​​dynamically determined according to the number of data particles.

[0013] In another possible implementation of the first aspect, each data particle is instantaneously symmetrically encrypted using a corresponding temporary key to obtain the processed electronic archive data, including: Get the byte stream of the current data particle; Extend the temporary key to the same length as the byte stream to generate a key stream; Perform a byte-by-byte XOR encryption operation on the byte stream and the key stream to obtain encrypted data granules; A timestamp and an integrity check code are appended to the header of the encrypted data particles to form encrypted data units; All encrypted data units are concatenated in the order of the target logical channel number to obtain the processed electronic archive data.

[0014] In another possible implementation of the first aspect, the electronic archive data decrypted at the receiving end is fragmented, and the fragmented data is reassembled to obtain electronic archive data, including: The receiving end acquires the processed electronic archive data and parses it according to the structural characteristics of the encrypted data units to extract each encrypted data unit; Extract the timestamp and integrity check code from each encrypted data unit; Integrity verification is performed based on integrity check codes. When the integrity verification passes, the receiving end reconstructs the same chaotic trajectory sequence using chaotic state values ​​synchronized with the sending end. Based on the chaotic trajectory sequence, a corresponding temporary key set is generated according to the same derivation rule as the sender; Each encrypted data unit is matched with its corresponding temporary key based on the target logical channel number; Perform a byte-by-byte XOR decryption operation on each encrypted data unit using a matching temporary key to restore the original data particles; Based on the chaotic state values, a nonlinear mapping matrix identical to that of the transmitting end is reconstructed to obtain the inverse mapping relationship; Based on the inverse mapping relationship, the disordered data particles are restored to their original logical order; All data particles after being restored to their original order are pieced together and reassembled to obtain complete electronic archive data.

[0015] Secondly, this application provides a mobile terminal, comprising: The memory is configured to store instructions; and The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the aforementioned AI electronic record management method based on a mobile terminal.

[0016] Thirdly, this application provides an AI-based electronic record management system for mobile terminals, comprising: Mobile terminal; The receiving end connects to the mobile terminal.

[0017] The above technical solution firstly collects the wearer's electromyography (EMG) and bone conduction acceleration (BCA) signals synchronously via a mobile terminal. A dual biometric verification benchmark is established using the inherent time delay threshold and cross-correlation coefficient threshold between these two physiological signals. Since EMG signals reflect the electrophysiological activity of muscle contraction, while BCA signals reflect the mechanical response of bone vibration, the temporal relationship between the two is individual-specific and difficult to simulate or replicate by external devices, thus ensuring the authenticity of the operator's identity at the source. When the system responds to the electronic file operation trigger signal, it calculates the similarity and time difference between the current EMG signal envelope and the BCA signal envelope in real time. A trigger pulse is generated only when the similarity is greater than the preset cross-correlation coefficient threshold and the time difference conforms to the inherent delay characteristics. This mechanism ensures that each file operation must be driven by the real user's immediate physiological activity. Even if an attacker intercepts the historical encrypted data stream, they cannot replay it because the replayed data lacks real-time synchronization with the current physiological signals. Within the time window of trigger pulse generation, bone conduction micro-vibration waveforms are further acquired and their instantaneous phase values ​​are extracted. These phase values ​​continuously change with the user's minute movements, breathing, heartbeat, and other physiological activities, exhibiting high randomness and unpredictability. The chaotic state value determined based on this instantaneous phase value generates unique encryption parameters for each data transmission. This ensures that even if the same user operates on the same file at different times, the encrypted data stream will be completely different, fundamentally eliminating the possibility of replay attacks. By applying the chaotic state value to the fragmentation processing and instantaneous encryption of electronic file data streams, dynamic encryption is achieved. Each data fragment carries spatiotemporal characteristics closely bound to the current physiological state. Even if an attacker forwards data in real time via a man-in-the-middle approach, the receiving end will reject the data due to biometric verification failure during decryption and reconstruction because it cannot synchronously acquire and simulate the user's real physiological signal characteristics. This effectively identifies and blocks illegal operations. Therefore, this technical solution does not rely on network connection status and real-time server verification. Even in offline or weak network environments, it can still ensure the currentity and authenticity of each file operation, effectively solving the security defects of traditional static key and periodic token mechanisms when facing signal interception and replay attacks and man-in-the-middle transparent forwarding attacks.

[0018] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0019] Figure 1 A flowchart illustrating an AI-based electronic record management method based on a mobile terminal, provided as an embodiment of this application; Figure 2 This is a schematic diagram illustrating a process for fragmenting and instantaneously encrypting an electronic archive data stream to be transmitted based on chaotic state values, as provided in an embodiment of this application. Figure 3 This is a schematic diagram of the structure of an AI electronic records management system based on a mobile terminal, provided as an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0021] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0022] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0023] Figure 1 The illustration shows a flowchart of an AI-based electronic record management method based on a mobile terminal according to an embodiment of this application. Figure 1 As shown in the figure, this application provides an AI-based electronic record management method based on a mobile terminal, which is applied to a mobile terminal including an AI module. The method may include the following steps.

[0024] S110. Collect the wearer's electromyography (EMG) signals and bone conduction acceleration signals through the AI ​​module, and determine the inherent time delay threshold and cross-correlation coefficient threshold between the EMG signals and the acceleration signals; S120, In response to the electronic record operation trigger signal, the AI ​​module collects the current electromyographic signal envelope and the current acceleration signal envelope in real time, and calculates the similarity and time difference between the current electromyographic signal envelope and the current acceleration signal envelope; S130. When the similarity is greater than the cross-correlation threshold and the time difference is less than or equal to the inherent time delay threshold, a trigger pulse is generated. S140. Within the time window of trigger pulse generation, the AI ​​module acquires bone conduction micro-vibration waveforms and extracts the instantaneous phase value of the bone conduction micro-vibration waveforms. S150. Determine the chaotic state value at the current moment based on the instantaneous phase value, and perform fragmentation processing and instantaneous encryption on the electronic archive data stream to be transmitted based on the chaotic state value to obtain the processed electronic archive data. S160. The processed electronic archive data is sent to the receiving end, wherein the receiving end is used to decrypt the processed electronic archive data to obtain fragmented data, and reassemble the fragmented data to obtain electronic archive data.

[0025] In this embodiment, the mobile terminal can be a smart wearable device, such as a smartwatch or ring, or it can be a robot, a robot dog, a robot cat, etc. This application embodiment does not impose any special restrictions on the specific form of the mobile terminal.

[0026] In addition, this embodiment also includes the following steps: Generate a unique operation identifier for each electronic record operation; Bind the operation identifier to the timestamp of the trigger pulse generation time; The operation identifier and the timestamp are embedded in the data particles; The receiving end verifies the validity of the timestamp. If the difference between the timestamp and the current time exceeds a preset time window, decryption is rejected.

[0027] This embodiment refers to using biometric recognition to achieve secure file management. In this embodiment, the AI ​​module integrated into the mobile terminal is used to review the data collected by the sensor. Specifically, it adopts an edge computing architecture, with the hardware foundation being an embedded neural network processor and the software framework based on TensorFlow Lite, supporting dynamic model loading and INT8 quantization inference. The AI ​​module deploys a signal quality classifier based on a one-dimensional convolutional neural network (1D-CNN) in the biosignal acquisition stage. The network structure includes: an input layer that receives raw electromyographic signals from 1000 sampling points; a first convolutional layer using 64 convolutional kernels with a kernel size of 16 and a stride of 2, combined with batch normalization and ReLU activation; a second convolutional layer using 128 convolutional kernels with a kernel size of 8 and a stride of 2; a global average pooling layer that outputs a 128-dimensional feature vector; and a fully connected layer that outputs three probability distributions: "high quality," "medium quality," and "low quality." The classifier was trained on 5000 labeled samples, using cross-entropy loss and the Adam optimizer. The initial learning rate was 0.001, which decayed by 0.8 every 5 epochs. After 20 epochs of training, the validation set accuracy reached 94.3%. This embodiment only uses high-quality data for the following data processing; medium / low-quality data was discarded or re-collected.

[0028] Specifically, the mobile terminal simultaneously acquires the wearer's electromyography (EMG) and bone conduction acceleration (BCA) signals. EMG signal acquisition is achieved through surface electrodes attached to the forearm flexor muscles. BCA acceleration signals are acquired and transmitted to the AI ​​module via a triaxial MEMS accelerometer fixed to the styloid process of the wrist. To determine the inherent time delay threshold, the user is first required to perform 20 standard fist-clenching movements as calibration samples, each movement lasting 2 seconds with a 3-second rest interval. For each standard movement, a dual-threshold detection algorithm identifies the starting point of the EMG signal. Specifically, the EMG signal amplitude is marked as the EMG initiation point when it exceeds three times the resting root mean square value for 50 ms consecutively. The acceleration signal initiation point is marked as the acceleration initiation point when the first derivative of the triaxial acceleration vector magnitude exceeds a threshold of 0.5 g / s for 30 ms consecutively. The time difference between the EMG initiation point and the acceleration initiation point in the 20 standard movements is calculated to obtain a time difference sequence. Perform statistical analysis on the sequence and calculate the mean. and standard deviation The effective time delay range is determined according to the 3σ criterion. The median value of this range is set as the inherent time delay threshold. This value is typically between 30 and 80 ms, reflecting the physiological delay between neuromuscular conduction and mechanical response.

[0029] The cross-correlation threshold is determined based on the normalized cross-correlation analysis between the electromyographic signal envelope and the acceleration signal envelope during 20 standard repetitions. First, Hilbert transforms are performed on both signals to extract the envelope curves. Then, the cross-correlation function is calculated within a time window of [-200ms, 200ms]. The mean of the cross-correlation peaks minus one standard deviation is taken as the cross-correlation threshold. Typical values ​​are between 0.65 and 0.85. This dual-threshold mechanism based on individual calibration can adapt to the physiological differences of different users, ensuring the accuracy and robustness of subsequent authentication.

[0030] When a user performs electronic document operations on the mobile terminal, such as clicking the document view button or performing a swipe to turn pages, the mobile terminal's operating system generates an electronic document operation trigger signal and activates the data acquisition module. This module continuously acquires raw electromyography (EMG) and acceleration signals within a 500ms time window following the trigger signal.

[0031] For electromyography (EMG) signals, a sliding window method is used to extract the signal envelope. Specifically, the raw EMG signal is rectified using a full-wave rectifier and then smoothed using a Hamming window with a window length of 50 ms and a step size of 10 ms to obtain an envelope curve reflecting the change in muscle contraction intensity over time. The extraction of the acceleration signal envelope first involves calculating the vector magnitude of the triaxial acceleration. Then, a sliding window smoothing algorithm with the same parameters is applied to the vector magnitude sequence to obtain the temporal envelope of hand movement intensity. This is done after obtaining the current electromyographic signal envelope. and the current acceleration signal envelope Then, the similarity and time difference between the two are calculated. The similarity is calculated using the Pearson correlation coefficient method, within a 200ms time window centered on the current time. After time shift The correlation coefficient between them is calculated, where the time shift τ traverses the range [-100ms, 100ms] in 5ms steps, and the maximum value of the correlation coefficient is taken as the similarity. The time shift corresponding to the maximum correlation coefficient is the current time difference. This real-time similarity and time difference calculation can accurately capture the physiological signal characteristics of the current operation, providing instant biometric data for subsequent identity verification and ensuring that each file operation is closely related to real physiological activities.

[0032] The mobile terminal's authentication module receives the similarity calculated in step S120. and time difference Compare it with the cross-correlation coefficient threshold determined in step S110. and inherent time delay threshold A comparison is performed. The verification logic employs a dual-condition judgment mechanism: first, it checks... This condition ensures that the correlation between the current electromyographic signal and the acceleration signal reaches the individual characteristic level, excluding interference from random noise or non-physiological vibrations; secondly, examination... This condition verifies that the current signal timing relationship conforms to the user's neuromuscular conduction characteristics, preventing signals simulated by external devices from passing the verification. The verification module will only generate a trigger pulse signal when both conditions are met simultaneously.

[0033] The trigger pulse is a square wave signal with a width of 10ms and an amplitude of 5V. This pulse signal serves as a time reference marker, accurately recording the moment when the verification is successful. On the other hand, it triggers the subsequent bone conduction micro-vibration acquisition module to start the data acquisition process. The trigger pulse generation mechanism ensures that only the real user's immediate physiological activity can drive the file operation process to continue. Any replay attack or simulated signal will be rejected because it cannot simultaneously meet the dual constraints of similarity and time difference. In addition, to prevent brute-force attacks, the verification module is equipped with a failure counter. When five consecutive verifications fail, the mobile terminal will enter a 30-second lockout state and record the abnormal event log, further enhancing the system's security.

[0034] Within a 50ms time window following the generation of the trigger pulse, a high-precision bone conduction sensor begins acquiring bone conduction micro-vibration waveforms and transmitting them to the AI ​​module in real time. The high-precision bone conduction sensor employs a piezoelectric ceramic transducer, tightly attached to the surface of the radial styloid process, with a sampling frequency set to 5000Hz to capture the high-frequency micro-vibration components of bone conduction. (Bone conduction micro-vibration waveform) It contains composite information including the user's heartbeat, respiration, muscle tremors, and environmental micro-vibrations, exhibiting high randomness and individual specificity. To extract the instantaneous phase value, the acquired raw waveform is first preprocessed: a bandpass filter (passband range 100-2000Hz) is used to remove low-frequency drift and high-frequency noise, and then the filtered signal is... Perform a Hilbert transform to obtain the analytic signal. ,in This represents the Hilbert transform operator. An analytic signal can be represented as... ,in Instantaneous amplitude, This refers to the instantaneous phase. The instantaneous phase value is calculated... The phase value is obtained in Continuous variation within a range. Within a 50ms acquisition window, at the trigger pulse time... The sampling point is 25ms later, and the instantaneous phase value at that moment is extracted. This serves as the seed parameter for subsequent chaotic state initialization. A 25ms delay was chosen to avoid transient interference that might be introduced by the trigger pulse, while ensuring that the sampling point is located in the stable phase of the user's operation. The extraction accuracy of the instantaneous phase value directly affects the subsequent encryption strength; therefore, a 64-bit floating-point number is used to store the phase value, ensuring at least 15 significant digits of precision. This random source based on the instantaneous phase of bone conduction micro-vibration is unpredictable and non-replicable. Even for the same user with similar actions, due to physiological factors such as heart rate variability and respiratory rhythm differences, the extracted instantaneous phase value will vary significantly each time, providing a high-quality entropy source for dynamic encryption.

[0035] Based on the extracted instantaneous phase value The process of determining the chaotic state values ​​uses the Lorentz chaotic system as the core dynamic model. First, the instantaneous phase values ​​are normalized to the interval [0, 1], using the normalization formula: The normalized phase values ​​are used as the initial perturbation parameters for the Lorentz equations, which are in standard form as follows: ; The parameters σ=10, ρ=28, and β=8 / 3 are the classic parameter settings. The initial state is set as follows: , , This initialization method ensures that different instantaneous phase values ​​will lead the chaotic system to completely different trajectories. The fourth-order Runge-Kutta method is used to numerically integrate the Lorentz equations, with an integration step size of 0.001. After 1000 iterations, the system enters a fully chaotic state. The three-dimensional state vector at the 1000th step is taken. As a chaotic state value, this state value is extremely sensitive to initial conditions; even a tiny difference in the initial phase value (10) -15 The magnitude of the chaos (on a certain order of magnitude) can lead to completely different state values ​​after 1000 steps. After obtaining the chaotic state values, the fragmentation process begins on the electronic archive data stream to be transmitted. Assuming the archive data stream size is N bytes, based on the first component of the chaotic state values... The formula for determining fragment size is as follows: The byte size allows the granular size to dynamically change between 256 and 1024 bytes. This divides the N-byte data stream according to... Cut into different sizes to obtain The data particles are then processed. Next, a nonlinear mapping matrix is ​​generated, and the Lorentz equations are iterated M times. In each iteration, (x, y) coordinate pairs are extracted, and the coordinate values ​​are modulo M to obtain index values ​​in the range [0, M-1]. A mapping relationship from source index to target index is constructed, ensuring that the mapping relationship is bijective (one-to-one correspondence with no duplicates). Based on this mapping matrix, the M data particles are then randomly arranged, from their original order... The data particles were rearranged as , where π is the permutation function defined by the mapping matrix. For generating the temporary key, the Lorentz equations are iterated M times. In each iteration, the three-dimensional state vector is used to derive a 256-bit temporary key using the SHA-256 hash function. The i-th data particle uses the corresponding temporary key. Encryption is performed in AES-256-CTR mode.

[0036] During encryption, the temporary key is expanded into a keystream of the same length as the data granules, and encryption is completed through a byte-by-byte XOR operation. The encryption formula is as follows: Where j is the byte index within the particle. A 16-byte metadata is appended to the header of each encrypted data particle, including an 8-byte timestamp (nanosecond precision), a 4-byte sequence number, and a 4-byte HMAC-SHA256 checksum. The checksum is calculated based on a combination of particle content and chaotic state values, ensuring data integrity and tamper-proof capabilities. Finally, all encrypted data units are concatenated according to the target logical channel number order to form a processed electronic archive data stream. This data stream has the following characteristics: fragmentation prevents individual particles from being parsed independently; out-of-order arrangement breaks the spatial continuity of the original data; instantaneous encryption ensures that the ciphertext transmitted each time is completely different; and the combination of multiple features renders replay attacks and man-in-the-middle attacks infeasible.

[0037] The processed electronic archive data is sent to the receiving end via the secure communication module of the mobile terminal. The communication protocol can use a TLS 1.3 encrypted channel as the transport layer protection, but core security relies on the application layer's biometric binding encryption mechanism. The receiving end first receives the complete data stream and parses it according to the predefined data unit structure (16 bytes of metadata + variable-length encryption granules) to extract M encrypted data units. For each data unit, the validity of the timestamp is first verified, and the time difference between the received time and the timestamp is calculated. If the time difference exceeds a preset time window (e.g., 30 seconds), it is determined to be a replay attack, decryption is refused, a security event log is recorded, and an alarm is sent to the administrator. After the timestamp verification is successful, the receiving end needs to reconstruct the same chaotic state value as the sending end, which requires the receiving end to be able to obtain the instantaneous phase value of the sending end. In practical implementation, a secure initial synchronization mechanism is adopted: when the user first pairs the mobile terminal and the receiver, the initial chaotic seed and the user's physiological characteristic parameters (intrinsic time delay threshold and cross-correlation coefficient threshold) are transmitted via physical contact (such as NFC near-field communication). In subsequent operations, the receiver independently generates the same instantaneous phase value through a biometric verification process synchronized with the transmitter. Specifically, the receiver is also equipped with a physiological signal acquisition module, which acquires the user's electromyography and bone conduction signals simultaneously with the received data, and performs the same S110-S140 steps to generate the receiver's instantaneous phase value. Because the sender and receiver are the same user and their operations are synchronized, the instantaneous phase values ​​extracted from both ends are statistically highly correlated. The receiver uses... Initialize the Lorentz equations, perform the same iterative process as the transmitter, and reconstruct the chaotic state values. Based on the reconstructed chaotic state values, the receiver generates the inverse matrix of the nonlinear mapping matrix using the same algorithm as the transmitter. and M temporary keys For the i-th encrypted data unit, the receiving end uses... Generate a keystream and perform an XOR decryption operation to obtain the decrypted data granules. After decryption, the HMAC-SHA256 checksum is verified. If the verification fails, it indicates that the data has been tampered with during transmission or the key does not match. The receiver rejects the data and requests a retransmission. After all data has been decrypted and verified, the reverse mapping matrix is ​​used... To restore disordered data particles to their original logical order, i.e. Finally, the M data particles, restored to their original order, are sequentially pieced together to reconstruct the complete electronic archive data. Throughout the decryption and reconstruction process, the receiving end ensures the authenticity of the operator's identity through independent biometric verification, ensures key consistency through chaotic state synchronization, ensures the currentity of the operation through timestamp verification, and ensures the immutability of the data through integrity verification. These multi-layered security mechanisms work together to achieve the goal of resisting replay attacks and man-in-the-middle attacks even in offline environments.

[0038] This embodiment uses the user's real-time physiological activity as the dynamic entropy source for the encryption process, ensuring that each file operation generates unique encryption parameters, fundamentally eliminating the replay attack threat faced by traditional static key and periodic token mechanisms. Dual biometric verification using electromyography (EMG) and bone conduction acceleration signals ensures the authenticity and current status of the operator's identity; even if an attacker intercepts the complete encrypted data stream, they cannot successfully replay it without synchronized physiological signals. A chaotic state generation method based on the instantaneous phase of bone conduction micro-vibrations provides high-quality randomness and unpredictability, significantly improving encryption capabilities. The introduction of fragmentation processing and nonlinear mapping further enhances the ciphertext's resistance to analysis, making it impossible to deduce the original file content even if some data is leaked. Timestamp verification and integrity checks improve protection against tampering and man-in-the-middle attacks. This solution maintains complete security even offline and in weak network environments, without relying on real-time server verification, making it particularly suitable for mobile management scenarios involving classified files, providing a practical technical path for high-security electronic file systems.

[0039] In one embodiment of this invention, determining the intrinsic time delay threshold between the electromyographic signal and the acceleration signal includes the following steps: S210. Collect the starting time points of electromyography (EMG) signals and acceleration signals when the user performs standard movements; S220. Calculate the time difference between the starting time of the electromyographic signal and the starting time of the acceleration signal in multiple standard movements. S230. Perform statistical analysis on the time difference to determine the mean and standard deviation of the time difference; S240. Determine the effective range of the time delay based on the mean and standard deviation; S250, Set the median value of the effective range as the inherent time delay threshold.

[0040] During the acquisition of the onset time points of electromyography (EMG) and acceleration signals when the user performs a standard movement, the mobile terminal first guides the user into calibration mode. The calibration interface guides the user to perform the standard movement through both visual and auditory prompts. The standard movement is defined as a rapid transition from a completely relaxed arm state to a clenched fist state and holding it for 2 seconds. The mobile terminal's EMG signal acquisition module uses a differential electrode configuration, attached to the surface of the radial wrist flexor muscles of the forearm. When the user performs the clenching fist movement, the EMG signal suddenly jumps from microvolt-level noise at rest to a millivolt-level action potential burst.

[0041] To accurately identify the onset time of electromyography (EMG) signals, a dual-threshold detection algorithm combined with sliding window energy analysis was employed: First, the root mean square value of the EMG signal at rest for 3 seconds was calculated as the baseline noise level. Then, during the action execution phase, the instantaneous root mean square value within a 50ms sliding window is calculated in real time. ,when For more than 50ms At that time, the start time of this 50ms window is marked as the start time point of the electromyographic signal. Bone conduction acceleration signals were acquired using a triaxial MEMS accelerometer fixed to the radial styloid process. The detection of the acceleration signal initiation point was based on the rate of change analysis of the vector modulus: the vector modulus of the triaxial acceleration was calculated. The rate of change of acceleration is obtained by performing a first-order difference on the vector magnitude sequence. When the rate of change exceeds the threshold of 0.5 g / s for 30 ms consecutively, the start time of this 30 ms window is marked as the start time point of the acceleration signal. To ensure detection accuracy, cubic spline interpolation is used to interpolate discrete sampling points to a sampling rate of 10 times, achieving a time resolution of 0.1 ms. The mobile terminal automatically records the data after each standard action is performed. and The system displays a signal quality score for the action on the screen, and only action data with a score of 85 or higher are included in subsequent analysis.

[0042] After collecting data from 20 standard repetitions, the mobile terminal's data processing module begins calculating the time difference between the onset time of the electromyographic signal and the onset time of the acceleration signal in each repetition. For the i-th standard repetition, the formula for calculating the time difference is: Because the electrophysiological activity generated by muscle contraction precedes mechanical movement, Typically a positive value, ranging from 30-80 ms, this delay reflects the combined time of neuromuscular conduction, tendon force transmission, and skeletal inertial response. To eliminate sampling clock bias, a clock synchronization procedure is performed before calibration begins: synchronization pulses are simultaneously sent to both sensors, the timestamps of the received pulses are recorded, and the clock offset is calculated and compensated for in subsequent data processing. For 20 standard movements, a time difference sequence is obtained. To assess the consistency of time differences, the coefficient of variation between adjacent time differences is calculated. If the deviation of a measurement's time difference from its adjacent measurement exceeds 20 ms, it is marked as an outlier, and the user is required to re-perform the action. In practice, the 20 standard movements are divided into 4 groups of 5 repetitions each, with a 30-second rest between groups to avoid systematic drift caused by muscle fatigue leading to neuromuscular conduction delay.

[0043] After obtaining the time difference sequence, the mobile terminal performs statistical analysis on the sequence to extract individual characteristic parameters. First, the arithmetic mean of the time differences is calculated. This mean reflects the average response delay characteristics of the user's neuromuscular system. The standard deviation of the time difference is then calculated. Standard deviation quantifies the degree of variability in time delay within an individual. To comprehensively assess the characteristics of the time difference distribution, higher-order statistics such as skewness and kurtosis are also calculated: skewness... Reflects the symmetry of the distribution, kurtosis This reflects the sharpness of the distribution. Under normal circumstances, the time difference sequence should exhibit an approximately normal distribution with skewness close to 0 and kurtosis close to 0. Statistical analysis also includes trend testing of the time difference sequence, using the Mann-Kendall trend test to determine whether the time difference exhibits a systematic increasing or decreasing trend with the number of actions. If a significant trend is detected, it suggests that the user may have a learning effect or fatigue effect, requiring recalibration to obtain stable baseline data.

[0044] Based on the mean of the time difference and standard deviation The effective range of the time delay is determined using a confidence interval method based on a normal distribution. In this embodiment, 2 standard deviations are used as the boundary of the effective range, calculated as follows: Lower bound of the effective range. Upper limit of effective range In practical applications, it is also necessary to conduct a reasonableness test on the effective range: if Then force setting ;like Then force setting Valid range after reasonableness test As a criterion for subsequent identity verification, the current operation is considered to conform to the user's physiological characteristics only when the time difference measured in real time falls within this range.

[0045] Set the median value of the effective range as the inherent time delay threshold. The calculation formula is: This threshold, serving as a characteristic parameter of the user's neuromuscular response latency, plays a crucial role in subsequent real-time authentication. To enhance the robustness of the threshold, the mobile terminal also stores the width of the effective range. As an auxiliary parameter, during real-time verification, it is used not only to compare the measurement time difference with... The deviation will also be based on The stringency of the judgment is dynamically adjusted. The inherent time delay threshold and related parameters are encrypted and stored in the secure storage area of ​​the mobile terminal, and are updated regularly to adapt to long-term changes in the user's physiological state. The update strategy is to re-execute the calibration process every 30 days, or to trigger recalibration when 10 consecutive verification failures occur.

[0046] This embodiment establishes an inherent time delay threshold based on individual physiological characteristics through a systematic calibration process and statistical analysis methods. This threshold accurately reflects the response characteristics of the user's neuromuscular system. The design of standard movements and the precise detection algorithm for signal initiation points improve the accuracy and repeatability of time difference measurements. Multiple repeated measurements and outlier removal enhance the statistical reliability of the calibration data. The effective range determination method based on normal distribution theory ensures both security and system usability. The inherent time delay threshold, as a unique physiological identifier for each user, is difficult to forge and replicate. This calibration method is applicable to users of different ages, genders, and physical conditions, achieving broad applicability through personalized parameter settings.

[0047] In one embodiment of this invention, the current electromyography signal envelope and the current acceleration signal envelope are acquired in real time through an AI module, including the following steps: S310. When the user performs a preset action on the mobile terminal, the potential changes of the arm muscles are monitored in real time, and the envelope curve of the potential changes is extracted as the electromyographic signal envelope. S320: Real-time monitoring of hand movement trajectory, extracting the acceleration change envelope of the hand movement trajectory as the acceleration signal envelope.

[0048] When a user performs a preset action on the mobile terminal, such as clicking the file view button, swiping to turn pages, or double-clicking to confirm, the mobile terminal's operating system immediately generates an electronic file operation trigger signal and activates the real-time electromyography (EMG) signal monitoring module. This module continuously collects the potential changes of the arm muscles through surface electrodes attached to the radial wrist flexor muscles of the forearm. The raw EMG signal contains rich frequency components, but is also mixed with noise components such as power line interference, motion artifacts, and baseline drift.

[0049] To extract the envelope curve reflecting muscle contraction intensity, the original signal was first preprocessed: a 50Hz notch filter was used to eliminate power frequency interference, and then a fourth-order Butterworth bandpass filter (20-450Hz) was applied to retain the effective electromyographic frequency band. The preprocessed signal... After full-wave rectification, the bipolar signal is converted into a unipolar signal. The rectification formula is as follows: The rectified signal undergoes envelope extraction using a low-pass filter, specifically a second-order Butterworth filter with a cutoff frequency of 10Hz. To further improve the smoothness and time resolution of the envelope curve, a sliding window root mean square (RMS) method is employed for secondary processing: a Hamming window with a window length of 50ms and a step size of 10ms is set, and the RMS value of the rectified signal within each window is calculated. , where N is the number of sampling points within the window. The envelope curve obtained by this method... It preserves the dynamic characteristics of muscle contraction while exhibiting good noise immunity. During real-time monitoring, the mobile terminal maintains a 500ms circular buffer to continuously update the latest electromyographic signal envelope data.

[0050] Real-time monitoring of hand movement trajectories is achieved through a triaxial MEMS accelerometer fixed to the wristband of the mobile terminal. This sensor, positioned close to the radial styloid process, accurately captures the hand's acceleration in three-dimensional space. The triaxial accelerometer outputs acceleration components along the X, Y, and Z axes. , and To obtain a scalar index reflecting the overall intensity of hand movement, the vector magnitude of triaxial acceleration is calculated. This vector mode eliminates directional information, retaining only the time-varying characteristics of motion intensity. The original acceleration vector mode signal contains both static components of gravitational acceleration and dynamic components of motion acceleration. To extract pure motion information, a high-pass filter is used to remove the static gravitational component, with the high-pass filter cutoff frequency set to 0.5Hz. The filtered dynamic acceleration signal... It reflects the instantaneous acceleration change of hand movements.

[0051] The extraction of the acceleration change envelope employs a processing procedure similar to that of electromyography (EMG) signals: first, the absolute value of the dynamic acceleration signal is taken and rectified; then, a low-pass filter with a cutoff frequency of 10Hz is applied to smooth high-frequency oscillations; finally, the root mean square value is calculated using a sliding window with a window length of 50ms and a step size of 10ms to obtain the acceleration signal envelope. The envelope curve clearly shows the complete process of hand movement from rest to acceleration, uniform speed, deceleration, and back to rest. To ensure time alignment between the electromyographic signal envelope and the acceleration signal envelope, a hardware timestamp synchronization mechanism is used for the two signals. Each sampling point carries a time stamp with nanosecond-level precision. In subsequent similarity and time difference calculations, precise timing matching is performed based on the timestamps.

[0052] This embodiment establishes a dual-modal physiological feature representation system for user operation actions by real-time acquisition and processing of electromyography (EMG) signal envelopes and acceleration signal envelopes. The EMG signal envelope reflects the intensity of electrophysiological activity during muscle contraction, while the acceleration signal envelope reflects the mechanical response characteristics of hand movements. Both characterize the physiological essence of the same operation action from different dimensions. An envelope extraction algorithm combining full-wave rectification, low-pass filtering, and sliding window root mean square (RMS) effectively suppresses high-frequency noise and baseline drift interference, resulting in an extracted envelope curve with good signal-to-noise ratio and temporal resolution. The calculation of the three-axis acceleration vector modulus eliminates the influence of hand movement direction, making the acceleration envelope a direction-independent indicator of motion intensity and enhancing the robustness of the features. Synchronous sampling of EMG and acceleration signals, along with a hardware timestamp mechanism, improves the time alignment accuracy between the two signals. Real-time monitoring and a circular buffer design enable the system to immediately acquire complete physiological signal features at the moment of operation triggering, enhancing the user experience. This dual-modal envelope acquisition method provides high-quality input data for biometric-based authentication.

[0053] In one embodiment of this invention, determining the chaotic state value at the current moment based on the instantaneous phase value includes the following steps: S410. Normalize the instantaneous phase value to a preset value range to obtain the normalized phase value. S420. Use the normalized phase value as the perturbation parameter of the chaotic equation. S430. Update the system state of the chaotic equation based on the perturbation parameters; S440. Iteratively calculate the chaotic equation to obtain the chaotic state value at the current moment; Among them, the chaos equation is a nonlinear dynamic equation.

[0054] Instantaneous phase value Extracted from the Hilbert transform of the bone conduction microvibration waveform, its original numerical range is: Radians. To ensure this phase value can serve as an effective perturbation parameter for the chaotic equations, it needs to be normalized to a numerical range suitable for initializing the chaotic system. Normalization first converts the phase value from radians to a dimensionless scalar, using the following formula: After this conversion, The range of values ​​for is strictly limited to Within the range.

[0055] To enhance the dynamic range and nonlinear characteristics of the normalized phase value, a piecewise nonlinear mapping strategy is adopted: The interval is divided into 10 equal-length subintervals. For a given subinterval, the value falls within the k-th subinterval. within Applying nonlinear spread functions This function introduces sinusoidal modulation within each subinterval, causing minute changes in the phase value to be nonlinearly amplified. The phase value after nonlinear expansion... It remains within the range of [0, 1]. To adapt to the typical initial value range of a Lorentz chaotic system, it is further... Scaling to the [-20, 20] range, the scaling formula is as follows: This numerical range covers the main active region of the Lorentz attractor in phase space. A floating-point precision protection mechanism is also introduced during the normalization process; all calculations use the IEEE 754 double-precision floating-point format to ensure at least 15 significant digits of precision.

[0056] Normalized phase value As perturbation parameters in the chaotic equations, it is necessary to establish a mapping relationship between the phase value and the initial state of the chaotic system. This embodiment uses a Lorentz chaotic system as the core dynamic model, which consists of three coupled first-order nonlinear differential equations: ; in , , For system parameters, the classic parameters are set as follows: , , .

[0057] The introduction of perturbation parameters employs a multi-dimensional initialization strategy: As the fundamental perturbation, a three-dimensional initial state vector is generated through a deterministic nonlinear transformation. Specifically, the initial value of the X-axis is set to... The initial value of the Y-axis is generated by a sine transformation of the phase value. The initial Z-axis value is generated by cosine transform of the phase value. This multidimensional nonlinear initialization strategy ensures that small differences in instantaneous phase values ​​lead to significant separation of the three-dimensional initial state vectors. Furthermore, to enhance the system's adaptability to different user physiological characteristics, a user identifier is introduced as an auxiliary perturbation parameter: the hash value of the user ID is normalized to [0, 1] after being modulo 1000, and used as a coefficient modulation factor. The corrected initial state is: .

[0058] Initial state determined based on perturbation parameters Updating the system state of the chaotic equations requires choosing a suitable numerical integration method to solve the Lorentz differential equations. This embodiment uses the fourth-order Runge-Kutta method for numerical integration. For the state vector... and derivative function First calculate Then calculate , , Last updated status Where h is the integration step size, Let h be the state vector at step n. In this embodiment, the integration step size is set to h = 0.001. Numerical stability monitoring is also implemented during the system state update process: the modulus of the state vector is checked every 100 steps. If the modulus exceeds 100 or is less than 0.1, the integration step size will be automatically adjusted or the system will be reinitialized.

[0059] The chaotic equations are iteratively calculated to obtain the chaotic state value at the current moment, with the number of iterations set to 1000. Starting from the initial state... Initially, continuous iterations were performed using the fourth-order Runge-Kutta method. Complete trajectory sequences were recorded during the iterations. To extract the chaotic state value at the current moment, a multi-point sampling strategy is adopted: the state vectors at steps 800, 900, and 1000 are selected, and a weighted average of the three points is calculated. The final chaotic state value is defined as a three-dimensional vector. The three components of this vector are used in different subsequent encryption stages: the X component determines the granular size of the data fragments, and the Y and Z components are used as seed parameters to generate the nonlinear mapping matrix. The chaotic state value is also bound to a timestamp and an operation identifier, forming a triplet. This triple serves as a global parameter for subsequent fragmentation and instantaneous encryption.

[0060] This embodiment establishes a complete transformation link from instantaneous features of physiological signals to chaotic encryption parameters through a systematic phase value normalization, perturbation parameter mapping, chaotic state update, and iterative calculation process. Nonlinear normalization of instantaneous phase values ​​effectively amplifies the impact of minute differences in physiological signals on the initial chaotic conditions. A multidimensional nonlinear initialization strategy expands the one-dimensional phase value into a three-dimensional initial state, fully utilizing the three-dimensional phase space structure of the Lorentz system. The fourth-order Runge-Kutta numerical integration method achieves a balance between computational efficiency and chaotic quality while ensuring computational accuracy. An iteration depth of 1000 steps makes the final state exponentially sensitive to initial perturbations; any minute difference in physiological signals will lead to completely different chaotic state values. A multi-point weighted sampling strategy improves the robustness of the chaotic state values. The chaotic state values ​​generated by this method have high entropy, high sensitivity, and strong irreversibility, providing a high-quality random source for subsequent data fragmentation and instantaneous encryption, ensuring that each file operation corresponds to a unique encryption parameter.

[0061] In one embodiment of this invention, the electronic archive data stream to be transmitted is fragmented and instantaneously encrypted based on chaotic state values ​​to obtain processed electronic archive data, as shown in the following example. Figure 2 It includes the following steps: S510. Cut the electronic archive data stream into multiple data particles according to a preset particle size; S520. Generate a nonlinear mapping matrix based on chaotic state values; S530. Assign a target logical channel number to each data particle according to the mapping relationship in the nonlinear mapping matrix; S540. Arrange the data particles in random order according to the target logical channel number; S550: Derive multiple temporary keys from chaotic state values, with each temporary key corresponding to a data particle; S560. Use the corresponding temporary key to perform instantaneous symmetric encryption on each data particle to obtain the processed electronic archive data.

[0062] Before fragmenting electronic archival data streams, the particle size needs to be dynamically determined based on chaotic state values. The X component of the chaotic state value... As a control parameter for particle size, it is converted into the number of bytes using a non-linear mapping function. The specific calculation formula is as follows: This formula is designed to ensure that the granularity dynamically varies between 256 bytes and 1024 bytes. Assuming the total size of the electronic archive data stream to be transmitted is N bytes, the granularity is determined accordingly. Cutting is performed to obtain the number of particles. ,in This indicates rounding up to the nearest integer.

[0063] The cutting process employs a sequential block strategy: the data stream starts from byte index 0, and each... A byte is divided into a data particle, and the first M-1 particles are all of the same size. bytes, the size of the last particle is Bytes. For the last insufficient The byte size is padded using the PKCS7 padding scheme: calculate the number of padding bytes. Add P bytes with a value of P to the end of each grain. After cutting, the resulting data grain sequence is obtained. Each particle For length A byte array. To facilitate subsequent reassembly, the original sequence number i and the starting byte offset in the data stream are recorded in the metadata of each particle. .

[0064] The process of generating a nonlinear mapping matrix based on chaotic state values ​​utilizes the continuous evolution characteristic of the Lorentz system. Starting from the chaotic state at step 1000... As a new initial condition, the Lorentz equations are iteratively calculated. The number of iterations is set to M, matching the number of data particles. In each iteration, the X and Y coordinates of the current state are extracted to form a two-dimensional coordinate pair. ,in To map continuous coordinate values ​​to a discrete granular index space, modulo normalization is performed on each coordinate pair: , ,in This indicates a modulo operation, ensuring that the index value falls within the range of the modulo operation. Within the range.

[0065] To resolve the index collision problem, a collision resolution algorithm is adopted: maintain a boolean array of length M. All values ​​are initially false. Each source index is processed sequentially. Calculate its initial target index ,like Then establish a mapping relationship. and set ;like The next unused index is found using linear probing. Where k starts from 1 and increments until a satisfying condition is found. Index, establish mapping This conflict resolution algorithm guarantees that the mapping function π is bijective. To enhance the nonlinearity of the mapping, a quadratic perturbation is introduced: for each mapping relationship, the perturbation amount is calculated. ,in For the Z-coordinate component at the i-th iteration, update the mapping as follows: And re-execute conflict resolution to ensure relevance.

[0066] Based on the mapping relationship in the nonlinear mapping matrix A target logical channel number is assigned to each data particle. Original data particle sequence. The i-th particle Its target logical channel number is defined as To implement this allocation process, a granular-channel mapping table is constructed, which is an array of length M. ,in The target channel position where the i-th particle should be placed is stored. An inverse mapping table is also constructed. ,satisfy To prevent the mapping table itself from leaking particle arrangement information, the mapping table is not transmitted with the data, but is independently reconstructed at the receiving end using the same chaotic state values.

[0067] When assigning target logical channel numbers, channel metadata is also appended to each granularity, including a 16-byte channel header: the first 4 bytes store the source index i, and the middle 4 bytes store the target channel number. The last 8 bytes store a nanosecond-level timestamp. The channel metadata uses the HMAC-SHA256 algorithm to generate a 4-byte integrity checksum. The checksum is calculated based on a combination of granular content, source index, target channel number, and chaotic state value, as shown in the formula: The first 4 bytes of the hash value are taken as a checksum and appended to the channel header.

[0068] Arranging data granules in random order according to the target logical channel number requires referring to a mapping table. Reorganize the particle sequence. Create a temporary array of length M. Traverse the original particle sequence, for the i-th particle Place it at the target location: After this rearrangement operation, the particles originally arranged according to the logical order of the archive data are scrambled, and adjacent data particles may be far apart after the rearrangement, disrupting the spatial continuity of the data. To further enhance security, a spurious particle insertion mechanism is introduced after the disordered arrangement: based on the Z component of the chaotic state value... Calculate the number of fake particles ,generate A spurious particle filled with random data. The spurious particle is randomly inserted into... In the array, a special flag is set in the channel header of the particle to identify it as a fake particle. The receiver identifies and discards these fake particles during decryption.

[0069] Multiple temporary keys are derived from chaotic state values ​​using a key derivation function based on chaotic trajectories. The Lorentz equations are iterated over, starting from the state after generating the mapping matrix, deriving one temporary key in each iteration. For the i-th data particle, the corresponding temporary key... This is generated through the following process: First, extract the... Three-dimensional state vector of step iteration The three components are converted into byte strings, each encoded into 8 bytes using IEEE 754 double-precision format, and then concatenated to obtain 24 bytes of seed data. Then, the seed data is compared with the granular index i and the timestamp. and operation identifiers Combine to form key-derived input .right Applying the SHA-512 hash function yields a 64-byte hash output. Take the first 32 bytes of the hash output as a 256-bit temporary key. To enhance key independence, a key stretching mechanism is also introduced: for the initial key... The PBKDF2 algorithm is executed, with 1000 iterations, and the salt value is the HMAC checksum of the particle, to obtain the final temporary key. .

[0070] Each data particle is instantaneously symmetrically encrypted using the corresponding temporary key, employing the CTR mode of the AES-256 algorithm. For the i-th data particle... Use temporary key Initialize the AES-256 encryptor; the initial counter value for CTR mode is set to... The AES-256-CTR encryption process encrypts the counter value to generate a keystream, and then XORs the keystream with the plaintext particles byte-by-byte to obtain the ciphertext. Specifically, for the particles... The j-th byte Calculate key stream bytes Then calculate the ciphertext bytes. .

[0071] After encryption is complete, the ciphertext particles will be... Combined with the previously generated 20-byte metadata, it forms a complete encrypted data unit. All encrypted data units are concatenated in the order of the target logical channel number to form the final processed electronic archive data stream. Add a 64-byte global header to the beginning of the data stream, containing the number of particles, particle size, timestamp, operation identifier, user identifier hash, and global HMAC-SHA256 checksum.

[0072] This embodiment constructs an electronic archive data protection system based on chaotic dynamics through a multi-layered security mechanism including fragmentation processing, nonlinear mapping, randomized arrangement, temporary key derivation, and instantaneous encryption. The dynamic particle size design ensures that the data segmentation method differs for each operation. A highly randomized rearrangement of particle positions is achieved based on the nonlinear mapping matrix generated by the Lorentz chaotic system. The temporary key derivation mechanism achieves true one-time key encryption through multiple bindings of chaotic states, timestamps, and operation identifiers. Instantaneous symmetric encryption in AES-256-CTR mode provides high-strength cryptographic protection, and combined with an integrity verification mechanism, it can effectively detect any tampering during transmission. The spurious particle insertion technique further enhances resistance to traffic analysis. This scheme deeply integrates user physiological characteristics, chaotic dynamics, and cryptographic techniques, making the encryption process inseparable from the user's current operating state.

[0073] In one embodiment of this invention, generating a nonlinear mapping matrix based on chaotic state values ​​includes the following steps: S610. Input the chaotic state values ​​as seed parameters into the Lorentz equations to initialize the three-dimensional phase space coordinates. S620. Perform numerical integration on the Lorentz equations according to the preset iteration step size to generate a chaotic trajectory sequence; S630. Extract the X-coordinate components and Y-coordinate components from the chaotic trajectory sequence to construct a two-dimensional chaotic coordinate pair sequence; S640. Normalize the two-dimensional chaotic coordinate pair sequence by modulo operation and map the coordinate values ​​to the data granular index space. S650. Based on the normalized coordinate pairs, construct a bijective mapping relationship from the source index to the target index; S660. Fill the bijective mapping relationship into the matrix structure to generate a nonlinear mapping matrix, wherein the dimension of the nonlinear mapping matrix is ​​dynamically determined according to the number of data particles.

[0074] In this embodiment, it is possible to detect whether there are repeated mappings in the rows and columns of the nonlinear mapping matrix. If there are repeated mappings, conflict resolution is performed based on the subsequent values ​​of the chaotic trajectory sequence until a conflict-free nonlinear mapping matrix is ​​obtained.

[0075] Using chaotic state values ​​as seed parameters to initialize the three-dimensional phase space coordinates in the Lorentz equations requires establishing a precise mapping between the chaotic state values ​​and the initial conditions of the Lorentz system. The chaotic state values ​​obtained from step S440... This is already a fully chaotic state of the Lorentz system after 1000 iterations, located in the core region of the Lorentz attractor. To generate the nonlinear mapping matrix, it is necessary to continue evolving from this state; therefore, the chaotic state value is directly used as the initial phase space coordinates for the new round of iterations. , , .

[0076] The standard form of the Lorentz equations remains unchanged: , , The system parameters continue to use classic chaos parameter settings. , , When initializing the three-dimensional phase space coordinates, it is also necessary to verify the validity of the initial state. Verification methods include calculating the magnitude of the initial state vector. The modulus should be within the range of [5, 50]. If it exceeds this range, it indicates that the state may deviate from the attractor and state correction is required. Furthermore, to enhance the personalized characteristics of the mapping matrix, a user feature perturbation is introduced during initialization: the first 24 bytes of the user identifier's SHA-256 hash value are parsed into three double-precision floating-point numbers. Normalized to the interval [-1, 1], and superimposed onto the initial coordinates: , , .

[0077] Numerical integration of the Lorentz equations is performed according to a preset iteration step size to generate a chaotic trajectory sequence. The number of iterations is set to M, equal to the number of data particles. The preset iteration step size continues to be used. The numerical integration method continues to use the fourth-order Runge-Kutta algorithm. Starting from the initial state... Initially, the state sequence is calculated step by step according to the fourth-order Runge-Kutta method. In each iteration, four intermediate slopes are calculated: , , , The derivative function The state update formula is: .

[0078] To monitor the quality of numerical integration, the Lyapunov exponent spectrum of the trajectory is calculated in real time during the iteration process. The QR decomposition method is used to calculate the Lyapunov exponent every 100 steps. If the maximum Lyapunov exponent deviates from the range [0.7, 1.1], the integration step size is automatically adjusted or reinitialized. The generated chaotic trajectory sequence... Complex non-periodic curves are formed in three-dimensional phase space.

[0079] Extract the X and Y coordinate components from the chaotic trajectory sequence to construct a two-dimensional chaotic coordinate pair sequence. For each state point in the trajectory sequence... Extract its X and Y coordinates to form a two-dimensional coordinate pair. The extracted two-dimensional coordinate pair sequence It contains M+1 coordinate pairs. To enhance the discriminative power of the coordinate pairs, a nonlinear transformation is performed on the extracted coordinate values: the hyperbolic tangent function is applied to the X coordinate. Apply the arctangent function to the Y coordinate. Coordinate pairs after nonlinear transformation The distribution is more uniform on a two-dimensional plane.

[0080] Modulo normalization is performed on the sequence of two-dimensional chaotic coordinate pairs to map the coordinate values ​​to a data granular index space. The data granular index space is defined as a set of integers. For the i-th two-dimensional coordinate pair First, shift the coordinate values ​​from the interval [-20, 20] to the interval [0, 40]. The shift formula is as follows: , Then, the absolute values ​​of the translated coordinates are taken and modulo M is performed to obtain the initial values ​​of the source and target indices: , .

[0081] To improve the uniformity of the mapping, a quadratic hash perturbation is introduced after the modulo operation: a hash function is applied to each initial index value. 997 is a large prime number. For fast hash functions. The target index sequence after hash perturbation. It has better randomness and uniformity.

[0082] A bijective mapping from the source index to the target index is constructed based on the normalized coordinate pairs. The bijective mapping is defined as a function... The construction process employs a greedy algorithm combined with a conflict resolution strategy: first, the mapping table is initialized. and occupied marker array According to the source index order Process them one by one. For source index i, obtain its initial target index. ,examine State: If Establish mapping ,set up ;like Conflict resolution is required.

[0083] The conflict resolution employs a linear probing method based on chaotic trajectories: starting from the current chaotic state... Continue iterating the Lorentz equation, extracting the Z-coordinate component at each iteration step. Calculate the detection offset Try a new target index ,examine If the source index is not occupied, a mapping is established; otherwise, iterative probing continues. After conflict resolution, all source indexes are successfully mapped to a unique target index, and the mapping table... The bijective relation is fully documented. .

[0084] The bijective mapping relation is filled into the matrix structure to generate a nonlinear mapping matrix. The nonlinear mapping matrix is ​​defined as follows: binary matrix Matrix elements ,in If and only if the source index i maps to the target index j, i.e. The filling process iterates through the mapping table. For each source index Obtain its target index Set matrix elements To save storage space, the matrix is ​​stored in a sparse matrix format, recording only the positions of non-zero elements, i.e., storing M pairs of elements. .

[0085] In the process of generating a nonlinear mapping matrix, detecting the existence of repeated mappings in the matrix's rows and columns is a crucial step in ensuring bijectivity. Repeated mapping detection is divided into two dimensions: row repeat detection and column repeat detection. Row repeat detection verifies whether each row has exactly one element with a value of 1; column repeat detection verifies whether each column has exactly one element with a value of 1. The detection algorithm traverses all rows and columns of the matrix, counting the number of 1s in each row and column. If any row or column has a number of 1s that is not equal to 1, it is marked as a conflict. For detected conflicts, subsequent values ​​based on the chaotic trajectory sequence are used for resolution. The conflict resolution process is iterative, and the bijectivity of the matrix is ​​re-checked after each resolution until a conflict-free nonlinear mapping matrix is ​​obtained.

[0086] This embodiment transforms the chaotic state derived from user physiological characteristics into a rearrangement rule for data particles. Initialization of three-dimensional phase space coordinates and user characteristic perturbation ensure that the mapping matrices generated by different users and at different operation times are completely different. Numerical integration and exponential monitoring guarantee the quality and long-term stability of the chaotic trajectory. Extraction of two-dimensional coordinate pairs and nonlinear transformation maintain the richness of chaotic characteristics while reducing computational complexity. The combination of modular normalization and hash perturbation achieves a high-quality mapping from continuous chaotic coordinates to discrete particle indices. A conflict resolution strategy based on chaotic trajectories efficiently solves the mapping conflict problem. The dynamic-dimensional matrix structure and sparse storage format enable this method to adapt to archive data of different sizes. The generated nonlinear mapping matrix, as a permutation matrix, possesses invertibility and orthogonality, providing mathematical guarantees for the disordered arrangement and subsequent recovery of data particles.

[0087] In one embodiment of this invention, each data particle is instantaneously symmetrically encrypted using a corresponding temporary key to obtain processed electronic archive data, including the following steps: S710, Obtain the byte stream of the current data particle; S720. Extend the temporary key to the same length as the byte stream to generate a key stream; S730: Perform a byte-by-byte XOR encryption operation on the byte stream and the key stream to obtain encrypted data particles; S740. A timestamp and integrity check code are appended to the header of the encrypted data particles to form an encrypted data unit. S750: Concatenate all encrypted data units in the order of the target logical channel number to obtain the processed electronic archive data.

[0088] Deriving multiple temporary keys from chaotic state values, with each temporary key corresponding to a data particle, involves the following steps: Extract continuous numerical segments from chaotic state values, with each numerical segment corresponding to a data particle; Perform a hash operation on each numerical fragment to generate a fixed-length key raw material; The temporary key is obtained by performing an XOR operation between the key raw material and the preset salt value.

[0089] The chaotic state continues to evolve after the mapping matrix is ​​generated in step S440, with additional M iterations of the Lorentz equations, where M is the number of data particles, and each iteration corresponds to the key generation of one data particle. For the i-th data particle, starting from the... Three-dimensional state vector of step iteration Extract numerical fragments from them.

[0090] The extraction of numerical fragments employs a multi-component fusion strategy: the three coordinate components are converted into byte representations in IEEE 754 double-precision floating-point format, with each component occupying 8 bytes, and the resulting fragments are concatenated to obtain a 24-byte original numerical fragment. To enhance the randomness of the numerical fragments, bit-level obfuscation is performed on the concatenated byte sequence: the 24 bytes are treated as a 192-bit bit string, and the bits are flipped according to the sign pattern of the chaotic state to obtain the obfuscated numerical fragments. Each obfuscated numerical fragment is hashed to generate a fixed-length key raw material; the hash algorithm chosen is SHA-512. The input to the hash operation includes not only the numerical fragment itself but also contextual information. Where i is the granular index, For the operation trigger timestamp, For operation identifiers, For the user identifier. Calculate the hash value. 64 bytes of key raw materials were obtained. .

[0091] To further enhance the key's resistance to attacks, a key strengthening mechanism is introduced: the raw key material is XORed with a preset salt value. The salt value is generated based on a combination of system-level parameters and the user's long-term key: first, a feature vector is extracted from the user's biometric calibration data, and then this feature vector is used to derive a 64-byte long-term key for the user via the PBKDF2 algorithm. The iteration count is set to 10000; then the user's long-term key is combined with the system's random salt. Perform an XOR operation to obtain the preset salt value. For the key raw material of the i-th particle, perform an XOR operation. This yields a 64-byte temporary key. To meet the key length requirements of the AES-256 algorithm, the first 32 bytes of the temporary key are used as the final encryption key. The last 32 bytes serve as the key for the message authentication code. M temporary keys generated It is stored in a protected memory area and uses a use-and-destroy strategy.

[0092] Obtaining the byte stream of the current data particle requires extracting it sequentially from the scrambled data particle sequence according to the encryption order. After the scrambling process in step S540, the data particles have been reorganized according to the target logical channel number. The encryption process proceeds in ascending order of the target channel numbers. For a particle with target channel number j, its corresponding source index is... Get the byte stream of the current data particle. The length of this byte stream is Bytes. Byte streams are represented as byte arrays. Each element This is an 8-bit unsigned integer. While acquiring the byte stream, the metadata information of the particle is recorded, including the source index i, the target channel number j, and the byte offset of the particle in the original data stream. .

[0093] The temporary key is expanded to the same length as the byte stream to generate the key stream; this process employs a key stream generator based on block ciphers. Temporary key The length is 32 bytes, while the data granule byte stream length is... Typically between 256 and 1024 bytes, the key needs to be expanded into a key stream of the same length as the byte stream. Key stream generation uses the AES-256 algorithm in counter mode: first, an initial counter value is constructed. Where i is the granular index, For timestamps, The numbers are random, and the resulting 16-byte initial counter is obtained by concatenating them.

[0094] The keystream generation process is performed in blocks: the length of the byte stream is divided into blocks. Divide by 16 to get the required number of AES blocks. For the k-th block, calculate the current counter value. Use temporary key Encrypt the counter value using AES-256. This yields a 16-byte keystream block. Concatenating all keystream blocks results in the complete keystream. The beginning of the key stream interception Bytes as the final key stream The generated keystream It has the same structure as byte streams.

[0095] Perform a byte-by-byte XOR encryption operation on the byte stream and the key stream. For data granular byte streams... and key stream Perform XOR operation byte by byte ,in , This indicates a bitwise XOR operation. The encrypted data granules are... To improve encryption performance, the XOR operation is implemented using vectorization: leveraging the SIMD instruction set, it processes 32 bytes or more of data at a time. Side-channel attack protection is also implemented during encryption: constant time is used to ensure that the execution time of the XOR operation does not depend on the data content.

[0096] A timestamp and integrity checksum are appended to the header of the encrypted data particles to form encrypted data units. The timestamp is represented by a 64-bit integer with nanosecond precision, recording the exact moment the trigger pulse is generated. The integrity check code is generated using the HMAC-SHA256 algorithm. The inputs to the HMAC calculation include the encrypted data granules, timestamps, granule indexes, and context information. HMAC calculation uses the MAC portion of the temporary key. As a key: Take the first 16 bytes of the HMAC output as the integrity checksum. .

[0097] The structure of the encrypted data unit is: 4-byte unit header + 8-byte timestamp + 16-byte integrity check code + Byte ciphertext particles, total length is Bytes. The cell header contains the magic number, version number, and flags. Once the cell is constructed, it is written to the output buffer.

[0098] All encrypted data units are concatenated according to the target logical channel number order to obtain the processed electronic archive data. The concatenation process is performed in ascending order of the target channel number: for each target channel number... The corresponding encrypted data units are then sequentially encrypted. Appended to the output data stream. The output data stream is constructed using a streaming approach: a dynamically expanding byte buffer is initialized, the target channel numbers are traversed, and for each number j, the encrypted data unit array is retrieved. Copy its contents to the output buffer.

[0099] To enhance the self-describing capability of the data stream, a global header is added at the very beginning of the data stream. The global header is 128 bytes long and includes the magic number, version number, number of granules M, and granule size. Original archive data size N, operation trigger timestamp Operation identifier The data includes a user identifier hash, a chaotic state fingerprint, and a global HMAC-SHA512 checksum. The final structure of the processed electronic archive data is: a 128-byte global header + ... A sequence of byte-encrypted data units.

[0100] This embodiment implements a transient symmetric encryption mechanism based on chaotic states through a complete process of temporary key derivation, keystream generation, byte-by-byte XOR encryption, integrity protection, and data stream concatenation. The temporary key derivation process integrates multi-dimensional information such as chaotic dynamics, timestamps, operation identifiers, and user identity to generate a temporary key with high entropy and strong unpredictability. Each data particle uses an independent temporary key, achieving a true one-time pad effect. Keystream generation based on the AES-256-CTR mode efficiently expands short keys into a keystream of equal length to the data, supporting parallel processing and vectorized optimization. The simplicity and symmetry of the byte-by-byte XOR operation ensure complete consistency between encryption and decryption. The addition of timestamps and HMAC integrity check codes provides time binding and tamper-proof protection for each encrypted data unit. Timestamps ensure the timeliness of the data and effectively identify and reject replay attacks. The structured design of the encrypted data units facilitates parsing and verification at the receiving end. Self-describing information in the global header and global HMAC checksum provide metadata management and end-to-end integrity protection for the entire data stream.

[0101] In one embodiment of this invention, the electronic archive data decrypted at the receiving end is fragmented, and the fragmented data is reassembled to obtain the electronic archive data, including the following steps: S810: The receiving end acquires the processed electronic archive data and parses it according to the structural characteristics of the encrypted data units to extract each encrypted data unit. S820: Extract the timestamp and integrity check code from each encrypted data unit; S830. Integrity verification is performed based on integrity check codes. When the integrity verification passes, the receiver reconstructs the same chaotic trajectory sequence using the chaotic state values ​​synchronized with the transmitter. S840. Based on the chaotic trajectory sequence, generate a corresponding temporary key set according to the same derivation rule as the sender; S850. Match each encrypted data unit with the corresponding temporary key according to the target logical channel number; S860: Perform a byte-by-byte XOR decryption operation on each encrypted data unit using a matching temporary key to restore the original data particles; S870: Reconstruct the same nonlinear mapping matrix as the transmitter based on the chaotic state values, and obtain the inverse mapping relationship; S880. Based on the inverse mapping relationship, restore the disordered data particles to their original logical order; S890. All data particles after restoration of order are spliced ​​together according to the original cutting boundaries to reassemble complete electronic archive data.

[0102] After acquiring the processed electronic archive data, the receiving end first needs to parse the data stream based on its structural characteristics to extract each encrypted data unit. The receiving end receives the complete data stream through a secure communication channel; the total length of the data stream is [missing information]. The parsing process begins at the start of the data stream. First, it reads the 128-byte global header and verifies whether the magic number field in the header is the expected value 0x5A5A5A5A. If the magic number does not match, it indicates that the data stream format is incorrect or the transmission is corrupted. The receiving end refuses to process the data and returns an error message.

[0103] After the magic number verification is passed, extract the key parameters from the global header: number of particles M, particle size. Original archive data size N, operation trigger timestamp Operation identifier The parameters extracted include the user identifier hash value and the chaotic state fingerprint. After extracting these parameters, the receiving end first performs a timeliness check: calculating the current time... With operation trigger timestamp Time difference between If the time difference exceeds the preset time window, it is judged as a replay attack and the attack is rejected. At the same time, a security event log is recorded and an alarm is sent to the administrator.

[0104] After the timeliness verification is passed, the global HMAC-SHA512 checksum is extracted from the global header and the user's long-term key is used. The HMAC value is calculated for the entire data stream and compared with the extracted checksum. If they do not match, it indicates that the data stream has been tampered with during transmission, and the receiving end rejects the request. After global integrity verification is successful, the encrypted data unit sequence is parsed: starting from the position after the global header, parsing is performed according to the unit length. The encrypted data units are extracted byte by byte. For the j-th unit, the data at the corresponding position is read. First, the unit header is verified, and the flag bit is parsed to determine whether the unit is a fake particle. If the flag bit indicates a fake particle, the unit is skipped; if it is a real particle, the complete data of the unit is stored in the encrypted data unit array. middle.

[0105] Extract the timestamp and integrity check code from each encrypted data unit. For the j-th encrypted data unit... Its structure is: 4-byte header + 8-byte timestamp + 16-byte integrity checksum + Byte-level encrypted particles. Timestamp extraction begins at a byte offset of 4 from the unit, reading an 8-byte big-endian integer and parsing it into a timestamp value. Timestamp verification includes absolute timeliness verification and relative consistency verification. The integrity check code is extracted starting at a byte offset of 12 from the cell, reading 16 bytes of the HMAC value. The extracted checksum and timestamp are stored together in a metadata array: .

[0106] Integrity verification is performed based on an integrity check code. When the verification passes, the receiver reconstructs the same chaotic trajectory sequence using chaotic state values ​​synchronized with the transmitter. The chaotic state reconstruction at the receiver employs a biometric verification process synchronized with the transmitter: the receiver is also equipped with electromyography (EMG) and bone conduction acceleration (BCA) signal acquisition modules. Simultaneously with receiving the data stream, the user performs the same file operation, triggering biometric acquisition at the receiver. The receiver executes the complete process from S110 to S140: acquiring EMG and BCA signals, calculating the signal envelope, verifying similarity and time difference, generating a trigger pulse, acquiring bone conduction micro-vibration waveforms, and extracting instantaneous phase values. .

[0107] Receiver uses The Lorentz chaotic system is initialized according to the process from S410 to S440, and normalization, perturbation parameter mapping, system state update, and 1000-step iterative calculation are performed to obtain the chaotic state value at the receiving end. To verify whether the chaotic state reconstructed by the receiver is consistent with that of the transmitter, a chaotic state fingerprint is calculated and compared with the chaotic state fingerprint extracted from the global header. If the fingerprints match, the chaotic state reconstruction is successful; if the fingerprints do not match, the receiver refuses to decrypt. After the chaotic state verification is successful, the receiver continues to evolve from this state, performing M iterations of the Lorentz equations to generate a chaotic trajectory sequence. .

[0108] Based on the chaotic trajectory sequence, a corresponding set of temporary keys is generated according to the same derivation rules as the sender. The receiver continues the evolution from the chaotic state values, performing an additional M iterations of the Lorentz equations. For the i-th data particle, a numerical fragment is extracted from the three-dimensional state vector of the corresponding iteration step. The processes of numerical fragment extraction, hashing, and salt XOR are completely identical to those of the sender, generating M temporary keys. Numerically, it should be highly consistent with the temporary key generated by the sender. After the temporary key is generated, the receiver can begin to perform integrity verification on each encrypted data unit: for the j-th unit, the checksum is calculated using the MAC part of the temporary key and the HMAC-SHA256 algorithm, and compared with the extracted... The system performs a comparison. If the units match, they pass the integrity check; otherwise, they are marked as damaged or tampered with.

[0109] Based on the target logical channel number, each encrypted data unit is matched with its corresponding temporary key. The encrypted data units are arranged in ascending order of the target logical channel numbers, with the j-th unit corresponding to target channel number j. To establish the mapping relationship between the unit and the temporary key, the source index i corresponding to the target channel number j needs to be determined. This mapping relationship is defined by the inverse mapping of a nonlinear mapping matrix. After inverse mapping reconstruction, for the target channel number j, the inverse mapping function is used... Get source index i and allocate temporary key. .

[0110] For each encrypted data unit, a byte-by-byte XOR decryption operation is performed using a matching temporary key to reconstruct the original data particles. For the encrypted data unit with target channel number j, its corresponding source index is... The matching temporary key is The decryption process first reconstructs the keystream: using a temporary key. Generates the same keystream as during encryption using AES-256-CTR mode. Extracts ciphertext particles from encrypted data units. Perform a byte-by-byte XOR decryption operation on the ciphertext and the key stream. The decrypted data particles are obtained. The decrypted data particles are stored in an array of plaintext particles. middle.

[0111] The receiver reconstructs the same nonlinear mapping matrix as the transmitter based on the chaotic state values, and obtains the inverse mapping relationship. Using the reconstructed chaotic state values, the receiver reconstructs the nonlinear mapping matrix following the complete process from S610 to S660. It initializes the three-dimensional phase space coordinates, introduces user-specific perturbations, performs M iterations of the Lorentz equations to generate a chaotic trajectory sequence, extracts coordinate components from the trajectory to construct a two-dimensional coordinate pair sequence, performs normalized mapping, and constructs a bijective mapping relationship from the source index to the target index. After the mapping matrix is ​​reconstructed, the inverse mapping relationship is calculated. For each target index j, search the mapping table for the element that satisfies the following conditions: For the source index i, establish an inverse mapping. .

[0112] Based on the inverse mapping relationship, the disordered data particles are restored to their original logical order. The decrypted data particles are stored in the plaintext particle array. In the diagram, the index is the source index i. Create an ordered particle array of length M. For each source index i, retrieve the i-th particle from the plaintext particle array and place it in the i-th position of the sorted array. .

[0113] All data particles, restored to their original order, are pieced together according to their original cutting boundaries to reconstruct the complete electronic archive data. An output buffer of length N bytes is created. For the i-th particle Calculate its starting position in the output buffer. The particle data is copied to the corresponding location in the output buffer. For the last particle, if PKCS7 padding was performed at the sending end, the padding bytes need to be removed before splicing. After splicing, the output buffer contains the complete original archive data. To verify the correctness of the reassembly, the SHA-256 hash value of the reassembled data is calculated and compared with the hash value of the original data appended at the sending end. If they match, the reassembly is confirmed to be successful.

[0114] The cutting boundary is determined through the following steps: First, the original archive data size N and particle size are extracted from the global header. Calculate the number of particles Then, for the first M-1 particles, the cutting boundary is... For the last particle, the cutting boundary is .

[0115] This embodiment achieves secure decryption and accurate reconstructing of encrypted electronic archive data at the receiving end. Data stream parsing and global integrity verification comprehensively check the correctness of data format and transmission integrity before processing begins. The timeliness verification mechanism effectively identifies and prevents replay attacks by checking the timestamp triggered by the operation. The chaotic state reconstruction process establishes a strong binding between the receiving end user's identity and decryption authority through biometric verification synchronized with the sending end. Temporary key generation adopts a derivation rule completely consistent with the sending end, ensuring precise key matching. Integrity check code verification provides fine-grained integrity protection for each data granularity. The symmetry and efficiency of the byte-by-byte XOR decryption operation make the decryption process completely equivalent to the encryption process in terms of performance and security. The reconstruction of the nonlinear mapping matrix and the acquisition of the inverse mapping relationship accurately restore the granular disorder rules of the sending end. The data splicing process ensures complete consistency between the reconstructed data and the original archive data through precise cutting boundary calculation and padding processing.

[0116] This application also provides a mobile terminal, including: The memory is configured to store instructions; and The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the aforementioned AI electronic record management method based on a mobile terminal.

[0117] Reference Figure 3 This application also provides an AI-based electronic record management system based on a mobile terminal, including: Mobile terminal 10; The receiver 20 is connected to the mobile terminal 10.

[0118] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0119] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0120] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0121] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0122] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0123] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0124] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media, such as modulated data signals and carrier waves.

[0125] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0126] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A mobile terminal-based AI electronic file management method, characterized by, Applied to a mobile terminal, which includes an AI module, the method includes: The AI ​​module collects the wearer's electromyography (EMG) and bone conduction acceleration signals, and determines the inherent time delay threshold and cross-correlation coefficient threshold between the EMG and acceleration signals. In response to the trigger signal of electronic record operation, the AI ​​module collects the current electromyographic signal envelope and the current acceleration signal envelope in real time, and calculates the similarity and time difference between the current electromyographic signal envelope and the current acceleration signal envelope; A trigger pulse is generated when the similarity is greater than the cross-correlation threshold and the time difference is less than or equal to the inherent time delay threshold. Within the time window of trigger pulse generation, the AI ​​module acquires bone conduction micro-vibration waveforms and extracts the instantaneous phase value of the bone conduction micro-vibration waveforms. The chaotic state value at the current moment is determined based on the instantaneous phase value, and the electronic archive data stream to be transmitted is fragmented and instantaneously encrypted based on the chaotic state value to obtain the processed electronic archive data. The processed electronic archive data is sent to the receiving end, whereby the receiving end decrypts the processed electronic archive data to obtain fragmented data and reassembles the fragmented data to obtain the electronic archive data.

2. The method according to claim 1, characterized in that, Determining the intrinsic time delay threshold between electromyographic signals and acceleration signals includes: Collect the start time points of electromyography (EMG) signals and acceleration signals when the user performs standard movements; Calculate the time difference between the start time of the electromyographic signal and the start time of the acceleration signal in multiple standard movements; Perform statistical analysis on the time difference to determine its mean and standard deviation; The effective range of the time delay is determined based on the mean and standard deviation. Set the median value of the effective range as the inherent time delay threshold.

3. The method according to claim 1, characterized in that, The AI ​​module collects the current electromyographic signal envelope and the current acceleration signal envelope in real time, including: When a user performs a preset action on a mobile terminal, the potential changes of the arm muscles are monitored in real time, and the envelope curve of the potential changes is extracted as the electromyographic signal envelope. Real-time monitoring of hand movement trajectory, extracting the acceleration change envelope of the hand movement trajectory as the acceleration signal envelope.

4. The method according to claim 1, characterized in that, The chaotic state value at the current moment is determined based on the instantaneous phase value, including: The instantaneous phase value is normalized to a preset range to obtain the normalized phase value; The normalized phase value is used as the perturbation parameter of the chaotic equation; Update the system state of the chaotic equation based on the perturbation parameters; The chaotic equations are calculated iteratively to obtain the chaotic state value at the current moment; Among them, the chaos equation is a nonlinear dynamic equation.

5. The method according to claim 1, characterized in that, Based on chaotic state values, the electronic archive data stream to be transmitted is fragmented and instantaneously encrypted to obtain the processed electronic archive data, including: The electronic archive data stream is cut into multiple data particles according to a preset particle size; Generate a nonlinear mapping matrix based on chaotic state values; Based on the mapping relationship in the nonlinear mapping matrix, assign a target logical channel number to each data particle; Arrange the data particles in random order according to the target logical channel number; Multiple temporary keys are derived from the chaotic state values, and each temporary key corresponds to a data particle. Each data particle is instantaneously symmetrically encrypted using the corresponding temporary key to obtain the processed electronic archive data.

6. The method according to claim 5, characterized in that, Based on the chaotic state values, a nonlinear mapping matrix is ​​generated, including: The chaotic state values ​​are used as seed parameters to input the Lorentz equations and the three-dimensional phase space coordinates are initialized. Numerical integration of the Lorentz equations is performed according to a preset iteration step size to generate a chaotic trajectory sequence. Extract the X and Y coordinate components from the chaotic trajectory sequence to construct a two-dimensional chaotic coordinate pair sequence; Normalize the two-dimensional chaotic coordinate pair sequence by performing a modulo operation, and map the coordinate values ​​to the data granular index space; Based on the normalized coordinate pairs, construct a bijective mapping relationship from the source index to the target index; The bijective mapping relationship is filled into the matrix structure to generate a nonlinear mapping matrix, where the dimension of the nonlinear mapping matrix is ​​dynamically determined according to the number of data particles.

7. The method according to claim 5, characterized in that, Each data particle is instantaneously symmetrically encrypted using the corresponding temporary key to obtain the processed electronic archive data, including: Get the byte stream of the current data particle; Extend the temporary key to the same length as the byte stream to generate a key stream; Perform a byte-by-byte XOR encryption operation on the byte stream and the key stream to obtain encrypted data granules; A timestamp and an integrity check code are appended to the header of the encrypted data particles to form encrypted data units; All encrypted data units are concatenated in the order of the target logical channel number to obtain the processed electronic archive data.

8. The method according to claim 1, characterized in that, The electronic archive data, after being decrypted and processed at the receiving end, becomes fragmented data. This fragmented data is then reassembled to obtain the final electronic archive data, including: The receiving end acquires the processed electronic archive data and parses it according to the structural characteristics of the encrypted data units to extract each encrypted data unit; Extract the timestamp and integrity check code from each encrypted data unit; Integrity verification is performed based on integrity check codes. When the integrity verification passes, the receiving end reconstructs the same chaotic trajectory sequence using chaotic state values ​​synchronized with the sending end. Based on the chaotic trajectory sequence, a corresponding temporary key set is generated according to the same derivation rule as the sender; Each encrypted data unit is matched with its corresponding temporary key based on the target logical channel number; Perform a byte-by-byte XOR decryption operation on each encrypted data unit using a matching temporary key to restore the original data particles; Based on the chaotic state values, a nonlinear mapping matrix identical to that of the transmitting end is reconstructed to obtain the inverse mapping relationship; Based on the inverse mapping relationship, the disordered data particles are restored to their original logical order; All data particles after being restored to their original order are pieced together and reassembled to obtain complete electronic archive data.

9. A mobile terminal, characterized in that, include: The memory is configured to store instructions; as well as The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the AI ​​electronic record management method based on a mobile terminal according to any one of claims 1 to 8.

10. An AI-based electronic record management system based on a mobile terminal, characterized in that, include: The mobile terminal according to claim 9; The receiving end connects to the mobile terminal.

Images