Data shredding method based on four-dimensional manifold chaotic dynamic system

Abstract

The application relates to the technical field of data security and discloses a data pulverization method based on a four-dimensional manifold chaotic dynamic system, which generates a space-time coupled chaotic entropy source through system initialization, dynamically allocates entropy decomposition channels to input data according to type characteristics, realizes irreversible pulverization of data and entropy value control, then completes fragment-level distributed verification by means of edge nodes, generates on-chain credentials with auditability, and combines the verification result to realize real-time closed-loop updating of chaotic parameters, thereby forming an adaptive feedback control mechanism. The data is realized to be non-tamperable and chain backtracking through improvement of a Merkle-Patricia tree structure, the long-term validity of verification information is ensured by using a security signature based on a lattice structure, and the stability of the entropy source is optimized through a chaotic disturbance feedback mechanism, so that the safety, reliability and adaptability of the system under the threat of a high-complexity calculation model are comprehensively improved.

Description

Technical Field

[0001] This invention relates to the field of data security technology, specifically a data shredding method based on a four-dimensional manifold chaotic dynamic system. Background Technology

[0002] As information systems become increasingly complex, the demand for strong encryption mechanisms and high-entropy random sources is growing in data processing, transmission, and verification. Particularly in typical scenarios such as blockchain, edge computing, and privacy computing, system security often relies on highly random and unpredictable entropy source generation mechanisms and the high reliability of multi-node collaborative verification. However, existing pseudo-random number generators and distributed verification systems generally suffer from high predictability, unstable entropy values, and a lack of feedback mechanisms, making it difficult to meet the requirements for data integrity and attack resistance in high-security scenarios.

[0003] Traditional encryption schemes rely heavily on classical algorithms and low-order dynamic perturbation mechanisms, which prove inadequate in the face of increasingly complex attack models (including speculative attacks, replay attacks, and perturbation injection). Furthermore, in resource-constrained edge nodes, entropy generators and encryption modules struggle to achieve dynamic adjustment and adaptive load control, leading to frequent problems such as entropy degradation, poor data consistency, and difficulty in maintaining system stability.

[0004] Therefore, the main drawback of the existing technology is the lack of a mechanism that can achieve adaptive entropy source adjustment and feedback control based on the operating state. In particular, in applications that support distributed architecture and high dynamic security requirements, there is a lack of a unified high-entropy data processing and system closed-loop update mechanism.

[0005] To address the aforementioned shortcomings, this invention proposes a data shredding method based on a four-dimensional manifold chaotic dynamical system. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a data shredding method based on a four-dimensional manifold chaotic dynamic system, which solves core security problems of existing blockchain systems, such as the fragility of encryption algorithms, the ease with which verification information can be tampered with, and the insufficient randomness of entropy sources under the threat of high-complexity computational models.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a data shredding method based on a four-dimensional manifold chaotic dynamical system, comprising the following steps:

[0008] System initialization: Configure and activate the four-dimensional manifold chaotic dynamic system, generating a spatiotemporally coupled chaotic entropy source;

[0009] Data classification processing: Identify the type characteristics of the input data and dynamically select the corresponding entropy decomposition channel;

[0010] Entropy decomposition processing: Based on the selected channel, the data is irreversibly compressed to output a set of fragments that conform to the entropy value standard;

[0011] Distributed verification: Collaborative verification of fragment sets is performed by edge computing nodes to generate auditable verification credentials;

[0012] Closed-loop update: The verification credentials are written to the blockchain, and the system's chaotic parameters are updated based on the evidence storage results to form a feedback control closed loop.

[0013] Preferably, the four-dimensional manifold chaotic dynamical system in the system initialization step is defined by the following set of fractional differential equations:

[0014]

[0015] Where α is the fractional derivative order, σ is the spatial coupling coefficient, ρ is the nonlinear feedback coefficient, β is the dissipation coefficient, γ is the time tensor coupling strength, η is the time flow feedback coefficient, κ is the spatiotemporal synchronization coefficient, λ is the time flow attenuation coefficient, T is the local time flow variable, and W t It is a time-dependent feature tensor.

[0016] Preferably, the data classification and processing steps include:

[0017] Multimodal feature extraction:

[0018] Text data: Calculate the average path length L of the dependency parsing tree avg ≥2.5;

[0019] Image data: The spectral radius R of the extracted convolutional feature map is ≥0.7;

[0020] Video data: Calculating the fractal dimension D of the optical flow field f ≥1.8;

[0021] Routing decision execution:

[0022]

[0023] The logical operator precedence is as follows: data type matching is performed first, followed by feature threshold judgment; L avg R is the average path length of the dependency syntax tree, R is the spectral radius of the convolutional feature map, and D is the input data stream.

[0024] Preferably, the entropy decomposition process includes the following steps when selecting text decomposition channels:

[0025] Construct syntactic entangled states;

[0026] Apply decoherence gate operation;

[0027] Irreversible fragments are generated through projection measurement, satisfying an entropy value ≤ 0.01 bit / word.

[0028] Preferably, the entropy decomposition processing step includes the following when selecting image decomposition channels:

[0029] Implementing metasurface optical convolution;

[0030] Generate a chaotic speckle map that satisfies the structural similarity index SSIM≤0.05;

[0031] Verify mutual information entropy.

[0032] Preferably, the entropy decomposition processing step includes the following when selecting video decomposition channels:

[0033] Constructing the spacetime crystal Hamiltonian;

[0034] Applying a critical temperature field triggers a phase transition;

[0035] Output the annihilated video cube, satisfying PSNR≤10dB.

[0036] Preferably, in the distributed verification step:

[0037] Key materials required for collaborative verification: Based on the chaotic entropy source generated during the system initialization steps, verification keys are generated in a time series.

[0038] Algorithm for generating auditable verification credentials: Tensor operations are performed collaboratively between edge nodes and the cloud to generate the final verification credentials.

[0039] Preferably, in the closed-loop update step:

[0040] Blockchain-based evidence storage: The verification credentials generated by the distributed verification process are written into the blockchain and stored using an improved Merkle-Patricia tree structure;

[0041] Parameter closed-loop update: Adjust the parameters of the chaotic system based on the evidence storage results.

[0042] Preferably, measures to resist highly complex attacks are also included:

[0043] A high-entropy random number generator is integrated into the entropy decomposition process, with a generation rate ≥100Mbps;

[0044] Embed lattice-based digital signatures in the verification credentials of the distributed verification step.

[0045] This invention provides a data shredding method based on a four-dimensional manifold chaotic dynamical system. It has the following beneficial effects:

[0046] 1. This invention adopts an "immutable blockchain evidence storage mechanism based on an improved Merkle-Patricia tree structure" to achieve highly reliable recording of multi-stage verification information and chain-like backtracking of historical states. Compared with the trust bottleneck caused by relying on local caching or centralized database storage in existing solutions, this invention avoids the risks of forgery and overwriting in the verification chain from the underlying architecture, providing structural support for traceability and auditing scenarios in complex distributed environments, and effectively filling the gap in existing technologies in terms of tamper-proof verification credentials and global consistency.

[0047] 2. This invention introduces a "lattice-based tamper-resistant signature embedding mechanism" and integrates it into the distributed verification credential structure, achieving dual protection against structural counterfeiting and complex attacks during the transmission and on-chain recording of verification information. Compared to traditional RSA or elliptic curve signature schemes, this invention offers higher security margins and scalability, providing a technological foundation for future high-security communication.

[0048] 3. This invention combines a "≥100Mbps high-entropy random number generator with chaotic disturbance control logic based on feedback formulas" to construct a dynamically updated chaotic entropy source feedback system. This system maintains unpredictable disturbances and stable output performance when processing high-frequency data, solving the problems of slow entropy source fluctuation response and single disturbance parameters in existing methods. This scheme, through real-time parameter closed-loop self-adjustment, addresses the risk of system instability or entropy degradation under high load or abnormal conditions, achieving a dual balance between safety and performance. Attached Figure Description

[0049] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0050] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Please see the appendix Figure 1 This invention provides a data shredding method based on a four-dimensional manifold chaotic dynamical system, comprising the following steps:

[0052] S1. System initialization: Configure and activate the four-dimensional manifold chaotic dynamic system to generate a spatiotemporally coupled chaotic entropy source;

[0053] In this embodiment, system initialization aims to construct a four-dimensional manifold chaotic dynamic system with a high-dimensional coupled structure and fractional-order evolution characteristics, thereby generating a chaotic entropy source with high entropy, strong nonlinearity, and time-space linkage, providing initial dynamic support for the entire data shredding process. Through precise modeling, parameter control, and tensor construction, this system achieves both physical feasibility and theoretical complexity in engineering implementation.

[0054] Model for constructing chaotic systems:

[0055] In general, to realize the complex evolutionary behavior and adjustable chaotic output of the system, a four-dimensional dynamic model is constructed using the following fractional differential form:

[0056]

[0057] In the above model, x(t), y(t), and z(t) represent three-dimensional spatially coupled state variables, and T(t) represents the system's local time-flow variables; α∈(0,1) represents the fractional order of the system, used to describe historical dependence and memory decay properties; σ is the spatial coupling coefficient, ρ is the nonlinear feedback coefficient, β is the dissipation coefficient, γ is the time tensor coupling strength, η is the time-flow feedback coefficient, κ is the spatiotemporal synchronization coefficient, λ is the time-flow decay coefficient, T is the local time-flow variable, and W... t It is a time-dependent feature tensor.

[0058] In some implementations, this fractional-order model is discretized using the definition of Caputo's differential, making it more suitable for solving practical problems in digital logic and hardware systems. To ensure the system possesses sufficient chaotic characteristics, the ranges of the variable parameters are set as follows:

[0059] α∈[0.8,0.9]: The non-integer order dynamic response of the control system evolution;

[0060] σ∈[8,12]: Adjusts the strength of the interaction between x and y, affecting the energy transfer path;

[0061] ρ∈[26,30]: As a nonlinear feedback factor, it regulates whether the system enters a chaotic state;

[0062] β∈[2.5, 3.5]: is the dissipation factor, which mainly controls the rate of system energy decay;

[0063] γ∈[0.1,0.5]: controls the degree of perturbation of spatial variables by the time-dependent tensor;

[0064] η∈[1.2,1.8]: describes the weights that describe the reaction of the local time stream to the spatial state;

[0065] κ∈[0.8,1.0]: is the coupling modulation factor, which determines the spatiotemporal dynamic relationship;

[0066] λ∈[2.3,2.7]: Adjusts the stability and self-feedback rate of T(t).

[0067] In a typical implementation, the time perturbation term W t As a coupling term in the time dimension, its construction form is as follows:

[0068]

[0069] in:

[0070] ω k ∈[0.1,0.5]: Exponentially decaying weighting coefficient, controlling the intensity of the disturbance;

[0071] τ k ∈[1,5]: Time scale adjustment parameter, which controls the decay rate;

[0072] ξ k A set of three orthogonal basis tensors that satisfy the orthogonality condition <ξ i ,ξ j >=δ ij And the Frobenius norm ||ξ k || F =1;

[0073] This tensor structure introduces non-stationary perturbations on the time axis, effectively increasing the unpredictability of the system.

[0074] Alternatively, tensor generation can be achieved using high-dimensional orthogonal transformation matrices, such as the Gram-Schmidt orthogonalization algorithm.

[0075] To ensure the system is in a strongly chaotic state from the initial state, the maximum Lyapunov exponent optimization method is generally used.

[0076] Initialize the parameter set Θ = {σ,ρ,β,γ,η,κ,λ};

[0077] Solving the maximum Lyapunov exponent λ based on orbital perturbation max (Θ);

[0078] Iterative updates are performed using the gradient ascent method:

[0079]

[0080] Where ∈ represents the learning rate, which can be dynamically adjusted based on system convergence. The optimized system state must satisfy λ. max ≥0.6 ensures that the entropy source of the output meets the requirements of high nonlinearity and high sensitivity.

[0081] In certain applications, chaotic systems can be implemented in hardware. For example:

[0082] The fractional derivative module is based on the discretization design of the Caputo formula and supports dynamic order control.

[0083] The tensor coprocessing unit implements the construction and real-time computation of WtWt, employing a three-dimensional tensor multiplication logic structure.

[0084] The parameter input module supports SPI / AXI communication interfaces, enabling synchronous interaction with the main control CPU or MCU.

[0085] During initialization, the module automatically loads the initial chaotic parameters via a DMA channel and executes the first round of chaotic sequence generation. The generated sequence is written to a local cache for subsequent use, ensuring that the system can seamlessly enter the data processing stage.

[0086] As the core initial module of the entire system, this chaotic dynamic system has the following connection with other steps:

[0087] The calculation of feature quantification indicators in the step-by-step data classification process relies on the distribution function generated by the chaotic entropy source for dynamic threshold adjustment;

[0088] The key generation operation in the distributed verification process uses the following formula:

[0089] K i =SHA3-512(x(t) i )⊕y(t i )⊕z(t i ));

[0090] Among them, t i This is a system time slice with an interval of 50ms;

[0091] The parameter feedback for the step-by-step closed-loop update depends on the current state of the chaotic system in order to achieve system-level adaptive closed-loop optimization update.

[0092] S2. Data classification and processing: Identify the type characteristics of the input data and dynamically select the corresponding entropy decomposition channel;

[0093] This implementation step is the data processing stage that begins after system initialization. It aims to achieve refined classification and recognition based on the modal features of the input data, and to select an appropriate entropy decomposition channel according to the identified feature structure. This step directly relies on the spatiotemporal chaotic entropy source constructed in the system initialization step (S1). Its output sequence not only participates in the dynamic adjustment of feature thresholds but also initializes and configures the perturbation templates required during feature extraction, thus forming a continuous data recognition and processing path.

[0094] Generally, after receiving the input data stream (denoted as D), the data classification module first obtains a dynamic perturbation parameter set from the chaotic entropy source module. This is used to enhance the feature responsiveness of multimodal data, and then proceeds to the modality recognition and channel selection submodule.

[0095] In this embodiment, multimodal feature extraction includes the following key processing steps:

[0096] In one possible implementation, the system extracts key structural features using a data type-based modal matching strategy, where:

[0097] For text data, we first construct a dependency syntax tree structure and calculate the average length of all syntax paths, defined as follows:

[0098]

[0099] in:

[0100] N represents the total number of syntactic edges;

[0101] l i This represents the path length corresponding to the i-th edge;

[0102] The judgment condition is: if the following conditions are met The initial assessment is that this is a text-based input.

[0103] For image data, a pre-trained convolutional neural network is used to extract intermediate layer feature maps, and its spectral radius R is calculated, defined as follows:

[0104] R = max{|λ i |∣λ i ∈Spec(F)};

[0105] Where F represents the convolutional feature map mapping matrix; λ i Let R be the set of eigenvalues ​​of the feature map; Spec(F) represents the spectrum of matrix F; and the structural similarity index SSIM(D) is used for joint judgment. If the condition is: R≥0.7∧SSIM(D)>0.7, then the input data is classified as an image modality.

[0106] For video data, in a typical implementation, the system calculates the optical flow field using multiple frames of images, and then estimates its fractal dimension D based on the Hausdorff method. f The expression is as follows:

[0107]

[0108] Where N(∈) represents the number of spheres required for the minimum coverage optical flow vector field; ∈ is the coverage scale; and the decision threshold is D. f ≥1.8.

[0109] In this step, based on the modal feature extraction results described above, the system selects subsequent entropy decomposition channels according to the following logical rules:

[0110]

[0111] In this rule, the priority of logical judgment follows the principle that data type identification takes precedence over feature determination. That is, the system first classifies the input modality by type, and then calculates the threshold function based on the classified modality to determine whether it meets the channel conditions.

[0112] Specifically, the system introduces chaotic entropy source variables constructed during the system initialization phase into the feature extraction process to modulate the input features:

[0113] In some embodiments, through perturbation transformation:

[0114] D′(t)=D(t)⊕(γW t +ηT(t));

[0115] Where D(t) is the original input data; ⊕ represents the XOR perturbation of the tensor level; γ is the time tensor modulation intensity, controlling the perturbation amplitude; W t Let the chaotic characteristic tensor be defined as: T(t) is a time-flow variable derived from the fourth term of the system's differential equations; η is a feedback coefficient used to adjust the strength of the time-flow feedback to the disturbance effect.

[0116] This perturbation process significantly improves the discriminative power of feature determination and enhances its adaptability to atypical data.

[0117] Alternatively, at the hardware implementation level, the data classification module is integrated into the high-frequency task sequence of the tensor coprocessor:

[0118] Operations such as convolutional spectrum extraction, dependency structure analysis, and optical flow estimation are all initialized using the distribution perturbation provided by the chaotic entropy source.

[0119] The coprocessor's internal tensor cache is preloaded with orthogonal basis tensors ξ. k To achieve W t Real-time construction;

[0120] The system receives the current input data stream from the main control MCU via the SPI bus and performs modulation processing;

[0121] In addition, in some scenarios, there is a DMA channel linkage mechanism between the modality determination module and the initialization module to ensure that the current state parameter set Θ={σ,ρ,β,γ,η,κ,λ} can still be quickly synchronized when the system is powered on again or the channel is switched.

[0122] The coupling mechanism between this step and the system initialization step (S1) is reflected in the following aspects:

[0123] The thresholds of 2.5, 0.7, and 1.8 required for modality recognition can all be dynamically fine-tuned based on the distribution density of the chaotic sequence output.

[0124] The selected perturbation tensor W t It originates from the exponentially decaying tensor structure defined in the initialization model;

[0125] The real-time values ​​of system state variables x(t), y(t), z(t), and T(t) are used as reference quantities for characteristic disturbances in characteristic judgment.

[0126] The parameter control mechanism works in tandem with the subsequent channel configuration to ensure the stability of the overall logic loop.

[0127] S3. Entropy decomposition processing: Based on the selected channel, perform irreversible information compression on the data and output a set of fragments that meet the entropy value standard.

[0128] In this embodiment, the entropy decomposition process is designed to perform irreversible information compression and structural perturbation on the input data based on the selected channels, generating a set of data fragments that conform to the entropy standard. The core of this step lies in achieving both structural destructiveness and feature traceability of the data through perturbation enhancement and feature coupling mechanisms. This ensures that the data meets entropy control indicators during compression while retaining sufficient feature information for subsequent processing. The entropy decomposition process selects and processes specific channels based on the chosen data type (e.g., text, image, or video).

[0129] Generally, the data entropy decomposition process includes different processing methods for three data modalities: text, images, and videos. Different decomposition mechanisms are set according to the characteristics of each data type to ensure that its entropy value meets the requirements.

[0130] In one possible implementation, for text data, entropy decomposition processing includes the following steps:

[0131] Construct the dependency syntax tree of the text;

[0132] A perturbation mapping matrix is ​​introduced to scramble the syntactic structure;

[0133] Information is compressed into a sparse embedding space through semantic mapping compression operations;

[0134] Finally, a projection mapping mechanism is applied to generate irreversible text fragments.

[0135] The output text fragments must meet the following entropy criteria:

[0136] H≤0.01bit / word;

[0137] Here, H represents the entropy value of the text fragment, measured in bits per word. The entropy value must be less than or equal to 0.01 to ensure high data compressibility and irreversibility.

[0138] This threshold ensures that while retaining a certain amount of information, the fragments are difficult to reconstruct into the original sentence structure.

[0139] In another implementation, for image data, the system employs Chaotic Perturbation Convolution (CD-CNN) combined with an image texture perturbation strategy for decomposition processing. Specifically, this includes:

[0140] Extract high-frequency structural features from the image and perform chaotic perturbation mapping;

[0141] Construct speckle perturbation maps to enhance the nonlinear distribution of data;

[0142] Evaluate its structural similarity index (SSIM) and mutual information entropy.

[0143] The processing result should meet the following two constraints:

[0144] SSIM(D)≤0.05;

[0145] Where D represents image data, and SSIM is an important indicator for measuring image quality. By controlling the SSIM value below 0.05, the chaotic characteristics of image fragmentation are preserved.

[0146] Mutual information entropy decreases, satisfying:

[0147] I(X;Y)=H(X)+H(Y)-H(X,Y)I(X;Y)=H(X)+H(Y)-H(X,Y)

[0148] Furthermore, during processing, it is necessary to verify the mutual information entropy of the image to further ensure that its entropy value meets the requirements. The verification formula for mutual information entropy is as follows:

[0149] I(X;Y)=H(X)+H(Y)-H(X,Y);

[0150] Where I(X;Y) represents the mutual information value between the original image data and the entropy decomposition result, H(X) and H(Y) are the individual information entropies of the image data and the corresponding entropy decomposition fragments, respectively, and H(X,Y) is the joint entropy value of the image data and N decomposition fragments. This step evaluates whether the decomposed fragments still retain information redundancy that can be used to reconstruct the original image by measuring the mutual information, so as to ensure that the image data meets the requirements of irreversible compression after entropy decomposition processing, while retaining key feature information to the maximum extent, which is convenient for subsequent structural verification and secure mapping.

[0151] In another possible implementation, for video data, the system constructs a video perturbation channel based on a spatiotemporal perturbation Hamiltonian mapping model:

[0152] First, optical flow features are extracted and their fractal dimensions are calculated.

[0153] Construct a perturbation cube representation structure;

[0154] Finally, by applying a temperature-controlled perturbation field, a local phase transition is induced, achieving irreversible compression.

[0155] The generated video fragments must meet the following peak signal-to-noise ratio (PSNR) limits:

[0156] PSNR≤10dB;

[0157] PSNR, or Peak Signal-to-Noise Ratio, is a metric used to measure the quality of video data after compression. A PSNR value of less than 10 dB indicates that the reconstructed quality of the video has been significantly reduced after entropy decomposition and structural perturbation compression, and the data content exhibits high irreversibility and structural fragmentation.

[0158] This limit ensures that the compressed result cannot be used to visually reconstruct the original video content.

[0159] In this embodiment, the channel selection for the entropy decomposition processing step depends on the data type identification result. Specifically, the system selects the corresponding entropy decomposition processing channel by determining the modality of the input data. If the input data is text, the system selects the text decomposition channel; if the input data is an image, the system selects the image decomposition channel; and if the input data is video, the system selects the video decomposition channel.

[0160] The connection between channel selection and entropy decomposition processing:

[0161] When selecting an entropy decomposition channel, the system first relies on the aforementioned data classification processing steps (such as modality recognition) to determine the decomposition channel based on the type of input data. At this point, the implementation method of entropy decomposition processing is closely related to the selected channel:

[0162] For text data, the system first constructs a dependency syntax tree, calculates the average path length, and then enters the text decomposition channel if the conditions are met.

[0163] For image data, the system extracts convolutional feature maps, calculates spectral radius and structural similarity index, and enters the image decomposition channel after meeting the conditions;

[0164] For video data, the system calculates the fractal dimension of the optical flow field, and if it meets the conditions, it enters the video decomposition channel.

[0165] This close integration of channel selection and decomposition process ensures that each data type can be fully processed in the appropriate entropy decomposition channel, generating an irreversible fragment set that meets the entropy value standard.

[0166] S4. Distributed verification: Collaborative verification of fragment sets is performed through edge computing nodes to generate auditable verification credentials;

[0167] After completing the entropy decomposition process (S3), the system enters the distributed verification step (S4). This step ensures the security, integrity, and auditability of the perturbation fragment sets generated by each entropy decomposition channel. This step aims to leverage the synergy between edge computing nodes and the central cloud platform to verify the structural consistency and content traceability of the irreversible fragment sets output by each channel, and generate verification credentials with auditing capabilities. This step is highly related to the chaotic entropy source construction mechanism in the system initialization step, as well as the modality recognition parameters and entropy decomposition channel selection mechanism provided by the data classification processing module.

[0168] Under normal circumstances, after completing the entropy decomposition process, the system will automatically call the distributed verification mechanism. The edge computing nodes will perform local verification according to the predetermined time sequence and key update strategy, and upload the intermediate verification results to the cloud for final convergence calculation to generate standardized verification credentials.

[0169] In this embodiment, the key material for collaborative verification comes from the chaotic entropy source sequence constructed during the system initialization step.

[0170] In one possible implementation, the system generates a key stream for verification based on the following time evolution function:

[0171]

[0172] Among them, K t The verification key generated at time t; α is a hash convergence function used to enhance non-linear key expansion; i The chaotic weighting factor defined for system initialization; ψ i (t) respectively correspond to the chaotic source variables x(t), y(t), z(t), T(t); ∈ t The perturbation residual term originates from the time-domain tensor perturbation W. t The residual distribution;

[0173] In general, the system uses K t Real-time updates are used to maintain unpredictability in the verification process, while avoiding the risk of data reconstruction caused by key reuse.

[0174] In this embodiment, the verification logic involving edge nodes can be abstracted as a tensor-level collaborative hash mapping process.

[0175] Specifically, the system denotes each set of fragments as a tensor S. i And process it according to the following steps:

[0176] First, normalize each fragment tensor:

[0177]

[0178] Where, μ i σ is the mean of the i-th fragment tensor; i The corresponding standard deviation;

[0179] Then, hash encoding is performed using the edge node collaborative mapping function:

[0180]

[0181] Among them, V i This is the local verification vector output by the edge node; The mapping function is typically a nonlinear compressed tensor mapping network; ⊕ denotes a bit-level hybrid perturbation operation;

[0182] In some embodiments, the mapping function employs a self-supervised tensor network structure, which has the ability to adaptively adjust the projection dimension, thereby improving the robustness of edge nodes in handling fragment differences.

[0183] In this embodiment, the cloud-based verification convergence process is implemented based on a tensor merging mechanism and a symmetric consistency metric.

[0184] In one specific implementation, the local verification vector V uploaded by all edge nodes... i It is incorporated into the global verification tensor G for fusion processing:

[0185] The system further constructs a symmetric tensor kernel function. Used to compare the consistency between the global tensor and the results of each local tensor:

[0186]

[0187] Among them, Λ i For consistency score; <·,·> for tensor inner product; ||·|| for tensor L2 norm; system-defined threshold Λ i A value of ≥0.85 is considered passing; a value below the threshold triggers a re-verification process and is recorded in the audit log.

[0188] In this embodiment, the algorithm for generating the final verification credential includes the following processing flow:

[0189] As an option, the system is based on timestamp T s Generate auditable verification credentials from the key index κ and the verification tensor digest value δ:

[0190] Π = Enc SHA-512 (T s ||κ||δ||θ);

[0191] Where Π represents the final generated verification credential; || represents the concatenation operation; and θ represents the current system status summary, including system load, link latency, temperature disturbances, etc. This is the hash digest value of the global verification tensor; this verification credential is traceable and tamper-proof. All verification credentials will be recorded in the system's distributed security chain (DSC). This mechanism draws on the audit chain structure in "Quantum Key Distribution (QKD)," but this system does not rely on real quantum physics mechanisms. Instead, it achieves verification security by perturbing the hash key stream and using a chaotic weight function, ensuring that each verification is traceable and tamper-proof.

[0192] The distributed verification step plays a crucial role in ensuring security and verifying data throughout the entire system process, and it has the following relationship with the aforementioned modules:

[0193] The verification key used originates from the chaotic entropy source generated during the initialization phase, ensuring the unity of time continuity and randomness;

[0194] The verification process applies to the fragment set {S} i The processing of the output structure is based on entropy decomposition to ensure that the verification input has a standardized tensor format;

[0195] The perturbation function used in the collaborative verification process and the tensor perturbation model W introduced in the entropy decomposition stage. t Maintain consistency;

[0196] The final verification credential Π will be embedded in the access authorization process of subsequent data reconstruction steps, serving as a chain authentication mechanism.

[0197] S5. Closed-loop update: The verification credentials are written to the blockchain, and the system's chaotic parameters are updated based on the evidence storage results to form a feedback control closed loop.

[0198] After completing the distributed verification step (S4), the system enters the closed-loop update step (S5). This step aims to ensure the long-term traceability of the verification results and to perform feedback control on the system based on the stored results, further optimizing the chaotic entropy source and other key parameters. By writing the verification credentials into the blockchain, the system can establish an immutable record chain and update the system's chaotic parameters according to the verification credentials, forming an adaptive closed-loop feedback mechanism to ensure that the system is always under effective control.

[0199] In one possible implementation, the system will write the generated verification credentials to the blockchain through the following steps:

[0200] Generate blockchain evidence: After the verification credential is generated, the system first calculates its hash value to ensure data consistency and tamper-proof capability. Then, this hash value is written to the blockchain along with relevant verification timestamps, verification key indexes, and system state summaries, as part of the blockchain transaction. This step not only ensures the secure storage of the verification credential but also provides a reliable basis for subsequent auditing and traceability.

[0201] Π block =SHA-256(Π||T) s ||κ||θ);

[0202] Among them, Π block This represents the evidence hash on the blockchain, where Π is the verification credential, and T... s κ is the timestamp, θ is the verification key index, and θ is the system's current state digest. The concatenated evidence is then hashed to ensure the information is tamper-proof.

[0203] Blockchain Storage and Verification: The stored information is disseminated through the blockchain network and ultimately verified and confirmed by each node. Once the information is written to the blockchain, any third party can access the stored information and verify the validity of the credential based on the information recorded in the blockchain. This process provides the system with robust auditing capabilities, ensuring that data is traceable and tamper-proof at every stage from generation to verification.

[0204] After completing the blockchain notarization, the system adjusts key system parameters such as the chaotic entropy source based on the notarization results. Specifically, the system dynamically updates chaotic parameters using a feedback control algorithm based on the verification credentials on the blockchain and the current operating state of the system, ensuring the stability and security of the system.

[0205] Under normal circumstances, the system will adjust the chaotic entropy source parameters based on the following feedback formula:

[0206] X t+1 =X t +γ·ΔX;

[0207] Among them, X t X represents the chaotic state parameters of the system at the current moment. t+1 The new chaotic parameters are represented by γ after feedback adjustment; γ is the feedback adjustment coefficient; ΔX is the adjustment amount calculated based on the verification results, usually equal to the verification consistency score Λ. i It is associated with the system state summary θ.

[0208] By continuously updating the chaos parameters, the system can adaptively optimize the entropy source disturbance in the entropy decomposition process, thereby enhancing the security and anti-interference capability of the data processing process.

[0209] In one possible implementation, the feedback mechanism involves the following steps:

[0210] Consistency score feedback: When the evidence stored in the blockchain is verified and the consistency score Λ is obtained... i When the consistency score exceeds a preset threshold, the system will fine-tune the disturbance of the chaotic source based on this information. If the consistency score falls below the threshold, the system needs to trigger a backtracking check or re-verification process to ensure the stability of the system.

[0211] Chaotic perturbation and optimization: The system adjusts the perturbation amplitude of the entropy source based on feedback from each verification process. Specifically, if the system detects an anomaly in the verification result, it may increase the perturbation of the entropy source to increase the randomness in the entropy decomposition process; conversely, it may reduce the perturbation to reduce unnecessary computational resource consumption.

[0212] Through this feedback mechanism, the system can continuously optimize performance during entropy decomposition and verification processes, ensuring that each data processing and verification process meets design standards and continuously adapts to external changes.

[0213] Example of closed-loop feedback control:

[0214] For example, when the system detects an anomaly in a certain verification stage (such as verification failure or low consistency score), the system automatically adjusts the perturbation parameter X of the chaotic entropy source, for example, by increasing the perturbation range to enhance the randomness of the data. In this way, the fragments generated in subsequent entropy decomposition processing steps will be more unpredictable, thereby enhancing the system's resistance to external attacks and ensuring data security and integrity.

[0215] It also includes measures to resist highly complex attacks:

[0216] A high-entropy random number generator is integrated into the entropy decomposition process, with a generation rate ≥100Mbps;

[0217] Embed lattice-based digital signatures in the verification credentials of the distributed verification step.

[0218] During entropy decomposition (S3) and distributed verification (S4), the system's resistance to high-complexity computation attacks is designed to ensure high security and unpredictability in data processing. Protection against high-complexity computation attacks is primarily achieved by integrating a high-entropy random number generator and lattice-based secure digital signature technology. The high-entropy random number generator generates high-speed, high-entropy random numbers, providing the system with a stronger entropy source. Meanwhile, the lattice-based secure digital signature technology is used to verify the signature embedded in the credential, enhancing the data's tamper-proof capabilities.

[0219] In system design, the need to resist high-complexity computational attacks permeates every step of entropy decomposition, distributed verification, and closed-loop update. A high-entropy random number generator plays a crucial role in the entropy decomposition process, providing not only a high-quality random number source but also sufficient unpredictability for subsequent key steps such as key generation and verification. Meanwhile, lattice-based digital signatures are embedded in the verification credentials generated by distributed verification, further enhancing the system's resistance to high-complexity computational attacks.

[0220] In this embodiment, the system integrates a high-entropy random number generator. Its working principle is based on the unpredictability of physical phenomena such as noise, generating random numbers with a sufficiently powerful entropy source to meet the high randomness requirements of data processing and security protocols. Specifically, the random number generation rate of the QRNG is set to ≥100Mbps to ensure high-speed and efficient data processing, meeting the high-frequency requirements of modern data transmission.

[0221] In some embodiments, the generation process of a high-entropy random number generator can be implemented in the following ways:

[0222] Random bit sequences are generated through interference phenomena;

[0223] These bit sequences are converted into binary random numbers and then post-processed (e.g., using a hash function) to improve the quality of the random numbers.

[0224] These generated random numbers play a role in various modules of the system, such as generating keys, random perturbation, and encryption. Their randomness and unpredictability can effectively resist traditional classical attacks as well as future high-complexity computational attacks.

[0225] To further enhance the system's resistance to high-complexity computational attacks, a lattice-based digital signature is embedded in the verification credential generated during the distributed verification process. Lattice-based digital signatures are a public-key cryptographic technique that leverages the difficulty of lattice problems to ensure the security of the encryption system.

[0226] In this embodiment, the lattice-based signature generation process can be as follows:

[0227] During the generation of verification credentials, verification data (such as verification results, timestamps, system status summaries, etc.) is combined with a pair of lattice-based keys;

[0228] Public-key cryptography is used to sign verification credentials to ensure their integrity and immutability.

[0229] The specific mathematical formal expression is as follows:

[0230] Π sig =Sign lattice (Π), where Π sig Indicates a verification credential with a coded signature; Sign lattice (·) denotes a lattice-based signature function, which typically employs the LWE (Learning with Noise) problem to ensure the security of the encryption system against high-complexity attacks.

[0231] Lattice-based signatures are resistant to high-complexity attacks, effectively preventing complex computational models from exploiting weaknesses in traditional public-key encryption algorithms. Therefore, by embedding them into credentials generated through distributed verification, the security and integrity of the verification process can be ensured, avoiding the risk of tampering during transmission or storage.

[0232] During distributed verification, the system effectively protects the verification information transmitted between nodes through the randomness provided by a high-entropy random number generator and the encryption protection of lattice-based digital signature technology. These protection measures, combined with the key generation mechanism and fragment verification in the previous entropy decomposition processing step, ensure the overall system's ability to withstand highly complex computational models.

[0233] Generally, high-entropy random number generators use random numbers to dynamically generate verification keys and perturbation functions, thereby enhancing the unpredictability of the key stream and ensuring that the verification key cannot be reverse-engineered even under the threat of highly complex attacks. Furthermore, lattice-based signature technology provides additional security in the generated verification credentials, guaranteeing their authenticity and immutability.

[0234] In this embodiment, by integrating a high-entropy random number generator and lattice-based digital signature technology, the system achieves multiple layers of protection against high-complexity attacks. The high-entropy random number generator ensures the high randomness and unpredictability of the system's internal keys and parameters, thereby enhancing the system's resistance to high-complexity attacks. Meanwhile, the lattice-based digital signature ensures the immutability and long-term verifiability of the verification credentials, providing robust anti-tampering protection.

[0235] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A data fragmentation method based on a four-dimensional manifold chaotic dynamical system, characterized in that, Includes the following steps: System initialization: Configure and activate the four-dimensional manifold chaotic dynamic system, generating a spatiotemporally coupled chaotic entropy source; Data classification processing: Identify the type characteristics of the input data and dynamically select the corresponding entropy decomposition channel; Entropy decomposition processing: Based on the selected channel, the data is irreversibly compressed to output a set of fragments that conform to the entropy value standard; Distributed verification: Collaborative verification of fragment sets is performed by edge computing nodes to generate auditable verification credentials; Closed-loop update: The verification credentials are written to the blockchain, and the system's chaotic parameters are updated based on the evidence storage results to form a feedback control closed loop.

2. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, The four-dimensional manifold chaotic dynamical system in the system initialization step is defined by the following set of fractional differential equations: ； in, It is the fractional order of the differential. The spatial coupling coefficient is... For nonlinear feedback coefficients, Here is the dissipation coefficient. The time tensor coupling strength, For time-flow feedback coefficients, For spatiotemporal synchronization coefficient, The time-flow attenuation coefficient, For local time stream variables, It is a time-dependent feature tensor.

3. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, The data classification and processing steps include: Multimodal feature extraction: For input data stream Calculate the following characteristic indicators respectively: Text data: Calculate the average path length of the dependency parsing tree, if it satisfies... If the value is ≥2.5, then the input data is preliminarily determined to be text modal input. Image data: Extract the spectral radius of the convolutional feature map and calculate the input data stream. Structural similarity index If satisfied ≥0.7 and A value of 0.7 indicates that the input data is initially determined to be an image modality. Video data: Calculating the optical flow field using multiple frames of images, and estimating the fractal dimension of the optical flow field based on the Hausdorff method. If satisfied If the value is ≥1.8, then the input data is preliminarily determined to be a video modality. Routing decision execution: Select the entropy decomposition channels according to the following logical order: ； The logical operator precedence is as follows: data type matching is performed first, followed by channel selection. The average path length of the dependency syntax tree. denoted as the spectral radius of the convolutional feature map. For input data stream.

4. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, The entropy decomposition process includes the following steps when selecting text decomposition channels: Construct the dependency syntax tree of the text; A perturbation mapping matrix is ​​introduced to scramble the syntactic structure; Information is compressed into a sparse embedding space through semantic mapping compression operations; Irreversible text fragments are generated by applying a projection mapping mechanism, wherein the text fragments satisfy an entropy value ≤ 0.01 bit / word.

5. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, The entropy decomposition process includes the following steps when selecting image decomposition channels: A decomposition process is performed using chaotic perturbation convolution combined with an image texture perturbation strategy. Extract high-frequency structural features from the image and perform chaotic perturbation mapping; A speckle perturbation map is constructed to enhance the nonlinear distribution of the data, so that the generated fragments satisfy the structural similarity index SSIM≤0.05; Verify the mutual information entropy of the decomposition results.

6. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, The entropy decomposition process includes the following steps when selecting video decomposition channels: Constructing a video perturbation channel based on a spatiotemporal perturbation Hamiltonian mapping model; Extract optical flow features and calculate the fractal dimension of the optical flow field; Construct a perturbation cube representation structure; Applying a temperature-controlled perturbation field induces a local phase transition, achieving irreversible compression; The output compressed video fragments must satisfy the peak signal-to-noise ratio (PSNR) ≤ 10dB.

7. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, In the distributed verification step: Key materials required for collaborative verification: Based on the chaotic entropy source generated during the system initialization steps, verification keys are generated in a time series. Algorithm for generating auditable verification credentials: Tensor operations are performed collaboratively between edge nodes and the cloud to generate the final verification credentials.

8. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, In the closed-loop update step: Blockchain-based evidence storage: Verification credentials generated during the distributed verification process are written to the blockchain and stored using an improved Merkle-Patricia tree structure; Parameter closed-loop update: Adjust the parameters of the chaotic system based on the evidence storage results.

9. The data fragmentation method based on a four-dimensional manifold chaotic dynamical system according to claim 1, characterized in that, It also includes measures to resist highly complex attacks: A high-entropy random number generator is integrated into the entropy decomposition process, with a generation rate ≥100 Mbps; Embed lattice-based digital signatures in the verification credentials of the distributed verification step.

Images