A Data Security Management Method Based on Chinese Cryptographic Algorithms

By adopting a data security management method based on national cryptographic algorithms, the format protection and security of data during cross-domain transfer are improved, solving the problems of data format corruption and network jitter, and enhancing the security and adaptability of the data management system.

CN123123015APending Publication Date: 2026-11-17SHAANXI BINCHANG XIAOZHUANG MINING CO LTD +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611578019.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-13
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In existing technologies, the data anonymization mechanism destroys data format characteristics when data is transferred across domains, leading to downstream parsing failures. Centralized verification has poor resistance to network jitter, plaintext data storage poses a risk of secondary leakage, and there is a lack of dynamic authentication and same-source anti-tampering audit mechanisms.

Method used

A data security management method based on national cryptographic algorithms is adopted. Data is diverted and encrypted through security gateway nodes, generating streaming commitment and rule semantic commitment graphs. Data is encapsulated in an out-of-order manner using dynamic policy state-derived keys and persisted on untrusted storage nodes. The authorization and authentication center node issues query view tokens, and the de-identified proxy layer node performs blind verification and memory decryption, generating downgraded placeholder text to ensure data security.

Benefits of technology

Without increasing communication overhead, out-of-order encapsulation of data packets is achieved, which improves the security of data in untrusted storage nodes, enhances the system's adaptability in network jitter environments, reduces the risk of secondary leakage caused by plaintext disking, and ensures the parsing continuity of downstream business applications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123123015A_ABST
    Figure CN123123015A_ABST
Patent Text Reader

Abstract

This application relates to the field of data information security technology and discloses a data security management method based on national cryptographic algorithms. The method includes: a gateway node splitting plaintext data into high-frequency data streams and low-frequency structured data based on the acquisition frequency, generating streaming commitment and rule-based semantic commitment graphs respectively; encrypting the low-frequency data using a dynamically derived key based on a deduced strategy state, generating a deterministic permutation sequence to scramble the ciphertext, combining various commitments to generate source evidence anchors, and uploading them along with the ciphertext to a storage node; a proxy node obtaining the scrambled ciphertext and source evidence anchors, performing data verification to restore the ciphertext order, performing decryption or downgraded placeholder operations based on a query view token, and assembling and outputting the target business view. By physically scrambling the ciphertext, logical speculation based on offsets is prevented; a compensation verification mechanism adapts to network transmission jitter; and the original data format and statistical characteristics are maintained even in unauthorized access states, ensuring the continuity of downstream business parsing.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of data information security technology, specifically a data security management method based on national cryptographic algorithms. Background Technology

[0002] In the process of heterogeneous data aggregation, storage, and cross-domain sharing, data security is typically ensured through desensitization and encryption mechanisms. Existing static desensitization mechanisms, when masking sensitive information, can easily disrupt the original data format and statistical distribution characteristics, leading to parsing failures in downstream application systems upon receiving the data. Traditional centralized decryption and verification mechanisms are highly dependent on the stability of the network transmission environment and lack adaptability to network fluctuations such as physical link packet loss and latency, easily resulting in the false interception of compliant data.

[0003] Furthermore, existing data access controls often employ static or coarse-grained permission models, lacking dynamic authentication methods that are deeply integrated with specific business scenarios and query sessions. Simultaneously, in long-term data storage and cross-domain calls, there is a lack of a common-origin anti-tampering audit mechanism that integrates data encryption, de-identification rules, and time-series status. Moreover, when existing data proxy servers perform de-identification and reconstruction operations, the plaintext data is typically cached locally on disk, further increasing the security risk of secondary data leakage at the proxy end.

[0004] Therefore, this invention proposes a data security management method based on national cryptographic algorithms to address the shortcomings of existing technologies. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a data security management method based on national cryptographic algorithms. This method solves the problems of data format failure caused by the desensitization mechanism damaging data format characteristics during cross-domain data transfer, as well as the poor anti-network jitter capability of centralized verification and the risk of secondary leakage due to plaintext storage on disk.

[0006] To address the above problems, the present invention provides the following technical solution:

[0007] The first aspect of this invention provides a data security management method based on national cryptographic algorithms, applied to a data security management system comprising a security gateway node, an untrusted storage node, an authorization and authentication center node, and a de-identification proxy layer node, comprising the following steps:

[0008] The security gateway node receives plaintext data reported by physical devices, extracts the frequency parameters of data frames to perform data splitting; for high-frequency data streams, it generates streaming commitments; for low-frequency structured data, it combines the desensitized dictionary version and runtime window to deduce the dynamic policy state, extracts rule dependency logic relationships, calculates and generates a rule semantic commitment graph and cross-field association consistency proof.

[0009] The security gateway node uses the dynamic policy state derived key to encrypt the plaintext field, drives the random number generator to generate a deterministic permutation sequence with the dynamic policy state, and performs a shuffling operation on the ciphertext array; combines the commitments and proofs generated in the above steps to generate a common source evidence anchor point, and uploads the disordered mixed ciphertext to the untrusted storage node;

[0010] The authorization and authentication center node receives the query request initiated by the application client, calculates the minimum rule slice that meets the request context, and uses an asymmetric algorithm to digitally sign the request identifier and context parameters to generate a query view token, which is then sent to the de-identification proxy layer node.

[0011] The de-identified proxy layer node obtains out-of-order mixed ciphertext data from the untrusted storage node, deduces and verifies the state within the volatile cache area, and sequentially performs blind verification of cross-time window continuity proof, cross-field association consistency proof, and rule consistency proof labels; after successful verification, the deterministic permutation sequence is regenerated to restore the ciphertext order; for ciphertext with authorization credentials, in-memory decryption is performed, and for ciphertext without authorization credentials, downgraded placeholder text is synthesized in memory according to the message template requirements; the decrypted plaintext and placeholder text are concatenated and assembled into the target business view object output, and the record view output commitment is calculated and recorded.

[0012] This invention introduces a deductive state derivation and deterministic permutation mechanism at the edge gateway side to achieve out-of-order encapsulation of data physical order without increasing communication overhead, preventing eavesdroppers from inferring business logic through message offsets. Simultaneously, the de-identification proxy layer performs compensation verification in memory using a tolerable drift interval, achieving blind verification to cope with network jitter; in the absence of verification credentials, it synthesizes a degradation placeholder based on a hash derivation seed, ensuring that data without access privileges maintains its original communication format and statistical distribution characteristics in the de-identified state, improving the availability and parsing consistency of the output view in downstream business applications.

[0013] Furthermore, the step of performing data diversion includes: parsing the acquisition frequency of the data frame and comparing it with a preset frequency threshold; when the acquisition frequency is greater than or equal to the preset frequency threshold, it is determined to be a high-frequency data stream and routed to the streaming commitment generation path; when the acquisition frequency is less than the preset frequency threshold and contains structured business fields, it is determined to be low-frequency, high-sensitivity structured data and routed to the multidimensional semantic commitment generation path.

[0014] Furthermore, the step of generating streaming commitments includes: extracting high-frequency data buffer blocks within the current physical time window according to a set time window span; and performing an overall hash operation on the globally unique identifier of the high-frequency data stream, the current physical time window sequence number, and the hash value of the high-frequency data buffer block using a hash algorithm to generate a lightweight streaming commitment.

[0015] Furthermore, the processing steps for low-frequency structured data specifically include: using a hash algorithm to concatenate and calculate the global policy version, the de-identified dictionary version, the device group identifier, and the current time window sequence number to deduce the dynamic policy state; using the dynamic policy state as a key parameter, calculating the message authentication code for the physical address, physical location number, constraint identifier, and binding result of devices with topological dependencies to generate the cross-field association consistency proof; extracting the visibility level of each plaintext field to generate a single rule digest; serializing the rule dependency logic relationship matrix reflecting rule inheritance and priority, and performing an aggregation operation with each single rule digest to generate the rule semantic commitment graph; extracting the message structure template identifier and attribute format constraints, calculating and generating a structural syntax commitment, and incorporating this structural syntax commitment into the combined calculation of the same-origin evidence anchor points.

[0016] Furthermore, the encryption and shuffling operation steps include: using the dynamic strategy state as the initial vector parameter to drive the counter mode of the commercial cryptographic standard symmetric algorithm to generate a pseudo-random bit stream; using the pseudo-random bit stream to drive a preset shuffling algorithm to perform a synchronous shuffling operation on the physical location of the ciphertext array and the corresponding single-rule digest, thereby removing the physical order mapping association between the ciphertext data and the original heterogeneous message topology.

[0017] Furthermore, the step of generating the same-source evidence anchor point includes: performing a hash ratchet iteration operation on the historical cross-time window continuity proof of the previous time window, the overall hash value of the current out-of-order mixed ciphertext array, and the current time window sequence number to generate the cross-time window continuity proof of the current time window; after performing placeholder alignment processing based on a preset commitment coverage graph, performing aggregate concatenation hash calculation on the streaming commitment, rule semantic commitment graph, structure syntax commitment, cross-time window continuity proof, and the de-identified dictionary version to generate the globally unique same-source evidence anchor point and write it into the trusted audit ledger.

[0018] Furthermore, the step of generating the query view token includes: obtaining the identity identifier of the access request subject, the business query purpose identifier, the target view identifier, and the current scenario identifier; generating a query session random factor uniquely bound to this request in real time using a hardware true random number generator; and performing a digital signature operation on the identity identifier, target view identifier, business query purpose identifier, current scenario identifier, minimum rule slice digest, token expiration timestamp, and query session random factor using an asymmetric algorithm private key to generate the query view token.

[0019] Furthermore, the steps of deducing and verifying the state and performing blind verification include: establishing a memory-locked isolated execution region on the desensitized proxy layer node; performing hash matching based on the nominal time window parameter; when the nominal time window blind verification fails, extracting the adjacent time window state parameter within the preset tolerance drift interval as a compensation parameter, and re-executing the cross-time window hash ratchet operation to perform anti-jitter blind verification.

[0020] Furthermore, the step of synthesizing downgraded placeholder text in memory includes: generating a dedicated placeholder synthesis seed that is valid within the lifecycle of the current query view based on the query view token, the current field identifier, the summary of the single rule to which it belongs, and the scene identifier using hash derivation; using the placeholder synthesis seed as a driving parameter to synthesize mask text with a character set and length consistent for unweighted fields that hit the format generalization rule; and synthesizing test waveform placeholder data for unweighted fields that hit the fake data replacement rule in combination with a preset statistical feature template.

[0021] Furthermore, the step of calculating the view output commitment includes: extracting the identifiers of each underlying time window participating in this view assembly, the same-source evidence anchor points corresponding to each plaintext and ciphertext block, and the logical assembly sequence, and performing serialization aggregation hash calculation to generate a multi-packet assembly consistency commitment; extracting the set of rule summaries actually hit within the lifecycle of this view reconstruction to generate a dynamic rule hit execution summary; and performing a hash operation on the target business view object, the query view token, the multi-packet assembly consistency commitment, and the dynamic rule hit execution summary to generate the final view output commitment and record it in the trusted audit ledger.

[0022] A second aspect of this invention provides a data security management system based on national cryptographic algorithms, the system comprising:

[0023] The security gateway node is used to receive plaintext data reported by physical devices, perform traffic splitting and multidimensional semantic commitment extraction, encrypt the ciphertext with derived keys, perform deterministic scrambling on the ciphertext, perform combined hash calculation to generate same-source evidence anchor points and output ciphertext data.

[0024] An untrusted storage node is used to receive and persistently store out-of-order mixed ciphertext data output by the security gateway node;

[0025] The authorization and authentication center node is used to receive data access requests, calculate the minimum rule slice, and issue query view tokens carrying the request context signature using an asymmetric cryptographic algorithm.

[0026] The de-identified proxy layer node is used to obtain out-of-order mixed ciphertext data and query view tokens, perform anti-jitter blind verification in memory, reverse the permutation to restore the business order, complete plaintext decryption and placeholder text synthesis for unauthorized fields, assemble and output the target business view object.

[0027] Furthermore, the security gateway node internally includes: a traffic splitting module, used to analyze the acquisition frequency parameters of the received data and perform splitting of high-frequency and low-frequency data processing paths; a semantic commitment generation module, used to generate streaming commitments, cross-field association consistency proofs, rule semantic commitment graphs, and structure syntax commitments; and an encapsulation and fusion module, used to derive keys to perform encryption, drive pseudo-random sequences to perform physical location scrambling, and perform hash ratcheting and aggregation calculations to generate the same-origin evidence anchor points.

[0028] Furthermore, the desensitized proxy layer node includes: a blind verification module, used to perform deduction within the established volatile memory isolation region using the adjacent time window parameters within the nominal time window or the tolerance drift interval, to verify the legality of each continuity proof and consistency proof label; and a structure reconstruction module, used to reverse restore the ciphertext array order after the blind verification passes, decrypt and restore the authorized fields, dynamically synthesize placeholder text for unauthorized fields based on exclusive placeholders as seeds, and generate a view output commitment containing a multi-package assembly consistency commitment.

[0029] A third aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described data security management method based on national cryptographic algorithms.

[0030] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described data security management method based on national cryptographic algorithms.

[0031] This invention provides a data security management method based on national cryptographic algorithms. It has the following beneficial effects:

[0032] 1. This invention analyzes the collection frequency of plaintext data through a secure gateway node, dividing the data into high-frequency data streams and low-frequency structured data, and generating streaming commitment and rule-based semantic commitment graphs accordingly. This mechanism can match the corresponding processing path based on data characteristics, reducing the processing overhead of massive high-frequency data while ensuring the control granularity of security management for low-frequency sensitive data, thus improving the overall processing efficiency of the data management system.

[0033] 2. This invention utilizes a dynamically derived key based on a deduced strategy state to encrypt data, generates a deterministic permutation sequence to shuffle the physical order of the encrypted data, and finally combines various commitments to generate anchor points for shared evidence. This process achieves out-of-order encapsulation of data packets without adding extra communication overhead, preventing external eavesdroppers from inferring the original business logic through fixed packet offsets or physical locations, effectively improving data security in untrusted storage nodes.

[0034] 3. This invention introduces adjacency time window compensation and verification logic based on a tolerable drift interval at the proxy node, and performs ciphertext order restoration in memory. For data without access permissions, it dynamically generates downgraded placeholder text based on the query view token. The compensation and verification mechanism improves the system's adaptability to network packet loss or transmission delay environments and reduces the false interception of compliant data; memory-level data reconstruction avoids the secondary leakage risk caused by plaintext being written to disk; the generation of downgraded placeholders allows data without access permissions to maintain its original message format and statistical distribution characteristics while masking the real information, ensuring the continuity of downstream business application parsing. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the data security management system architecture based on national cryptographic algorithms according to an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of the data security management method based on national cryptographic algorithms according to an embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram illustrating the working principle of data two-dimensional splitting and multi-dimensional semantic commitment extraction in an embodiment of the present invention;

[0038] Figure 4 This is a schematic diagram illustrating the deterministic out-of-order encapsulation and homologous evidence anchor point generation principle of an embodiment of the present invention;

[0039] Figure 5 This is a schematic diagram illustrating the principle of pure memory-state adaptive blind verification and structurally compatible reconstruction in an embodiment of the present invention.

[0040] Figure 6 This is a schematic diagram of the cumulative distribution curve of end-to-end response latency in a high-concurrency scenario according to an embodiment of the present invention;

[0041] Figure 7 This is a schematic diagram of high-fidelity frequency domain comparison of target range data in an embodiment of the present invention.

[0042] Figure 8 This is a schematic diagram of the robustness test for industrial network jitter tolerance verification in an embodiment of the present invention.

[0043] Among them, 10 is the security gateway node; 110 is the traffic splitting module; 120 is the semantic commitment generation module; 130 is the encapsulation and fusion module; 20 is the untrusted storage node; 30 is the authorization and authentication center node; 310 is the authorization module; 40 is the de-identification proxy layer node; 410 is the blind verification module; and 420 is the structure reconstruction module. Detailed Implementation

[0044] 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.

[0045] See attached document Figure 1 This invention provides a data security management method based on national cryptographic algorithms, which is applied to a data security management system based on national cryptographic algorithms. The system includes: a security gateway node 10, an untrusted storage node 20, an authorization and authentication center node 30, and a de-identification proxy layer node 40.

[0046] Security gateway node 10 is deployed at the underlying control ring network aggregation layer to receive plaintext data sent by field sensing devices and encapsulate it. Security gateway node 10 internally includes a traffic splitting module 110, a semantic commitment generation module 120, and an encapsulation and fusion module 130. Traffic splitting module 110 analyzes the acquisition frequency parameters of the received data and performs data processing path splitting. Semantic commitment generation module 120 extracts the attribute parameters of the structured message and the global policy version, generating hash credentials and logical constraint proofs bound to the policy state. Encapsulation and fusion module 130 derives a key to perform encryption and decryption operations on the fields, generates a pseudo-random sequence based on the deduced state to change the physical order of the fields, and aggregates various calculated credentials to produce common-source evidence anchors.

[0047] Untrusted storage node 20 is deployed in the aggregation layer computer room or cloud data center to receive and persist the encrypted data set output by the security gateway node 10.

[0048] The authorization and authentication center node 30 is deployed in the network management area and is used to manage global access control policies and the de-identification rule dictionary. The authorization and authentication center node 30 internally includes an authorization module 310. The authorization module 310 obtains the identity attributes of the access subject and data request context parameters, calculates rule slices, and issues query view tokens using an asymmetric cryptographic algorithm.

[0049] The de-identification proxy layer node 40 is deployed at the front end of the data center application service to handle encrypted parsing and view assembly. Internally, the de-identification proxy layer node 40 includes a blind verification module 410 and a structure reconstruction module 420. The blind verification module 410 performs cryptographic verification within the server's volatile storage area, verifying the legality of the time window continuity proof and rule consistency proof tags. The structure reconstruction module 420, based on the authentication token's permission parameters and a preset message template, completes the decryption and recovery of plaintext fields and the synthesis and assembly of unauthorized field placeholder text in memory.

[0050] See attached document Figure 2This invention provides a data security management method based on national cryptographic algorithms, comprising the following steps:

[0051] S100, the security gateway node 10 receives plaintext data reported by the physical device, extracts the frequency parameters of the data frame and performs data splitting; for high-frequency data streams, it generates a lightweight streaming commitment; for low-frequency structured data, it extracts the currently effective de-identification dictionary version and device group identifier, combines the running time window to deduce the dynamic policy status, extracts the visibility level of plaintext fields and the rule dependency logic relationship, and calculates the rule semantic commitment graph and cross-field association consistency proof.

[0052] S200, the security gateway node 10 uses the dynamic policy state derived key to perform symmetric encryption on the plaintext field, uses the dynamic policy state as the initial vector to drive the random number generator to generate a deterministic permutation sequence, and performs a shuffling operation on the ciphertext array; it combines the lightweight streaming commitment, the rule semantic commitment graph, the structure syntax commitment, and the cross-time window continuity proof to generate the same source evidence anchor point, and uploads the generated disordered mixed ciphertext to the untrusted storage node 20;

[0053] S300, when the application client initiates business data access, the authorization and authentication center node 30 receives the query request message, obtains the identity identifier of the requester, the business query purpose identifier, the target view identifier, and the current scenario identifier; calculates the minimum rule slice that conforms to the request context, uses the asymmetric algorithm private key to digitally sign the identifier and parameters to generate a query view token, and sends the query view token and the minimum rule slice to the de-identification proxy layer node 40.

[0054] S400, the de-identification proxy layer node 40 obtains the mixed ciphertext data corresponding to the request from the untrusted storage node 20, and simultaneously obtains or parses the current physical time window sequence number, de-identification dictionary version, and underlying physical device group identifier of the data source corresponding to the mixed ciphertext data. It obtains the nominal time window parameters or adjacent time window parameters within the tolerance drift interval in the server cache area to deduce the dynamic policy state, and sequentially verifies the cross-time window continuity proof, cross-field association consistency proof, and rule consistency proof label. After the verification is passed, the same deterministic permutation sequence is regenerated according to the dynamic policy state to restore the initial order of the ciphertext array. For ciphertext with authorization credentials, the decryption operation is performed. For ciphertext without authorization credentials, the minimum rule slice content and message template requirements are read to synthesize the downgraded placeholder text in memory. The plaintext output by decryption and the generated placeholder text are concatenated and assembled into the target business view object and output to the application client. The execution commitment summary of the view object is calculated and recorded in the trusted audit ledger.

[0055] To further clarify the implementation of each technical aspect of the present invention, the following will provide a detailed description of the implementation of each functional module involved above and its internal processing flow.

[0056] See attached document Figure 3 The security gateway node 10 receives plaintext data reported by the underlying physical device and performs two-dimensional data splitting and multi-dimensional semantic commitment extraction operations. This process is specifically implemented through the following sub-steps:

[0057] S110, the security gateway node 10 receives data frames reported by the underlying physical devices through the splitting module 110. In this embodiment, the underlying physical devices include underground gas monitoring substations and field acquisition devices such as microseismic sensors. To avoid imbalance in the allocation of system computing resources due to mixed processing of different types of data, the splitting module 110 prioritizes parsing the acquisition frequency and structural characteristics of the data frames, and compares the parsed acquisition frequency with a preset frequency threshold.

[0058] As a preferred approach, the preset frequency threshold serves as the physical boundary distinguishing between real-time continuous streams and periodic status reporting data. It is set according to the communication specifications of various control devices in the mine, with a typical range of 1Hz to 10Hz. When the acquisition frequency of a data frame is greater than or equal to the preset frequency threshold, the splitting module 110 classifies it as a high-frequency data stream. This type of high-frequency data stream includes continuous microseismic waveform data reported by the microseismic system or continuous time-series streams from the hydrological monitoring system. The splitting module 110 routes it to the streaming commitment generation path. When the acquisition frequency of a data frame is less than the preset frequency threshold and contains structured business fields, the splitting module 110 classifies it as low-frequency, high-sensitivity structured data. Low-frequency, high-sensitivity structured data includes parameters such as precise personnel location coordinates, gas concentration alarm thresholds, and ventilation fan control commands. The splitting module 110 routes it to the multidimensional semantic commitment generation path.

[0059] S120, for the high-frequency data stream entering the streaming commitment generation path, to ensure the integrity of the streaming data, the semantic commitment generation module 120 uses cryptographic hash operations to generate a state digest. Specifically, the semantic commitment generation module 120 extracts high-frequency data buffer blocks within the current time window according to a set time window span. In this embodiment, the time window span is typically set to 100ms to 500ms. To ensure accurate time alignment of data from multi-source heterogeneous devices, each data frame undergoes timestamp calibration via the Network Time Protocol before entering the security gateway node 10.

[0060] Furthermore, combining the unique identifier to which the data stream belongs, the semantic commitment generation module 120 generates a lightweight streaming commitment, the calculation logic of which is as follows:

[0061] ;

[0062] In the formula, For time windows Lightweight streaming commitments; This refers to the Chinese national cryptographic hash algorithm function. This serves as a globally unique identifier for the high-frequency data stream; This is the sequence number of the currently captured physical time window; This is a buffer block of high-frequency data captured within the current time window; This is a data byte stream concatenation operator. By performing an initial hash operation on the buffer block internally and nesting time window parameters externally, this design compresses the data volume while defending against time-series replay attacks on streaming data.

[0063] S130, for low-frequency, highly sensitive structured data entering the multi-dimensional semantic commitment generation path, the semantic commitment generation module 120 extracts the currently effective global policy version, the de-identification dictionary version, and the identifier of the underlying physical device group from which the data originates. In this embodiment, the de-identification dictionary version is used to indicate the update status of the underlying de-identification rule set.

[0064] The semantic commitment generation module 120 uses the current time window sequence number as a deduction variable to anchor parameters such as the de-identified dictionary version into the deduction process of the dynamic policy state. The specific calculation logic is as follows:

[0065] ;

[0066] In the formula, For the current time window The dynamically generated strategy state is derived. This refers to the currently active global policy version. This is a de-identified dictionary version; This identifies the underlying physical device group from which the data originates. Because various system-level parameters with physical dependencies are introduced into the hash calculation, this deductive operation is used to limit desensitization and degradation behavior, ensuring that any illegal modification of dictionary parameters will trigger hash mismatch results in subsequent verification stages.

[0067] S140, to verify the legality of the physical deployment topology of the equipment, the semantic commitment generation module 120 extracts field pairs with business topology dependencies from low-frequency, high-sensitivity structured data. Field identifiers, association constraint identifiers, and corresponding verification metadata used to generate cross-field association consistency proofs are provided as verification metadata along with the mixed encrypted data to the desensitized proxy layer node 40, allowing it to recalculate and verify the cross-field association consistency proof without restoring all high-sensitivity business values. Specifically, the system extracts the binding relationship between the physical media access control address of the underground gas monitoring substation and the roadway number of its respective mining area, and obtains the corresponding association constraint identifier and binding matching result.

[0068] Based on the aforementioned generated dynamic policy state, the system calculates the cross-field association consistency proof, and its calculation logic is as follows:

[0069] ;

[0070] In the formula, For the first A cross-field association consistency proof; This is a message authentication code calculation function based on the Chinese cryptographic hash algorithm, using a dynamic strategy state. As a key parameter; The physical media access control address for the sensing device; Number the roadways in the mining area; For the first Item association constraint identifier; To bind the matching results, a Boolean data representation is used, with the parameter set to 1 when the topology binding is valid. This operation deeply couples physical location information with device identification, preventing the risk of erroneous emergency commands from the ground control center due to malicious tampering with sensor logical coordinates.

[0071] S150, for the de-identification transformation within low-frequency, high-sensitivity structured data, the semantic commitment generation module 120 traverses each plaintext field in the message, extracting the applicable de-identification rule encoding, data type, and visibility level for each field. In this embodiment, the visibility level is divided according to the system's preset 6-level de-identification standard, which includes 6 gradient levels from full plaintext visibility to complete replacement. The system's logic for calculating the single-rule digest is as follows:

[0072] ;

[0073] In the formula, For the first A single rule summary for a plaintext field, superscript Iterate through the index for each field; For the first The specific de-identification rules applicable to each plaintext field are encoded; For the first The data type of each plaintext field; For the first The visibility level of each plaintext field.

[0074] To present the rule topology at the overall level, the semantic commitment generation module 120 extracts a management matrix reflecting the rule inheritance and priority relationships between fields, defined as a rule dependency logic relationship matrix. Before hash calculation, the system pre-unfolds and serializes this two-dimensional matrix into a one-dimensional byte stream. The single rule summary of each plaintext field is then aggregated with this one-dimensional matrix to generate a rule semantic commitment graph. The calculation logic is as follows:

[0075] ;

[0076] In the formula, The summary value of the rule semantic commitment graph; The maximum number of fields in a single structured message is preset. This threshold is set according to the maximum carrying capacity of the message in the business communication protocol, and its typical value is 128 or 256. When the actual number of parsed fields is insufficient, the system will automatically fill in the gaps with 0 values ​​for alignment. This is a byte stream of the rule dependency logic relation matrix after serialization.

[0077] S160, to ensure compatibility with the communication protocols of heterogeneous data systems, the semantic commitment generation module 120 extracts the message structure template identifier and attribute format constraints from low-frequency, high-sensitivity structured data. Combining these parameters, the system calculates the structure syntax commitment:

[0078] ;

[0079] In the formula, Commitment to the generated structural syntax; For message structure template identification; This is an attribute format constraint byte string. In this embodiment, the attribute format constraint byte string specifies the encoding format and legal length range of each plaintext field. The extraction of this commitment value provides a fixed verification benchmark for the degradation placeholder and format compatibility checks in the subsequent proxy reconstruction stage.

[0080] See attached document Figure 4 The following describes the detailed implementation of deterministic out-of-order encapsulation and the generation of anchor points for same-origin evidence. The security gateway node 10 executes this business logic, specifically through the following sub-steps:

[0081] S210, the encapsulation and fusion module 130 inside the security gateway node 10 performs field-level subkey derivation and symmetric encryption operations. Considering that the security gateway node 10 is usually deployed in operating environments such as underground aggregation layer data centers, its edge computing processor has objective limitations in computing power.

[0082] In cryptographic engineering practice, to prevent the leakage of a single key from leading to the decryption of global data, systems typically use key derivation functions to generate exclusive keys for independent data units. Based on this technical principle, the encapsulation and fusion module 130 uses the dynamic policy state obtained from the aforementioned deduction as the core input material, and quickly generates field-level subkeys through the key derivation function of the national cryptographic algorithm. Its calculation logic is as follows:

[0083] ;

[0084] In the formula, For the first Each plaintext field corresponds to a field-level subkey; This is a key derivation function for the national cryptographic standard. In this embodiment, it is specifically implemented using key derivation based on the SM3 algorithm. The dynamic strategy state for the current time window; For the first A system-globally unique identifier for a plaintext field; This is the fixed byte length for the required output key. To accommodate subsequent commercial cryptographic standard encryption algorithms, this fixed byte length is set to 16 bytes.

[0085] After completing the key derivation operation, the encapsulation and fusion module 130 calls the symmetric encryption mode of the commercial cryptographic standard SM4 algorithm to perform encryption operations on the highly sensitive plaintext field, thereby generating the corresponding ciphertext data. Specifically, the system typically uses the SM4 codebook mode or cipher block chaining mode to process fixed-length structured fields. This process relies on the lightweight characteristics of the single symmetric key system, which helps ensure that the single encryption and decryption latency at the edge is less than or equal to 5ms, thus meeting the strict requirements of coal mine safety regulations for real-time data flow.

[0086] S220, to prevent external eavesdroppers from inferring business logic by analyzing the fixed storage offset of data frames, the encapsulation and fusion module 130 performs a deterministic physical permutation and rearrangement operation on the generated ciphertext data array.

[0087] In specific implementation, the encapsulation and fusion module 130 uses the dynamic policy state as the initial vector parameter to drive the counter mode of the commercial cryptographic standard SM4 algorithm to generate a pseudo-random bit stream. Since the counter mode does not depend on the preceding ciphertext state and has parallel output characteristics, the generated bit stream has high determinism and random distribution characteristics. The encapsulation and fusion module 130 uses this pseudo-random bit stream as input parameters to drive a preset shuffling algorithm to perform a physical scrambling operation on the ciphertext data array and the corresponding single-rule digest, generating a scrambled mixed ciphertext array.

[0088] As a preferred approach, the aforementioned shuffling algorithm employs the standard Fisher-Yates permutation algorithm, utilizing a pseudo-random bitstream to be bitwise converted into the random index values ​​required during the algorithm's execution. This out-of-order operation removes the physical order mapping between the ciphertext data array and the original heterogeneous message topology. After generating the out-of-order mixed ciphertext array, the encapsulation and fusion module 130 further calculates a rule consistency proof tag for each out-of-order ciphertext block. The rule consistency proof tag is generated jointly by the corresponding out-of-order ciphertext block, the single rule digest corresponding to that block, and the current position index, and is used by the desensitized proxy layer node 40 to verify the consistency between the ciphertext block and the rule semantics and physical location without recovering all plaintext. The generated rule consistency proof tag is uploaded to the untrusted storage node 20 along with the corresponding out-of-order mixed ciphertext array.

[0089] Since the generation of pseudo-random bit streams depends entirely on dynamically extrapolated state parameters, this mechanism effectively avoids the underlying differences in the implementation of pseudo-random number generators across hardware platforms, thereby ensuring that subsequent agent layer nodes can restore the initial business sequence without deviation under the premise of obtaining the same extrapolated state.

[0090] S230 introduces hash ratchet technology to construct a data anti-tampering chain spanning different physical time windows. This technology utilizes the unidirectional irreversible characteristic of hash functions, using historical states as input parameters for the current operation, thereby forming a strongly coupled dependency relationship in the time dimension.

[0091] Based on this principle, the encapsulation and fusion module 130 performs a cross-time window hash ratchet operation to generate a cross-time window continuity proof. The calculation logic is as follows:

[0092] ;

[0093] In the formula, For the current time window Proof of continuity across time windows generated internally; Proof of the continuity of historical time-spanning windows generated from the previous time window; The overall national cryptographic hash value of the disordered mixed ciphertext array generated within the current time window; This is the current physical time window sequence number.

[0094] To address the computational dead zone caused by the lack of prior credentials during system initialization, when the system is running within the first physical time window, The parameters are replaced by a system-preset static initial seed value. In this embodiment, the static initial seed value is a 256-bit fixed random sequence burned into the security module during system deployment. The above iterative calculation mechanism ensures that the current continuity proof is deeply bound to all state changes within the system's historical running time window. Illegal tampering with past time window data will cause the current continuity proof to fail hash matching during the verification process.

[0095] S240, the encapsulation and fusion module 130 extracts the various cryptographic computation credentials generated in this process, performs hash aggregation, and generates a globally unique same-origin evidence anchor point. The calculation logic is as follows:

[0096] ;

[0097] In the formula, As anchor points for evidence of common origin; Lightweight streaming commitment; For rule semantic commitment graph; For structural grammar commitment; Proof of the continuity across time windows for the current time window; This is the current version of the de-identified dictionary.

[0098] Considering the uncertainty of when data streams arrive at the security gateway node 10 in actual business operations, to avoid algorithmic dead zones caused by inconsistent hash concatenation lengths due to missing data of a certain type, the encapsulation and fusion module 130 executes strict placeholder alignment logic. Specifically, if only low-frequency, high-sensitivity structured data is processed within the current time window and no high-frequency data streams are received, the encapsulation and fusion module 130... The corresponding positions are padded and aligned using a fixed-length sequence of 256-bit zero-value bytes. Similarly, if no low-frequency structured data is captured within the current time window, then the corresponding... and The parameters are also filled with a sequence of zero-value bytes of the same bit width as above, and the corresponding position of the cross-time window continuity proof is also filled with a sequence of zero-value bytes of a preset length.

[0099] In this embodiment, the security gateway node 10 uploads the generated out-of-order mixed ciphertext array to the untrusted storage node 20 for persistent ciphertext storage; at the same time, the security gateway node 10 synchronously writes the generated same-source evidence anchor points into the trusted audit ledger inside the distributed data encryption and backup system.

[0100] As a further optimization, to eliminate potential placeholder ambiguities caused by the absence of multiple commitment objects within the same time window, the encapsulation and fusion module 130 constructs a commitment coverage graph corresponding to the current time window before generating the same-source evidence anchor. The commitment coverage graph is represented by a fixed-length binary bitmap, with each bit corresponding to a lightweight streaming commitment, a rule-based semantic commitment graph, a structured syntax commitment, a cross-time-window continuity proof, and a de-identified dictionary version, in a preset order. The bit is set to 1 when the corresponding object participates in the aggregation operation of the current time window, and to 0 when it does not. The encapsulation and fusion module 130 incorporates the commitment coverage graph, each commitment object, and the de-identified dictionary version into a hash aggregation, enabling the trusted audit ledger to not only verify the consistency of the commitment value itself during subsequent disaster recovery verification or cross-scenario reuse verification, but also to rigorously verify whether the same-source evidence anchor is established within the same commitment coverage range. This mechanism reduces the risk of collisions caused by the use of fixed zero-value padding, resulting in different business source data generating identical value summaries in form, and improves the distinguishability and anti-counterfeiting capability of same-source verification under heterogeneous business diversion paths.

[0101] The aforementioned trusted audit ledger mechanism provides a fixed physical benchmark for cross-system audit verification. During subsequent historical disaster data tracing and disaster recovery operations, the security gateway node 10 can retrieve encrypted data from the untrusted storage node 20 to recalculate the same-source evidence anchor point and accurately compare it with the same-source evidence anchor point stored in the trusted audit ledger, thereby verifying the integrity and legality of the backup data.

[0102] The following describes the detailed implementation of multidimensional query parameter parsing and hard-bound token issuance. The authorization and authentication center node 30 executes this business logic, specifically through the following operation process:

[0103] Given that traditional role-based access control models are insufficient to effectively handle high-risk query operations across network segments or during abnormal time periods, this embodiment introduces a dynamic access control mechanism based on the fusion of roles and attributes. Under this mechanism, the authorization module 310 receives access requests from the business application and parses the permission level of the access request subject and the current network environment context parameters.

[0104] In this embodiment, the entities making the access requests include various operational personnel in the mining area, such as mine leaders, dispatchers, and anti-collision technicians. Their corresponding permission levels are pre-assigned based on a role-based access control model. Meanwhile, network environment context parameters include attribute-based access control dimensions such as the network segment address of the access initiating node, the request timestamp, and the terminal security baseline status.

[0105] To achieve fine-grained data security management, the system utilizes the parsed permission levels and network environment context parameters to trigger the internal dynamic access control engine. In practice, the dynamic access control engine uses the permission identifier of the access request subject and the network environment attribute tags as matching conditions to perform an intersection matching operation on the global de-identification policy rule base, thereby accurately extracting the smallest rule slice specific to this business query.

[0106] As a preferred approach, this minimal rule slicing limits the subset of field anonymization rules that the current requesting entity can only access within the current context, effectively removing rule information for irrelevant business fields from the source. In case of incomplete request parameters or network environment context parameter parsing failures, the system executes security degradation logic, forcibly setting the visibility level of the fields involved in this query to a fully masked anonymized state, thus avoiding algorithmic dead zones caused by missing judgment conditions leading to unauthorized rule distribution.

[0107] To prevent unauthorized token use across physical network environments, the authorization module 310 extracts the scenario identifier corresponding to the current access request and packages it into a cryptographic computation variable. Specifically, the authorization module 310 predefines three core business scenarios for the mining area: a daily scheduling and monitoring scenario corresponding to the production network environment, a disaster recovery audit and verification scenario corresponding to the backup system environment, and a digital twin target range data supply scenario corresponding to the penetration testing environment. The authorization module 310 converts the parsed current business scenario into a fixed-length hexadecimal identifier code, for example, using 0x01 to represent the daily scheduling and monitoring scenario and 0x02 to represent the disaster recovery audit and verification scenario, which serves as the scenario identifier parameter required for subsequent calculations.

[0108] After obtaining the aforementioned constraint parameters, to ensure the tamper-proof and non-repudiation nature of the distribution strategy during transmission, the authorization module 310, based on an asymmetric cryptographic system, uses the private key within the authorization and authentication center node 30 to sign the core constraint parameters, ensuring that the recipient can verify the legitimate source of the token through the corresponding public key. Based on this general technical principle, the authorization module 310 within the authorization and authentication center node 30 synthesizes the final verification credential.

[0109] In practice, the authorization module 310 pre-calculates the hash digest of the minimum rule slice, and then uses its internally held SM2 algorithm private key to perform a digital signature operation on the business purpose identifier, scenario identifier, minimum rule slice digest, and token expiration timestamp to generate a query view token. The specific calculation logic is as follows:

[0110] ;

[0111] ;

[0112] In the formula, Minimal rule slice summary; This is the smallest regular slice byte stream after serialization. For query view tokens; This is an asymmetric digital signature operation function executed using the internal SM2 private key of the authorized authentication center node 30; A unique identifier pre-assigned to the business application; Identify the target view; Identifier for business query purposes; For scene identification parameters; The token expiration timestamp set by the system; A random factor for query sessions generated on the fly.

[0113] As a further optimization method, the authorization module 310 also introduces a query session random factor uniquely bound to this network request during the generation of the query view token. Query session random factor The random number is generated instantly by the authorized authentication center node 30 upon receiving the access request, relying on a hardware true random number generator. Preferably, this query session random factor is... The data length is set to 256 bits (32 bytes) to strictly align with the security strength of the national cryptographic algorithm and the hash block width. This factor is specifically used to distinguish different query instances initiated by the same identity in the same business scenario. The authorization module 310 will identify the target view. Random factor for this query session These elements are integrated into the digital signature body, ensuring that even if the same access subject repeatedly initiates queries against the same view in the same scenario, it must obtain an independent token strictly bound to the current instantaneous session. After successful signature verification, the de-identified proxy layer node 40 only allows subsequent inverse permutation, plaintext decryption, and placeholder reconstruction operations to be performed if the target view identifier, scenario identifier, and query session random factor carried in the query view token are consistent with the current request context. This anti-replay mechanism is used to restrict the malicious lateral reuse of issued legitimate tokens between different target views and to ensure that replay requests outside the historical query time window can be identified during the signature verification or context comparison stage.

[0114] To ensure the integrity of the algorithm logic, in extreme cases where the identity identifier, target view identifier, business purpose identifier, or scenario identifier is parsed as empty due to incomplete request messages, the authorization module 310 uses a preset rejection placeholder during the byte concatenation stage. In this embodiment, the rejection placeholder uses a stream of all-zero bytes of a specific length to avoid logical dead zones caused by hash concatenation anomalies or signature function execution failures due to missing parameters.

[0115] In the above process, the token expiration timestamp is dynamically set according to the real-time requirements of the business scenario. For example, in daily scheduling and monitoring scenarios, it is usually set to 30 minutes after the request occurs. By forcibly binding the time constraint and scenario identifier into the signature body, this mechanism establishes a physical isolation and timeliness control benchmark at the cryptographic level, ensuring that unauthorized operations such as transferring production network tokens to the test environment or using them beyond the authorized time range will be directly rejected in the subsequent signature verification process at the proxy layer. After the signature is generated, the authorization module 310 issues the above query view token to the de-identified proxy layer node 40 to support subsequent reconstruction and de-identification services.

[0116] See attached document Figure 5 The following describes the detailed implementation of pure memory-state adaptive blind verification and structurally compatible reconstruction. The de-identified proxy layer node 40 executes this business logic, specifically through the following operation process:

[0117] S410, to block data leakage paths from physical storage media at the underlying hardware level, establishes an isolated execution region in the volatile cache within the de-identified proxy layer node 40. Based on the general security protection principle of trusted execution environments, physical memory partitioning and hardware-level access control effectively resist unauthorized reading by privileged processes.

[0118] Based on the above principles, the blind verification module 410 inside the de-identification proxy layer node 40 uses system call interception and memory locking mechanisms to strictly limit the encryption and decryption operations of plaintext data to the memory space of the isolated execution area, thereby preventing plaintext data from being accidentally replaced and written to disk files in the form of memory paging. For the memory allocation and isolation control of volatile caches, those skilled in the art can use memory protection keys or secure enclave technology to implement this; the underlying memory addressing and isolation mechanisms are well-known technologies in the field and will not be elaborated upon here.

[0119] In distributed network architectures, due to the inherent delay in packet routing, maintaining strict clock synchronization between the sender and receiver is often difficult. Given that physical environmental factors such as electromagnetic interference and damaged fiber optic links during deep mining operations can easily cause network jitter and high packet loss, the blind verification module 410 executes a tolerant time-window jitter-resistant matching algorithm. When network transmission delay causes the blind verification of the currently received data within the nominal time window to fail, the blind verification module 410 extracts the state parameters of adjacent time windows within the tolerable drift interval for rapid recalculation and hash matching.

[0120] In this embodiment, the aforementioned tolerable drift interval is empirically set based on the extreme value distribution of network latency tests at the mine site. Its typical value range is set to ±2 physical time windows centered on the nominal time window. The system uses the time window sequence number and historical credentials within the tolerable drift interval to re-execute the hash ratchet operation for cross-time window continuity proof.

[0121] To avoid false blocking or missed blocking due to a single indicator, the system performs a multi-dimensional joint judgment based on the time window span, data packet sequence number, and historical hash credentials. Once a calculation record with a successful hash match is found within the aforementioned tolerable drift interval, the system determines that the current ciphertext block originates legitimately. This mechanism constructs fault-tolerant logic adapted to the physical environment of the mine, effectively mitigating the false blocking of normal business data caused by physical link jitter.

[0122] S430, after a successful blind check match, in order to restore the physical topological order of the original message, the structure reconstruction module 420 inside the desensitized proxy layer node 40 performs deterministic inverse permutation restoration and multi-dimensional semantic verification operations. Specifically, the structure reconstruction module 420 uses the successfully verified state parameters as the initial vector to reconstruct the pseudo-random bitstream sequence of the SM4 algorithm counter mode in the commercial cryptographic standard.

[0123] Since this sequence is completely consistent with the bitstream generated during the encapsulation stage, the desensitization proxy layer node 40 uses this pseudo-random bitstream to drive the reverse permutation operation of the preset shuffling algorithm, restoring the disordered mixed ciphertext array to the topological order of the initial heterogeneous message. As a preferred method, this reverse permutation operation is achieved by recording the random index array generated during the encapsulation stage and performing reverse element swaps according to the index sequence completely opposite to that of the encapsulation stage, thus ensuring the deterministic nature of data order restoration. The aforementioned initial heterogeneous messages cover underlying formats in the mining area, such as private protocol messages from the hydrological monitoring system.

[0124] S440, for ciphertext blocks with access permissions, the de-identification proxy layer node 40 performs authorized-state plaintext memory decryption. The system calls the symmetric decryption mode of the commercial cryptographic standard SM4 algorithm in the isolated execution region's memory, and uses the same national cryptographic standard key derivation function as in the encapsulation stage to quickly generate the corresponding field-level subkey, thereby accurately restoring the ciphertext block to the corresponding plaintext data in memory.

[0125] S450, for unauthorized fields lacking access permissions, the structure reconstruction module 420 performs key-deficient state compatible reconstruction based on the 6-level de-identification standard mapped in the minimum rule slice issued by the authorization module 310. This 6-level de-identification standard is graded according to data sensitivity. When a field rule matches a format generalization rule, the de-identification proxy layer node 40 calls a format-preserving transformation mechanism driven by a national cryptographic symmetric algorithm to synthesize masked text, ensuring that the replaced ciphertext data remains consistent with the original plaintext in terms of character set range and data length.

[0126] As a further optimization method, before generating downgraded placeholders for unauthorized fields, the structure reconstruction module 420 first generates a unique placeholder synthesis seed based on the current query view token, the current field identifier, the corresponding single rule digest, and the scenario identifier using one-way hash derivation. The placeholder synthesis seed is calculated by concatenating the QVT, the current field identifier, the corresponding single rule digest, and the scenario identifier in a preset order, and is only valid within the lifecycle of the current query view, serving as a unified driving parameter for subsequent key-deficient state placeholder synthesis algorithms. For multiple unauthorized fields with logical associations within the same business view, the structure reconstruction module 420 uses the same source placeholder synthesis seed to generate placeholder texts that are mutually compatible in terms of encoding format, field length, and logical references. However, for the same physical field in different query sessions or different business scenarios, because their corresponding placeholder synthesis seeds are different, the output placeholder results no longer maintain a consistent form that can be directly cross-compared. By introducing this dynamic derivation mechanism, the system not only maintains the compatibility and stability of front-end business protocol parsing, but also helps to reduce the risk of inferring the statistical distribution characteristics of real high-sensitivity data by frequently initiating queries of different dimensions and comparing fixed patterns of placeholders.

[0127] When a field rule matches a fake data replacement rule, the de-identification proxy layer node 40 uses a preset data synthesis function to generate test waveform placeholder data that matches the target field's data type and structural constraints. To maintain the parsability of the test data in the business system, the system can adjust the test waveform placeholder data using pre-extracted statistical feature templates to meet the data length, sampling organization method, and basic frequency domain constraints corresponding to the current view template. This placeholder synthesis mechanism for different rules maintains the original message's syntax structure, ensuring that the business system does not trigger format validation anomalies when processing degraded data.

[0128] S460, subsequently, the structure reconstruction module 420 performs a closed-loop operation of heterogeneous view assembly and output. The structure reconstruction module 420 sequentially concatenates the decrypted plaintext data with the synthesized placeholders to generate a business view object compatible with the target simulation environment. In this embodiment, the target simulation environment includes an OpenStack digital twin testbed, and the generated business view object uses JSON or XML format.

[0129] As a further optimization, when the business view object finally output by the structure reconstruction module 420 is assembled from multiple time windows or multiple independent mixed encrypted data packets, the system further calculates and generates a multi-packet assembly consistency commitment before the view assembly is completed. Specifically, the structure reconstruction module 420 extracts the identifiers of each underlying time window participating in this view assembly, the source evidence anchor points corresponding to each plaintext / ciphertext block, and the actual logical assembly sequence of these blocks in the current business view object. After performing serialization aggregation according to a preset format, it calculates the multi-packet assembly consistency commitment. The consistency commitment certificate is used to characterize that the currently output business view object is not freely and disorderly spliced ​​together from arbitrary encrypted fragments in memory, but is composed of a group of data objects that satisfy global consistency constraints in the time dimension, source dimension, and business topology assembly order. The de-identified proxy layer node 40 records this consistency commitment as an important component of the final view output commitment to the trusted audit ledger, thus providing a basis for verifying whether complex combined views originate from the same legitimate business chain in subsequent target range twin reproduction, disaster recovery integrity restoration, or post-event security audit scenarios.

[0130] To establish a security audit baseline for the final view output, the system calculates the view output commitment based on a hash algorithm. The calculation logic is as follows: VOC is determined by the business view object. It is calculated together from the query view token (QVT), the multi-package assembly consistency commitment (if it exists), and the dynamic rule hit execution digest (if it exists).

[0131] As a further optimization, to prevent the de-identified proxy layer node 40 from being controlled by malicious processes after obtaining the minimum rule slice and arbitrarily expanding the scope of rule application, the structure reconstruction module 420, after completing the assembly of the business view object, further extracts the set of rule summaries actually hit within the lifecycle of this view reconstruction and generates a dynamic rule hit execution summary. The rule hit execution summary is obtained by aggregating and hashing each single rule summary that has undergone real driver field transformation and downgrading during the current in-memory view reconstruction process according to the physical field reorganization order, and is arranged together with the minimum rule slice summary issued by the authorization module 310, jointly participating in the aforementioned view output commitment. The generation operation. Through this two-way verification mechanism, the distributed trusted audit ledger can not only verify whether the current view object is legitimately derived from the corresponding query view token, but also accurately verify afterward whether the dynamic view reconstruction process strictly converges within the minimum rule slice boundary pre-authorized by the system. This prevents the hidden overreach of authority behavior where the control policy is correctly issued, but rule degradation and expansion occur during the actual execution on the end side.

[0132] To prevent algorithm dead zones caused by parsing errors or empty data packets, if the current view assembly fails, The content is empty; the structure reconstruction module 420 uses a preset fixed-length all-zero placeholder instead. Participate in the splicing calculation. After completing the hash calculation, the structure reconstruction module 420 synchronously records the view output commitment to the trusted audit ledger inside the distributed data encryption and backup system, thereby completing the view assembly and full-process security audit closed loop to support the range penetration test.

[0133] The present invention also provides a computer device, including: a processor and a memory, the memory storing a computer program executable by the processor, the computer program performing the method described above when executed by the processor.

[0134] The present invention also provides a storage medium storing a computer program, which is executed by a processor to perform the method described above.

[0135] The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0136] To further illustrate the technical effects of the present invention, the following detailed description is provided in conjunction with specific embodiments and experimental data from a complex disaster monitoring scenario in deep coal mining.

[0137] This embodiment is based on the actual monitoring network architecture of a high-gas deep mine. The bottom ring network connects to two types of typical monitoring terminals: one is a microseismic monitoring sensor with a sampling rate of 500Hz, and the other is an underground gas monitoring substation with a reporting cycle of 1 time / minute.

[0138] On the edge processing side, the security gateway node 10 deployed in the underground substation triggers the diversion module 110 after receiving the reported data. Microseismic waveform data, due to its frequency exceeding a preset frequency threshold, is routed to the streaming commitment generation path. The security gateway node 10 only extracts a lightweight streaming commitment to ensure temporal integrity. Low-frequency, high-sensitivity structured data reported by the gas substation (including plaintext fields such as physical media access control addresses, roadway absolute coordinates, and real-time gas concentration) enters the multidimensional semantic commitment generation path. The security gateway node 10 topologically binds the physical media access control addresses of the gas substation with the roadway numbers and extracts cross-field association consistency proofs. Combined with the inferred dynamic strategy state, it drives a pseudo-random sequence to shuffle the physical order of each high-sensitivity ciphertext field, ultimately aggregating and generating a common-source evidence anchor point, which, along with the disordered mixed ciphertext, is uploaded to the untrusted storage node 20 of the ground dispatch center. This common-source evidence anchor point is synchronously written into the trusted audit ledger within the disaster recovery system as a subsequent verification benchmark.

[0139] On the dynamic authorization and reconstruction side, there are two types of business access requests: daily security monitoring queries from the dispatch and command center, and data supply requests for penetration testing from the digital twin test range.

[0140] When a dispatcher initiates a query, the authorization and authentication center node 30 extracts the daily dispatch monitoring scenario identifier and the dispatcher's identity identifier, and issues the minimum rule slice and query view token. The de-identification proxy layer node 40 performs blind verification within the memory-isolated execution area. For the absolute coordinate field of the roadway in the gas message, the de-identification proxy layer node 40, based on the generalized de-identification instructions in the minimum rule slice, converts the precise coordinates into a downgraded placeholder for "A mining area return airway" without storing plaintext, and then outputs a business view object compatible with the dispatch dashboard's JSON parsing engine.

[0141] During target range penetration testing, the authorized authentication center node 30 issues an independent token carrying a disaster recovery audit verification scenario identifier. After parsing and verification by the de-identified proxy layer node 40, a fake data replacement rule is triggered for the highly sensitive micro-seismic waveform. Using the mean amplitude and frequency variance of the original data as constraint parameters, the data synthesis function is driven to generate test waveform placeholder data with the same statistical characteristics in memory. This process provides compliant data-driven support for automated attack chain deduction while concealing the real warning threshold and sensitive coordinates.

[0142] To verify the availability and performance of the data security management system and method of the present invention, a test environment including edge computing nodes and high-computing-power proxy servers was built, and desensitized stream data collected from a mining area was introduced for comparative testing.

[0143] like Figure 6As shown, under test conditions with continuously increasing concurrent requests up to 5000 QPS, the traditional DDM solution, relying on the database engine's string regular expression replacement and temporary table disk persistence mechanism, achieves a 95th percentile (P95) latency exceeding 120ms. This invention, based on a two-dimensional traffic splitting strategy and a purely memory-based deterministic inverse permutation operation, de-identifies the proxy layer node 40 by performing memory decryption only on authorized fields, and quickly concatenates unauthorized fields using placeholders to form seeds. Experimental data shows that the P95 latency of this invention stably converges to within 35ms under extremely high concurrency conditions, meeting the real-time data retrieval requirements of coal mine safety regulations.

[0144] like Figure 7 As shown, traditional static desensitization schemes replace sequence data with fixed masks, which can easily lead to format or business logic anomalies during simulation model parsing. Experimental results show that the downgraded placeholder data synthesized according to the rules of this invention has an energy distribution profile in the microseismic main frequency range of 10Hz-50Hz that basically fits the real original waveform. The downstream vulnerability assessment component did not experience parsing failure or format errors when receiving the synthesized data, verifying that this invention maintains the consistent availability of complex simulation services while ensuring the security of the underlying data.

[0145] like Figure 8 As shown, in the case of sudden packet loss caused by downhole electromagnetic interference, traditional mechanisms fail to achieve a cascaded verification failure due to hash chain breakage when the packet loss rate reaches 3%, resulting in a drop in valid throughput to below 60%. Test data shows that this invention, by setting a tolerable drift range through the blind verification module 410, maintains an effective verification pass rate of over 98.5% and a view reconstruction success rate even under channel conditions with a packet loss rate of 8%, demonstrating a high degree of adaptability to harsh communication environments in industrial settings.

[0146] 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 security management method based on a national secret algorithm, characterized in that, include: The security gateway node receives plaintext data and, based on the acquisition frequency of the plaintext data, splits the plaintext data into high-frequency data streams and low-frequency structured data. For the high-frequency data stream, a streaming commitment is generated; for the low-frequency structured data, a dynamic strategy state is deduced and a rule semantic commitment graph is generated. The security gateway node uses a dynamic policy state to derive a key to encrypt the low-frequency structured data to obtain ciphertext data, and uses the dynamic policy state to generate a deterministic permutation sequence to shuffle the ciphertext data. The streaming commitment and the rule semantic commitment graph are combined to generate a common source evidence anchor point, and the common source evidence anchor point and the shuffled ciphertext data are uploaded to the storage node. The authentication node receives a query request and digitally signs the request parameters of the query request to generate a query view token. The proxy node obtains the scrambled ciphertext data and the same-source evidence anchor point, and performs data verification based on the same-source evidence anchor point; After verification, the order of the ciphertext data is restored using the deterministic permutation sequence. Based on the query view token, the restored ciphertext data is decrypted or downgraded for placeholder purposes. The operation results are then assembled into the target business view output.

2. The data security management method based on the SM algorithm according to claim 1, characterized in that, The step of splitting plaintext data into high-frequency data streams and low-frequency structured data based on the acquisition frequency of the plaintext data includes: The plaintext data collection frequency is analyzed, and the collection frequency is compared with a preset frequency threshold. When the acquisition frequency is greater than or equal to a preset frequency threshold, the corresponding plaintext data is determined to be the high-frequency data stream; when the acquisition frequency is less than the preset frequency threshold and contains structured fields, the corresponding plaintext data is determined to be the low-frequency structured data.

3. The data security management method based on the SM algorithm according to claim 1, characterized in that, The steps for generating streaming commitments for the high-frequency data stream include: The high-frequency data stream is intercepted according to the set physical time window to obtain a high-frequency data buffer block; The streaming commitment is generated by concatenating the identifier of the high-frequency data stream, the sequence number of the physical time window, and the hash value of the high-frequency data buffer block using a hash algorithm.

4. The data security management method based on the SM algorithm according to claim 1, characterized in that, The steps of inferring dynamic policy states from the low-frequency structured data and calculating and generating rule semantic commitment graphs include: The dynamic policy state is generated by concatenating and calculating the preset policy version, the preset de-identification dictionary version, and the current time window sequence number using a hash algorithm. Extract the visibility level of each field in the low-frequency structured data and generate a single-rule summary; Obtain the rule dependency logic relationship matrix that reflects the rule priority, and perform aggregate hash calculation on the rule dependency logic relationship matrix and the single rule digest to generate the rule semantic commitment graph.

5. The data security management method based on the SM algorithm according to claim 1, characterized in that, The step of generating a deterministic permutation sequence using the dynamic strategy state to shuffle the ciphertext data includes: Using the dynamic strategy state as the initial vector parameter input into the counter mode of the symmetric encryption algorithm, a pseudo-random bit stream is generated; The pseudo-random bitstream is used as a random seed input in the position permutation algorithm to generate the deterministic permutation sequence. The deterministic permutation sequence is then used to perform physical position swapping on the ciphertext data, thereby breaking the original order mapping association of the ciphertext data.

6. The data security management method based on the SM algorithm according to claim 1, characterized in that, The step of combining the streaming commitment with the rule-based semantic commitment graph to generate anchor points for shared evidence includes: The current time window continuity proof is generated by performing a hash iteration operation on the historical cross-time window continuity proof, the hash value of the ciphertext data, and the current time window sequence number; The streaming commitment, the rule semantic commitment graph, and the continuity proof of the current time window are combined and calculated to generate the same source evidence anchor point.

7. The data security management method based on the SM algorithm according to claim 1, characterized in that, The step of digitally signing the request parameters of the query request to generate a query view token includes: Obtain the request subject identifier, target view identifier, and current scene identifier from the query request; A random number generator is used to generate a query session random number that is uniquely bound to the query request. An asymmetric encryption algorithm is used to digitally sign the request subject identifier, the target view identifier, the current scene identifier, and the query session random number to generate the query view token.

8. The data security management method based on the SM algorithm according to claim 1, characterized in that, The proxy node obtains the scrambled ciphertext data and the same-origin evidence anchor point, and performs data verification based on the same-origin evidence anchor point, including: Extract the nominal time window parameter from the anchor points of the same source evidence and perform hash matching verification; When the nominal time window parameter verification fails, the adjacent time window state parameters within the preset drift interval are extracted as compensation parameters, and the hash matching calculation is re-executed using the compensation parameters to complete the anti-jitter compensation verification.

9. The data security management method based on the SM algorithm according to claim 1, characterized in that, The steps of performing decryption or downgrade placeholder operations on the restored ciphertext data based on the query view token include: Parse the query view token to obtain access permission credentials; determine whether the encrypted data after restoration has access permission credentials; For encrypted data with the aforementioned access credentials, a decryption operation is performed using a key to obtain the decrypted plaintext; For encrypted data without the access permission credential, a placeholder character seed is generated by hash derivation based on the query view token, and a downgraded placeholder text is generated using the placeholder character seed.

10. The data security management method based on the SM algorithm according to claim 9, characterized in that, The steps for assembling the operation results into the target business view output include: The decrypted plaintext is concatenated and assembled with the downgraded placeholder text to obtain the target business view; Extract the time window identifiers involved in the assembly and the rule summaries that were hit, perform aggregate hash calculations to generate view output commitments, output the target business view, and record the view output commitments in the trusted audit log.