Multi-node heterogeneous server networking data secure intercommunication method and system

Abstract

The invention discloses a multi-node heterogeneous server networking data secure intercommunication method and system, and relates to the technical field of data processing. Comprising the steps that a source edge node based on a heterogeneous server intercepts a real-time data request through an embedded edge agent, and request identity verification based on multi-dimensional trust evaluation is carried out; if the identity verification is passed, contextual information of the real-time data request is collected based on the edge agent, the contextual information is combined into a to-be-verified information set, data texture verification is carried out on the to-be-verified information set, and the real-time data request is sent to the data relay according to a data texture verification result; the signature validity and the data integrity of the real-time data request are verified through data transfer, and the real-time data request is forwarded to a target edge node through routing decision after secondary re-encryption; and the target edge node receives the data forwarded by the data relay, performs request response, generates a digital signature and returns the digital signature to the source edge node through the data relay. According to the invention, the security of networking data intercommunication of the multi-node heterogeneous server is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method and system for secure data interoperability in a multi-node heterogeneous server network. Background Technology

[0002] With the rapid development of edge computing and heterogeneous server networking, the demand for data interoperability between multiple nodes is increasing. Existing technologies typically use single identity authentication or transport layer encryption mechanisms to ensure basic security.

[0003] However, when faced with hardware differences from heterogeneous servers, dynamic network topologies, and concurrent access to multi-source data, existing technologies rely on a single dimension for trust assessment, focusing only on identity credential verification. This makes it difficult to effectively prevent security risks such as forged requests, data tampering, and man-in-the-middle attacks. Furthermore, the lack of collaborative trust mechanisms between nodes leads to security gaps during data transfer. Summary of the Invention

[0004] This invention provides a method and system for secure data communication in a multi-node heterogeneous server network, aiming to solve the technical problem of insufficient security in data communication in existing multi-node heterogeneous server networks.

[0005] In view of the above problems, the present invention provides a method and system for secure data interoperability in a multi-node heterogeneous server network.

[0006] In a first aspect, the present invention provides a method for secure data interoperability in a multi-node heterogeneous server network, comprising:

[0007] The source edge node based on heterogeneous servers intercepts real-time data requests through an embedded edge proxy and performs request identity verification based on multi-dimensional trust assessment on the real-time data requests. If the identity verification passes, the context information of the real-time data request is collected based on the edge agent, merged into a set of information to be verified, the set of information to be verified is subjected to data texture verification, and the real-time data request is sent to the data relay according to the data texture verification result. The data relay verifies the signature validity and data integrity of the real-time data request, and performs a second re-encryption on the real-time data request. After the second re-encryption, the request is forwarded to the target edge node via routing decision. The target edge node receives the data forwarded by the data relay, responds to the request, generates a digital signature, and returns it to the source edge node through the data relay.

[0008] Secondly, this invention provides a secure data interconnection system for multi-node heterogeneous server networking, comprising: The identity verification module is used to intercept real-time data requests from the source edge nodes of heterogeneous servers through an embedded edge proxy, and to perform request identity verification on the real-time data requests based on multi-dimensional trust assessment. The texture verification module is used to collect the context information of the real-time data request based on the edge agent if the identity verification is successful, merge it into a set of information to be verified, perform data texture verification on the set of information to be verified, and send the real-time data request to the data relay according to the data texture verification result. The relay encryption forwarding module is used to verify the signature validity and data integrity of the real-time data request during data relay, and to perform secondary re-encryption on the real-time data request. After secondary re-encryption, the request is forwarded to the target edge node through routing decision. The response signature module is used by the target edge node to receive the data forwarded by the data relay, make a request response, generate a digital signature, and return it to the source edge node through the data relay.

[0009] One or more technical solutions provided in this invention have at least the following technical effects or advantages: This invention discloses a method and system for secure data interoperability in a multi-node heterogeneous server network. Through multi-dimensional trust assessment and data texture verification of the source edge nodes, fine-grained identity authentication and contextual risk identification are achieved from the source of the request, effectively intercepting illegal requests and abnormal data. Furthermore, signature verification, integrity verification, and secondary re-encryption are simultaneously performed during data transfer, ensuring secure isolation and tamper-proof capabilities of data during cross-node transmission. Finally, combined with the digital signature and bidirectional return mechanism of the target edge node's response, end-to-end traceability and non-repudiation are achieved. This invention effectively improves data security and inter-node collaborative trust capabilities. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating the method for secure data interoperability in a multi-node heterogeneous server network provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of a multi-node heterogeneous server networking data security interoperability system provided in an embodiment of the present invention; The components represented by each number in the attached diagram are explained below: Identity verification module 11, texture verification module 12, relay encryption and forwarding module 13, response signature module 14. Detailed Implementation

[0011] This invention provides a method and system for secure data interoperability in multi-node heterogeneous server networking, which addresses the technical problem of insufficient security in data interoperability in existing multi-node heterogeneous server networking.

[0012] Example 1, as Figure 1 As shown, this invention provides a method for secure data interoperability in a multi-node heterogeneous server network, the method comprising: S100: The source edge node based on the heterogeneous server intercepts real-time data requests through the embedded edge proxy and performs request identity verification on the real-time data requests based on multi-dimensional trust assessment.

[0013] In this embodiment of the invention, the source edge node based on heterogeneous servers intercepts real-time data requests through an embedded edge proxy and performs request identity verification based on multi-dimensional trust assessment. In a multi-node heterogeneous server network environment, edge nodes exhibit differences in hardware architecture, operating systems, and communication protocols, resulting in varying node security capabilities. Traditional request identity verification uses only a single certificate verification method, which cannot cover multiple security risks such as the device's own security status and abnormal access behavior. This makes it highly susceptible to unauthorized access issues such as certificate theft, access by vulnerable devices, and impersonation of legitimate requests. Furthermore, different business scenarios have varying data security levels, and fixed verification rules cannot adapt to differentiated security protection needs. Therefore, it is necessary to complete request identity verification through multi-dimensional trust assessment, comprehensively determining the legitimacy of the request from three dimensions: certificate credentials, device status, and access behavior, providing a source of trusted assurance for secure data interoperability in heterogeneous server networks.

[0014] Step S100 in the method provided in this embodiment of the invention includes: Based on the certificate issuance level, certificate key length, and revocation status check results, the real-time data request is scored for authentication strength. Based on the device firmware version, the operating status of security protection components, and system integrity metrics, a device health score is performed on the real-time data request. Based on the degree of deviation between the current request frequency, requested data volume, and accessed resource type and the historical behavior baseline, the real-time data request is scored for behavioral baseline deviation. The normalized weighted identity authentication strength score, device health score, and behavioral baseline deviation score are used to obtain a comprehensive trust score. When the comprehensive trust score meets the preset trust threshold, the identity verification is deemed to have passed. The normalized weighted weights are selected from preset weight combinations based on the data security level of the target scenario.

[0015] First, based on the certificate issuance level, certificate key length, and revocation status check results, the real-time data request is scored for identity authentication strength. The identity authentication strength score f_identity is a score that quantifies the security and reliability of the requested identity credentials based on the issuance level, key length, and revocation status of the device certificate carried in the request. The score ranges from 0 to 1, with a higher score indicating a more secure and trustworthy identity credential.

[0016] In this context, the certificate issuance level refers to the depth of the trust chain within the Public Key Infrastructure (PKI) system for the digital certificate used by the requester. It is typically represented by the number of intermediate CAs from the root Certificate Authority (CA) to the end certificate. A deeper issuance level indicates a more complex certificate issuance path, and this scheme considers it as one of the factors in evaluating certificate authority and chain integrity. The certificate key length refers to the number of bits in the encryption algorithm key corresponding to the public key in the certificate, such as RSA 2048 bits, RSA 4096 bits, or ECC 256 bits. A longer key length increases the difficulty of cracking and strengthens the anti-counterfeiting capabilities of the identity credential. The revocation status check result verifies whether the certificate has been added to the Certificate Revocation List (CRL) by the issuing authority or can be queried as valid through the Online Certificate Status Protocol (OCSP). If the certificate has been revoked, it is directly determined to be untrustworthy.

[0017] Specifically, after the edge proxy embedded in the source edge node intercepts real-time data requests, it extracts the device certificate from the request and performs three checks and assigns scores accordingly: Checking the certificate issuance level by tracing the certificate chain back to the root CA to ensure the chain is complete and unbroken; checking the certificate key length to confirm it meets the security requirements of the corresponding encryption algorithm; and checking the certificate revocation status by querying the Certificate Revocation List (CRL) or the Online Certificate Status Protocol (OCSP) to confirm the certificate has not been revoked. If all three checks pass, `f_identity` is assigned a value of 1.0; if two checks pass, it is assigned 0.7; if one check passes, it is assigned 0.3; and if all checks fail, it is assigned 0.0.

[0018] For example, taking the logistics asset management server in the digital logistics scenario of government agencies as the source edge node, the real-time data request initiated by this node carries the SM2 device certificate: the certificate chain is root CA-industrial intermediate CA-node certificate, with complete hierarchy; the key length is 256 bits, which conforms to the SM2 algorithm security standard; the certificate has not been revoked by querying through the OCSP protocol, and all three verifications pass, so f_identity=1.0.

[0019] Secondly, based on the device firmware version, the operating status of security protection components, and system integrity metrics, a device health score is performed on the real-time data request. The device health score f_device is a quantitative assessment of whether the node device has vulnerabilities or tampering risks based on the edge node's own hardware and software security status. The score ranges from 0 to 1, with a higher score indicating a better security status for the node device.

[0020] Specifically, the edge proxy collects three security metrics from the source edge nodes to complete the device health score: It checks the device firmware version to confirm it is the latest stable version officially released and free of known high-risk security vulnerabilities; it checks the operational status of security protection components to confirm that the firewall, intrusion detection, and security protection components are started and running normally; and it verifies the system integrity metric by comparing the current system core file metric values ​​with pre-stored baseline values ​​to confirm that the system has not been illegally tampered with. If all three metrics meet the standard, f_device is assigned a value of 1.0; if two metrics meet the standard, f_device is assigned a value of 0.6; if one metric meets the standard, f_device is assigned a value of 0.3; and if all metrics fail to meet the standard, f_device is assigned a value of 0.0.

[0021] For example, the firmware of the aforementioned logistics asset management server is the latest stable version from the manufacturer, with no high-risk vulnerabilities; the security protection components such as firewalls and intrusion detection systems are all operating normally; the system integrity metric is completely consistent with the pre-stored benchmark value, and all three indicators meet the standards, therefore f_device=1.0.

[0022] Next, based on the deviation of the current request frequency, requested data volume, and accessed resource type from the historical behavior baseline, a behavior baseline deviation score is calculated for the real-time data request. The behavior baseline deviation score f_behavior is a score that quantifies the degree of abnormal access behavior by comparing the behavioral characteristics of the current real-time request with the node's historical normal behavior baseline. The score ranges from 0 to 1, with a higher score for smaller deviations.

[0023] Specifically, the edge proxy extracts three behavioral characteristics of current real-time data requests and compares them with the historical normal behavior baseline stored locally for scoring: Request frequency: the number of requests initiated per unit time; Request data volume: the size of data transmitted in a single request; Accessed resource type: the type of business resource requested. If the deviation of any of the three characteristics from the baseline is ≤10%, f_behavior is assigned a value of 1.0; if the deviation of any feature is between 10% and 30%, a value of 0.6 is assigned; if the deviation of any feature is ≥30%, a value of 0.2 is assigned.

[0024] For example, the historical normal behavior baseline of the logistics asset management server is: request frequency 4 times / minute, single request data size 8KB, only accessing logistics asset ledger data; the current real-time request frequency is 3.8 times / minute, deviation 5%, single request data size 8.6KB, deviation 7.5%, accessed resource type is consistent with the baseline, and the deviation of the three characteristics is ≤10%, therefore f_behavior=1.0.

[0025] Furthermore, a comprehensive trust score is obtained by combining the normalized weighted identity authentication strength score, device health score, and behavioral baseline deviation score. When the comprehensive trust score meets a preset trust threshold, the identity verification is deemed successful. The normalized weighting is determined by selecting a preset weight combination based on the data security level of the target scenario. Normalized weighting means weighting scores of different dimensions and units according to preset weights, with the sum of all dimension weights being 1. The weights are preset based on the data security level of the target scenario. The comprehensive trust score TS is the final result of the weighted calculation of the identity authentication strength, device health, and behavioral baseline deviation scores, ranging from 0 to 1, and is used to directly determine whether the identity verification has passed.

[0026] Specifically, based on the data security level of the target scenario, a corresponding preset weight combination is selected; the comprehensive trust score is calculated using the formula: TS=w1×f_identity+w2×f_device+w3×f_behavior, where w1 is the identity authentication strength weight, w2 is the device health weight, w3 is the behavior baseline deviation weight, and w1+w2+w3=1; the comprehensive trust score TS is compared with the preset trust threshold. If TS is greater than or equal to the trust threshold, the identity verification passes; otherwise, the verification fails, and the request is rejected.

[0027] For example, the digital logistics scenario of this organization is of a high security level, with preset weights w1=0.4, w2=0.3, w3=0.3 and a preset trust threshold of 0.8; substituting into the scoring calculation: TS=0.4×1.0+0.3×1.0+0.3×1.0=1.0; 1.0≥0.8, therefore the identity verification of the real-time data request is successful.

[0028] In this embodiment of the invention, real-time data request interception is achieved through an edge proxy. A three-dimensional scoring model is adopted, which combines identity authentication strength, device health, and behavioral baseline deviation with a normalized weighted average to calculate a comprehensive trust score to complete identity verification, thus ensuring the legality and compliance of network requests from the source. It abandons the limitations of traditional single certificate authentication and simultaneously covers triple verification of credential validity, device security status, and access behavior compliance, which can accurately identify risky requests such as credential misuse, abnormal device access, and unauthorized data access. The normalized weighted average can be flexibly configured according to the security level of different management scenarios to adapt to differentiated security needs, providing a reliable trust foundation for subsequent data texture verification and secure data transmission.

[0029] S200: If the identity verification passes, the context information of the real-time data request is collected based on the edge agent, merged into a set of information to be verified, the set of information to be verified is subjected to data texture verification, and the real-time data request is sent to the data relay according to the data texture verification result.

[0030] In this embodiment of the invention, if identity verification passes, the context information of the real-time data request is collected based on the edge proxy, merged into a set of information to be verified, and data texture verification is performed on the set of information to be verified. Based on the data texture verification result, the real-time data request is sent to the data relay. In a multi-node heterogeneous server network environment, identity verification alone cannot fully guarantee the security of data requests; risks such as request content tampering, abnormal context, and disguised legitimate requests still exist. Traditional data verification methods rely solely on integrity verification and cannot identify hidden abnormal requests from the perspectives of temporal features, data texture, and behavioral patterns. Furthermore, data requests and context information contain various data types, including numerical and categorical types, and the fluctuation patterns of different information vary significantly, making direct feature comparison difficult. Fixed-scale analysis windows cannot guarantee verification accuracy. In addition, data requests often contain sensitive information, and direct plaintext transmission poses a risk of leakage. Therefore, after identity verification passes, data texture verification needs to be further performed, and the verified requests need to be de-identified and encrypted before being forwarded to the data relay.

[0031] Step S200 in the method provided in this embodiment of the invention includes: If the identity verification passes, the context information of the real-time data request is collected based on the edge proxy, merged into a set of information to be verified, and data texture verification is performed on the set of information to be verified, including: Obtain the context information of the real-time data request, and merge the real-time data request and the context information into a set of information to be verified, wherein each information item includes an information category label, an information value, and a timestamp; Based on local historical records, a priori information set marked as normal state is extracted, wherein the priori information set includes prior data requests and prior context information; By using preset encoding rules, the prior information set and the information set to be verified are uniformly encoded into an encoded data sequence, and the encoded data sequence is sorted according to time series labels and organized according to information categories to generate a two-dimensional encoded data matrix.

[0032] The texture feature difference analysis of the two-dimensional encoded data matrix is ​​performed using a multi-scale sliding window method, and the corresponding data texture verification results are output.

[0033] First, the context information of the real-time data request is obtained, and the real-time data request and the context information are merged into a set of information to be verified. Each information item includes an information category label, an information value, and a timestamp. Context information refers to auxiliary information associated with the real-time data request, including the request source, access module, data length, and time information, used to enrich the basis for data verification. The set of information to be verified is the information set formed by merging the real-time data request body and the corresponding context information. Each information item includes an information category label, an information value, and a timestamp, used for subsequent data texture verification.

[0034] Specifically, after identity verification is successful, the edge agent collects the entire content of the real-time data request and extracts the context information corresponding to the request. The real-time data request and context information are merged in a unified format, and each piece of information is labeled with an information category tag, information value and timestamp to form a set of information to be verified.

[0035] For example, consider an asset ledger query request initiated by the asset management server: Real-time data request: Information category tag = request content, information value = asset ledger query, timestamp = 20260324100000; Context information: Information category tag = terminal identifier, information value = HQ-ZC-001, timestamp = 20260324100000; Information category tag = access module, information value = asset management, timestamp = 20260324100000; Information category tag = request data length, information value = 8KB, timestamp = 20260324100000; The above information is merged into a set of information to be verified.

[0036] Secondly, based on local historical records, a priori information set marked as normal is extracted. This priori information set includes prior data requests and prior context information. The priori information set refers to the collection of historical data requests and historical context information marked as normal from local historical records, serving as the benchmark data for data texture verification. The edge agent reads locally stored historical access records, filters out verified prior data requests and prior context information marked as normal, forming a priori information set used for texture feature comparison analysis with the information set to be verified.

[0037] For example, extract all legitimate and normal asset ledger query requests and corresponding context information from the local history of the asset management server within the past 30 days to form a priori information set containing normal time sequence, normal terminal, normal access module, and normal data length.

[0038] Furthermore, by using preset encoding rules, the prior information set and the information set to be verified are uniformly encoded into an encoded data sequence, and the encoded data sequence is sorted according to time series labels and organized according to information categories to generate a two-dimensional encoded data matrix.

[0039] Specifically, by using preset encoding rules, the prior information set and the information set to be verified are uniformly encoded into an encoded data sequence. This encoded data sequence is then sorted according to time-series labels and organized according to information categories to generate a two-dimensional encoded data matrix, including: The numerical information values ​​in the prior information set and the information set to be verified are normalized. Based on preset encoding rules, categorical information values ​​are numerically encoded and converted into numerical vector representations; The normalization result and the numerical vector representation are time-aligned according to the timestamp to form the two-dimensional encoded data matrix. The horizontal axis of the encoded data matrix is ​​the time dimension, the vertical axis is the information category dimension, and the value of each matrix element is the encoded value of the corresponding information category at the corresponding time point.

[0040] First, the numerical information values ​​in the prior information set and the information set to be verified are normalized. Numerical information values ​​refer to the values ​​of information items expressed in numerical form, such as requested data length, request frequency, time interval, etc., which have differences in dimensions and value ranges. Normalization means linearly mapping numerical information values ​​with different dimensions and value ranges to [0,1], eliminating the influence of differences in dimensions and magnitudes on subsequent texture analysis.

[0041] Specifically, the edge agent traverses the prior information set and the information set to be verified, identifying numerical information items; it then uses an extreme value normalization method, processed according to the following formula: Where: x is the original numerical information value, This represents the historical minimum value for this category of information. This represents the historical maximum value for this category of information. This is the result after normalization.

[0042] For example, if the requested data length in the set of information to be verified is 8KB, and the historical minimum value for this category is 4KB and the maximum value is 40KB, then: That is, the normalized value is 0.11.

[0043] Secondly, based on preset encoding rules, categorical information values ​​are numerically encoded and converted into numerical vector representations. Categorical information values ​​refer to non-numerical information items represented in the form of text, symbols, or identifiers, such as terminal identifiers, access modules, and request types, which cannot be directly used in numerical calculations. Numerical encoding: According to preset encoding rules, categorical information values ​​are mapped one-to-one to fixed-length numerical values ​​or numerical vectors, making them usable for matrix construction and texture calculations.

[0044] Specifically, the edge agent matches categorical information items in the prior information set and the information set to be verified with a locally preset encoding dictionary; it uniquely maps each categorical information value to a decimal number or a binary number vector, thereby realizing the numerical representation of non-numerical information.

[0045] For example, the preset encoding rules are as follows: Terminal identifier HQ-ZC-001: encoded as 001; Access module asset management: encoded as 101; Request content asset ledger query: encoded as 1001. The above categorical information values ​​are uniformly converted into numerical values ​​that can participate in matrix operations.

[0046] Finally, the normalized result and the numerical vector representation are time-series aligned according to the timestamps to form the two-dimensional encoded data matrix. The horizontal axis of the encoded data matrix represents the time sequence dimension, and the vertical axis represents the information category dimension. Each matrix element represents the encoded value of the corresponding information category at the corresponding time point. The two-dimensional encoded data matrix is ​​a matrix used to structurally represent time-series, multi-category information. The horizontal axis represents the time sequence dimension, and the vertical axis represents the information category dimension. Each element in the matrix represents the encoded result of the corresponding time point and information category. Time sequence alignment means arranging the encoded values ​​of all information categories at the same time in the same column based on the timestamp, ensuring accurate correspondence between time sequence and category.

[0047] Specifically, using timestamps as the sole alignment criterion, the normalized results of numerical information and the numerical encoding results of categorical information under the same timestamp are arranged in the same column of the matrix according to a preset information category order; the horizontal axis is arranged in chronological order, and the vertical axis is arranged in a fixed order according to information category, ultimately forming a two-dimensional coded data matrix.

[0048] For example, the codes corresponding to timestamp 20260324100000 are: Terminal Identifier: 001, Access Module: 101, Request Content: 1001, Request Data Length: 0.11. Using the timestamp as the horizontal axis and the information category as the vertical axis, fill in the above values ​​in the corresponding positions to form one column of a two-dimensional encoded data matrix; arranging the columns of multiple time points sequentially yields the complete two-dimensional encoded data matrix.

[0049] Furthermore, the texture feature difference analysis of the two-dimensional encoded data matrix is ​​performed using a multi-scale sliding window method, and the corresponding data texture verification results are output.

[0050] Specifically, a multi-scale sliding window method is used to perform texture feature difference analysis on the two-dimensional encoded data matrix, and the corresponding data texture verification result is output. Prior to this, the process includes: Based on the prior information set, periodic analysis is performed using information categories as indexes to obtain information fluctuation characteristics corresponding to multiple information categories; Based on the preset encoding rules, the position range corresponding to each information category is marked in the two-dimensional encoded data matrix; Based on the information fluctuation characteristics, information categories whose ratio of information fluctuation characteristics belongs to the integer neighborhood are combined, and multiple position interval combinations are formed accordingly. Based on the item with the largest information fluctuation feature of the information category contained in each of the aforementioned location interval combinations, the corresponding sliding window size is determined, and the sliding window size for different location interval combinations is configured differently.

[0051] First, based on the prior information set, periodic analysis is performed using information categories as indexes to obtain information fluctuation characteristics corresponding to multiple information categories. Information category indexing refers to a method of classifying, retrieving, and analyzing information in the prior information set using information category labels as unique identifiers. Periodic analysis refers to an analytical method that mines characteristics such as the periodicity and fluctuation amplitude of information values ​​by statistically analyzing the patterns of change in each information category over time. Information fluctuation characteristics characterize the changes in each information category over time, mainly including fluctuation amplitude, change period, and volatility rate, and are used to reflect the temporal change patterns of information.

[0052] Specifically, the edge agent uses information category labels as indexes to classify and split the prior information set, obtaining the time-series data sequence corresponding to each information category; it performs periodic analysis on the time-series data sequence of each information category: it uses the sliding window statistical method to calculate the data change period, obtains the volatility by calculating the coefficient of variation, and calculates the volatility amplitude by combining the maximum and minimum values; it summarizes the period, volatility, and volatility amplitude of each information category to form the information volatility characteristics corresponding to each information category.

[0053] For example, three information categories are extracted from the prior information set, and their fluctuation characteristics are analyzed: Request data length: small fluctuation amplitude, fluctuation rate 0.08, and change period of 1 hour; Access module: extremely small fluctuation amplitude, fluctuation rate 0.02, and no obvious change period; Request content: relatively large fluctuation amplitude, fluctuation rate 0.35, and change period of 2 hours.

[0054] Secondly, based on the preset encoding rules, the position interval corresponding to each information category is marked in the two-dimensional encoded data matrix. The position interval refers to the range of consecutive rows on the vertical axis corresponding to each information category in the two-dimensional encoded data matrix, which is used to clarify the specific position of different information categories in the matrix, facilitating subsequent sliding window scanning.

[0055] Specifically, based on the preset coding rules, the coding identifier corresponding to each information category is clearly defined; referring to the vertical axis of the two-dimensional coding data matrix, that is, the information category dimension, according to the coding order of the information categories, the continuous row range corresponding to each information category on the vertical axis is marked, that is, the position interval of each information category is obtained; the correspondence between each information category and the position interval is recorded to ensure that subsequent analysis can accurately locate the information.

[0056] For example, the vertical axis of the two-dimensional encoded data matrix is ​​arranged in the order of information category encoding, marking the position interval of each information category: terminal identifier (encoding 001): vertical axis row 1, position interval [1,1]; access module (encoding 101): vertical axis row 2, position interval [2,2]; request content (encoding 1001): vertical axis row 3, position interval [3,3]; request data length: vertical axis row 4, position interval [4,4].

[0057] Furthermore, based on the aforementioned information fluctuation characteristics, information categories whose ratios of information fluctuation characteristics fall within the integer neighborhood are grouped together, forming multiple position interval combinations. The ratio of information fluctuation characteristics refers to the ratio of corresponding fluctuation characteristics between any two information categories, used to determine whether the fluctuation patterns of the two types of information are similar. The integer neighborhood refers to a reasonable range where the ratio falls near an integer, such as between [k-0.1, k+0.1], where k is a positive integer, used to determine whether the fluctuation patterns of the two types of information are consistent. Position interval combinations refer to merging the position intervals corresponding to information categories with similar fluctuation patterns to form a joint analysis interval, used for subsequent unified configuration of the sliding window.

[0058] Specifically, calculate the fluctuation feature ratio between any two information categories; determine whether the ratio belongs to an integer neighborhood; if it does, group the two information categories together; repeat the above steps to divide all information categories into multiple groups, merge the position intervals corresponding to each group to form multiple position interval combinations; ensure that each information category belongs to only one position interval combination, with no repetition or omission.

[0059] For example, the ratio of volatility of each information category is calculated: the volatility of the request data length is 0.08, the volatility of the access module is 0.02, the ratio is 4.0, which belongs to the integer neighborhood [4-0.1, 4+0.1], and is grouped together; the volatility of the request content is 0.35, and the ratio of the other two categories is not in the integer neighborhood, so it is grouped separately; two position interval combinations are formed: combination 1: access module + request data length, the combined position intervals are [2,2] and [4,4]; combination 2: request content, position interval [3,3].

[0060] Finally, based on the information category with the largest fluctuation feature contained in each of the aforementioned position interval combinations, the corresponding sliding window size is determined, with different sliding window sizes configured for different position interval combinations. The information category with the largest fluctuation amplitude and rate in each position interval combination has the most complex fluctuation pattern, thus determining the sliding window size for that combination. Differentiated sliding window configuration means that different position interval combinations correspond to different sized sliding windows; the more complex the fluctuation feature, the larger the window size, ensuring the accuracy of texture feature capture.

[0061] Specifically, for each location interval combination, the information category with the largest information fluctuation feature is selected; the correspondence rules between the preset window size and the fluctuation feature are: fluctuation feature value ≤ 0.1, configure a 3×3 sliding window; 0.1 < fluctuation feature value ≤ 0.4, configure a 5×5 sliding window; fluctuation feature value > 0.4, configure a 7×7 sliding window; according to the fluctuation value of the item with the largest fluctuation feature in each combination, the corresponding window size is matched to complete the differentiated configuration of the sliding window for all location interval combinations.

[0062] For example, window sizes are configured for the two location interval combinations: Combination 1: the largest fluctuation feature is the length of the requested data, with a fluctuation rate of 0.08≤0.1, and a 3×3 sliding window is configured; Combination 2: the largest fluctuation feature is the requested content, with a fluctuation rate of 0.35, 0.1<0.35≤0.4, and a 5×5 sliding window is configured; finally, differentiated sliding window configurations for the two location interval combinations are achieved to adapt to their respective information fluctuation patterns.

[0063] Specifically, a multi-scale sliding window method is used to perform texture feature difference analysis on the two-dimensional encoded data matrix, and the corresponding output data texture verification results include: Based on the differentiated configuration of the sliding window size, combined with the preset backtracking window number constraint and backtracking step size ratio, a sliding scan is performed on the combination of multiple position intervals on the encoded data matrix. In each sliding scan, local texture feature values ​​are calculated for the window position, and the number of backtracking windows constrained by the local texture feature values ​​are arranged in chronological order to obtain multiple data texture sequences; Based on the data texture sequence, a priori region texture subsequences are extracted, and combined with statistical feature extraction and trend regression analysis, the difference between the priori region texture subsequences and the corresponding data texture sequences is calculated. The difference between multiple data texture sequences is weighted to obtain a global anomaly index. When the global anomaly index exceeds a preset anomaly threshold, the data texture verification result is output as abnormal; otherwise, the data texture verification result is output as normal. The weighting weights are determined based on the information category importance preset by the target manufacturer.

[0064] First, based on the differentiated sliding window size configuration, combined with preset backtracking window quantity constraints and backtracking step ratio, a sliding scan is performed on multiple combinations of the aforementioned position intervals on the encoded data matrix. Sliding scan refers to the process of moving a window with a preset window size across the two-dimensional encoded data matrix in a fixed direction and step size to complete local region traversal. The backtracking window quantity constraint limits the maximum number of window positions selectable within the prior information region, used to filter out excessively distant historical data and prevent invalid historical information from interfering with the current verification. The backtracking step ratio is the ratio of the backtracking step size to the window length, used to determine the step distance of the sliding window along the temporal axis. A smaller backtracking step ratio is used for high-fluctuation feature combinations to preserve detailed changes, while a larger backtracking step ratio is used for low-fluctuation feature combinations to improve processing efficiency.

[0065] Specifically, based on the differentiated sliding window size, the sliding scan is performed only along the temporal axis of the two-dimensional encoded data matrix for each position interval combination; the corresponding backtracking step size is determined according to the backtracking step size ratio: a smaller backtracking step size ratio is used for position interval combinations with high volatility characteristics, and a larger backtracking step size ratio is used for position interval combinations with low volatility characteristics; based on the constraint of the number of backtracking windows, the corresponding number of window positions within the prior information area are extracted to complete the traversal scan of the target area.

[0066] For example, combination 1 is a low volatility feature with a window length of 3, a backtracking step ratio of 2 / 3, corresponding to a backtracking step of 2 time units, and a backtracking window number constraint of 10 positions; combination 2 is a high volatility feature with a window length of 5, a backtracking step ratio of 1 / 5, corresponding to a backtracking step of 1 time unit, and a backtracking window number constraint of 10 positions; the sliding scan is completed along the time axis for the two position intervals respectively.

[0067] Secondly, during each sliding scan, local texture feature values ​​are calculated for the window position, and the number of backtracking windows constrained by the local texture feature values ​​are arranged in chronological order to obtain multiple data texture sequences.

[0068] In each sliding scan, local texture feature values ​​are calculated for the window position, including: For each window position, the value at the center of the window is defined as a threshold. Based on the relationship between the threshold and the values ​​of the neighboring positions at the center position within the window, the neighboring positions are binarized to generate binary codes. The merged output contains multiple binary codes representing positional relationships, which are then used to define the local texture feature values.

[0069] First, for each window position, the value of the window center position is defined as a threshold. The window center position refers to the geometric center point of the area covered by the sliding window in the two-dimensional encoded data matrix, and it is the reference position for calculating local texture features. The threshold is directly taken as the encoded value of the window center position.

[0070] Specifically, within the two-dimensional coded data matrix area covered by the current sliding window, locate the coordinates of the center position; read the coded value corresponding to the center position, and define this value as the threshold for this LBP calculation.

[0071] For example, if a 3×3 sliding window is used, and the center of the window corresponds to a coding value of 0.11 in the two-dimensional coding data matrix, then 0.11 will be used as the threshold for this binarization process.

[0072] Secondly, based on the relationship between the threshold and the values ​​of the neighboring positions within the window's center, the neighboring positions are binarized to generate binary codes. Neighboring positions refer to all valid positions surrounding the center within the sliding window, excluding the center position; these are feature acquisition points for local texture features. Binarization refers to the numerical conversion operation that maps the values ​​of the neighboring positions to 0 or 1 based on the relationship between the neighboring position values ​​and the threshold. Binary encoding refers to arranging the binarized results of each neighboring position into a binary string in a fixed positional order.

[0073] Specifically, it iterates through all neighboring positions outside the center position within the sliding window; it compares the value of each neighboring position with the threshold: if the value of the neighboring position is greater than or equal to the threshold, it is recorded as 1; if the value of the neighboring position is less than the threshold, it is recorded as 0; according to the preset fixed spatial position order, the binarized results of all neighboring positions are arranged sequentially to generate the corresponding binary code.

[0074] For example, a 3×3 window has 8 neighboring locations. After comparing them with the threshold of 0.11, the binarized result is obtained. Arranged in clockwise order, it is: 10101101, which generates the binary code: 10101101.

[0075] Next, the multiple binary codes containing spatial relationships are merged and output as the local texture feature values. Local texture feature values ​​are numerical values ​​obtained by merging and transforming the binary codes containing spatial relationships, used to uniquely represent the data texture features of the current window. Spatial relationships refer to the relative spatial order of neighboring locations within the window, which is strictly preserved in the binary codes.

[0076] Specifically, the spatial order of the binary encoding of the neighborhood is preserved without changing the encoding arrangement rules; the generated binary encoding is directly output as the local texture feature value; if it is necessary to participate in subsequent calculations in decimal form, the binary encoding can be converted into decimal values ​​while still retaining the texture feature information.

[0077] For example, the binary code 10101101 containing the positional relationship is directly merged and output as the local texture feature value of the current window; converted to decimal, it is 173, which can be directly used for subsequent texture sequence construction and difference calculation.

[0078] Furthermore, the local texture feature values ​​are arranged in chronological order according to the constraint of the number of backtracking windows to obtain multiple data texture sequences. A data texture sequence refers to a temporal feature sequence composed of multiple local texture feature values ​​combined from the same location interval and arranged in chronological order, used to reflect the texture change pattern of data in the temporal dimension. The constraint of the number of backtracking windows refers to the pre-set maximum number of selected window positions, used to limit the number of local texture feature values ​​participating in sequence construction, avoiding excessive historical data from interfering with verification accuracy.

[0079] Specifically, after completing the full window sliding scan and local texture feature value calculation for the current position interval combination, all calculated local texture feature values ​​are sorted according to the time sequence from far to near. Based on the preset backtracking window number constraint, a corresponding number of local texture feature values ​​are selected from the sorted feature values. The selected feature values ​​are then concatenated in chronological order to form the data texture sequence corresponding to the position interval combination. The above operation is repeated for all position interval combinations to finally obtain multiple data texture sequences with the same number of position interval combinations.

[0080] For example, the number of backtracking windows corresponding to a certain location interval combination is constrained to 10, and a total of 15 local texture feature values ​​are calculated. After sorting them in chronological order, the first 10 local texture feature values ​​are selected, namely: 10101101, 10101001, 10001101... The above 10 feature values ​​are arranged in chronological order to form a set of data texture sequences. The same operation is performed on another location interval combination, and finally multiple sets of data texture sequences are obtained.

[0081] Next, based on the data texture sequence, a priori region texture subsequences are extracted, and combined with statistical feature extraction and trend regression analysis, the difference between the priori region texture subsequences and the corresponding data texture sequences is calculated. The priori region texture subsequence refers to the baseline subsequence formed by extracting local texture feature values ​​belonging to prior normal information regions from the data texture sequence. The difference refers to the numerical value of the deviation between the texture sequence to be verified and the prior baseline subsequence, obtained through statistical feature and trend regression analysis.

[0082] Specifically, from the corresponding data texture sequence, the prior region texture subsequence corresponding to the prior information region is accurately extracted; the mean, variance, standard deviation and other statistical features are extracted from the two sets of sequences respectively, and the time series change trend is fitted by trend regression analysis; based on the difference of statistical features and the trend fitting deviation, the difference between the data texture sequence to be verified and the prior region texture subsequence is calculated.

[0083] For example, the prior region texture subsequence corresponding to combination 1 is extracted and compared with the texture sequence to be verified: the statistical feature difference is 0.03, the trend fitting deviation is 0.05, and the overall calculation shows that the sequence difference is 0.08.

[0084] Finally, the differences between multiple data texture sequences are weighted to obtain a global anomaly index. If the global anomaly index exceeds a preset anomaly threshold, the data texture verification result is output as an anomaly; otherwise, the data texture verification result is output as normal. The weighting weights are determined based on the information category importance preset by the target manufacturer. The global anomaly index is a comprehensive anomaly index obtained by weighting the differences between multiple location interval combinations, and it serves as the final judgment basis for data texture verification. The weighting weights are preset according to the information category importance to distinguish the degree of influence of different information categories on anomaly determination.

[0085] Specifically, according to the information category importance preset by the target manufacturer, the difference degree of each data texture sequence is assigned a corresponding weight; the difference degree is multiplied by the corresponding weight and then summed to obtain the global anomaly index; the global anomaly index is compared with the preset anomaly threshold: if it exceeds the threshold, the output data texture verification result is anomaly; if it does not exceed the threshold, the output verification result is normal.

[0086] For example, the preset weights are: Combination 1 difference weight 0.6, Combination 2 difference weight 0.4; Combination 1 difference 0.08, Combination 2 difference 0.12; Global anomaly score = 0.6 × 0.08 + 0.4 × 0.12 = 0.096; The preset anomaly threshold is 0.5, 0.096 < 0.5, therefore the output data texture verification result is normal.

[0087] The process of sending the real-time data request to the data relay based on the data texture verification result includes: When the data texture verification result is normal, the edge proxy performs edge lightweight desensitization on the sensitive data fields in the real-time data request; The de-identified real-time data request is encrypted at the field level using a pre-negotiated session key and a symmetric encryption algorithm. The real-time data request, encrypted at the field level, is sent to the data relay through a secure transmission channel established based on a cryptographic key negotiation protocol.

[0088] First, when the data texture verification result is normal, the edge proxy performs edge lightweight desensitization on sensitive data fields in the real-time data request. Sensitive data fields refer to key data items in the real-time data request that are prone to information leakage, such as device number, resource identifier, configuration parameters, etc. Edge lightweight desensitization refers to performing non-business-destructive obfuscation processing on sensitive data fields at the edge proxy end, reducing the risk of sensitive information leakage without affecting data transit and business parsing.

[0089] Specifically, the edge proxy reads the message structure of real-time data requests, identifies and locates sensitive data fields, and uses lightweight desensitization methods such as mask replacement and partial character hiding to process the sensitive data fields. The desensitization process only modifies the sensitive content and does not change the format, field length and business logic of the request message, ensuring that subsequent processes can be parsed normally.

[0090] For example, if a real-time data request contains the sensitive field "device unique identifier", some characters of this field can be masked and replaced to remove the sensitive information. After masking, only the field structure required for business identification is retained, and the sensitive information is hidden.

[0091] Secondly, the anonymized real-time data request is encrypted at the field level using a pre-negotiated session key and a symmetric encryption algorithm. The pre-negotiated session key is a temporary private key generated in advance between the edge agent and the data relay through a key negotiation protocol, used only for this data interaction. Field-level encryption involves encrypting each field in the data request as the smallest encryption unit, making it more flexible and lightweight compared to overall encryption. The symmetric encryption algorithm uses the same key for both encryption and decryption, featuring low computational overhead and high encryption efficiency, making it suitable for real-time data transmission at edge nodes.

[0092] Specifically, the edge proxy calls the session key that has been pre-negotiated with the data relay; selects national cryptographic or general symmetric encryption algorithms to encrypt the de-identified real-time data requests field by field; and retains the field separation structure after encryption to ensure that the data relay can be decrypted and parsed field by field.

[0093] For example, the de-identified real-time data request includes independent fields such as request identifier, request content, request time, and device number; the edge proxy uses the SM4 symmetric encryption algorithm and the same session key pre-negotiated with the data relay to perform independent encryption operations on each of the above fields; during the encryption process, the field separators and the overall message structure are preserved, and only the plaintext of each field is converted into the corresponding ciphertext, so as to finally obtain a field-level encrypted request message with a complete structure and normal parsing.

[0094] Furthermore, the real-time data request, encrypted at the field level, is sent to the data relay via a secure transmission channel established based on a cryptographic key agreement protocol. A cryptographic key agreement protocol is a protocol used by two communicating parties to securely exchange keys and establish an encrypted channel in an insecure network, ensuring the confidentiality and tamper-proof nature of the channel. A secure transmission channel is an encrypted transmission link established based on a key agreement protocol, preventing data from being eavesdropped on, tampered with, or forged during transmission.

[0095] Specifically, the edge agent completes identity authentication and key confirmation with the data relay through a cryptographic key negotiation protocol, establishing a stable and secure transmission channel; real-time data requests encrypted at the field level are encapsulated and transmitted through this secure channel; the data is encrypted throughout the transmission process, and the data relay can only receive it by decryption using the pre-negotiated session key.

[0096] In this embodiment of the invention, by constructing a set of information to be verified and a set of prior information, unified encoding and two-dimensional matrix organization of multiple types of data are achieved. Combined with multi-scale differentiated sliding windows, data texture feature difference analysis is completed, which can effectively identify content anomalies, temporal anomalies, and behavior spoofing requests. By configuring differentiated windows to adapt to the fluctuation patterns of different information, the accuracy and applicability of texture verification are improved. After verification, edge lightweight desensitization and field-level encryption are performed, which improves data transmission security while ensuring business availability. Through dual protection of identity verification and data texture verification, the legitimacy and security of data requests in heterogeneous server networking are further improved, providing reliable protection for the secure processing of subsequent data transfers.

[0097] S300: The data relay verifies the signature validity and data integrity of the real-time data request, and performs a second re-encryption on the real-time data request. After the second re-encryption, the request is forwarded to the target edge node via routing decision.

[0098] In this embodiment of the invention, the data relay verifies the signature validity and data integrity of the real-time data request, and performs a second re-encryption on the real-time data request. After the second re-encryption, it is forwarded to the target edge node via routing decision. In a multi-node heterogeneous server network environment, the data relay, as the intermediate interaction hub between the source edge node and the target edge node, needs to undertake the responsibilities of data security verification, protocol compatibility adaptation, and secure forwarding. Since different edge nodes use different communication protocols, direct forwarding will lead to data parsing failure. At the same time, although the data transmitted by the source edge node has been de-identified and encrypted, it still needs to be verified twice to confirm that the data has not been tampered with and the signature is valid. Furthermore, the de-identification and encryption methods need to be dynamically adjusted according to the security policy of the target edge node to avoid security risks caused by differences in the security requirements of the target node. In addition, traditional data relay only performs simple forwarding and lacks verification of data integrity and signature validity, which cannot guarantee the security and controllability of data transmission. Therefore, it is necessary to achieve secure, compatible, and accurate forwarding of data through signature verification, protocol conversion, dynamic de-identification, second re-encryption, and routing decision.

[0099] Step S300 in the method provided in this embodiment of the invention includes: Verify the validity of the digital signature of the real-time data request, and verify the data integrity by comparing the cryptographic hash value of the data; Based on the protocol adapter library, the real-time data requests of different communication protocols are parsed and converted into a unified data format, and the field-level decryption of the real-time data requests is performed using a pre-negotiated decryption key; Based on the decryption results, identify data classification and hierarchical labels, and perform dynamic desensitization based on a preset role-differentiated desensitization strategy; Based on the security policy of the target edge node, the dynamic de-identification result is re-encrypted and then forwarded to the target edge node through the path determined by the routing decision.

[0100] First, the validity of the digital signature of the real-time data request is verified, and data integrity is verified by comparing the cryptographic hash values ​​of the data. Digital signature validity verifies whether the digital signature carried in the real-time data request was generated by a legitimate sender and has not been tampered with, thus confirming the legitimacy of the data source. Data integrity is verified through cryptographic hash value comparison to confirm that the real-time data request has not been tampered with, lost, or destroyed during transmission, ensuring data consistency.

[0101] Specifically, the data relay receives real-time data requests encrypted at the field level, extracts the digital signature carried in the request and the cryptographic hash value corresponding to the original data; it calls the public key of the source edge node to decrypt and verify the digital signature. If the decryption result is consistent with the core information of the data request, the signature is deemed valid; the received real-time data request is re-hashd to obtain the current hash value, which is compared with the original hash value carried in the request. If the two are completely consistent, the data is deemed complete; only when the signature is valid and the data is complete can the subsequent processing steps be initiated; otherwise, the request is rejected and an exception message is returned.

[0102] For example, the data relay receives an asset ledger query request sent by the source edge node, extracts the digital signature and SHA-256 hash value from the request; calls the source edge node's public key to decrypt the digital signature, and if the decryption result is consistent with the request identifier, the signature is deemed valid; the SHA-256 hash value of the received request data is recalculated, and if it is completely consistent with the original hash value carried in the request, the data is deemed complete.

[0103] Secondly, based on the protocol adapter library, the real-time data requests from different communication protocols are parsed and converted into a unified data format, and the field-level decryption of the real-time data requests is performed using a pre-negotiated decryption key. The protocol adapter library refers to a collection of adaptation modules built into the data relay, containing multiple mainstream communication protocols, used to parse and convert data from different protocols. A unified data format, such as JSON, is used to solve parsing problems caused by differences in protocols between heterogeneous nodes. Field-level decryption refers to using the pre-negotiated decryption key to decrypt each data field encrypted by the source edge node one by one, restoring the anonymized plaintext data, providing a foundation for subsequent dynamic anonymization.

[0104] Specifically, the data relay uses a protocol adapter library to identify the communication protocol used by the real-time data request; it calls the corresponding protocol adaptation module to parse and convert the real-time data request of different protocol formats into a preset unified data format, retaining all field information and structure; it calls the decryption key pre-negotiated with the source edge node, and uses a symmetric encryption algorithm consistent with the source edge node, such as SM4, to perform decryption operations on the converted encrypted fields one by one, obtaining the desensitized plaintext data request.

[0105] For example, the data relay identifies that the received asset ledger query request uses the TCP protocol. Using the TCP protocol adaptation module in the protocol adapter library, it parses and converts the request into unified data in JSON format. The pre-negotiated SM4 decryption key is then used to decrypt the encrypted fields in the request, such as the request content and device identifier, one by one. After decryption, the anonymized plaintext request is obtained, where the device identifier has been anonymized using edge-lightweighting to HQ-ZC-. .

[0106] Secondly, based on the decryption results, data classification and grading labels are identified, and dynamic desensitization is performed based on a pre-defined role-based differentiated desensitization strategy. Data classification and grading labels are pre-assigned classification and security level identifiers for each field in the data request, used to distinguish the importance and sensitivity of the data. The role-based differentiated desensitization strategy refers to performing differentiated desensitization processing on data fields of different grades based on the access roles and business permissions of the target edge node, balancing security and business availability. Dynamic desensitization means adjusting the desensitization method and degree in real time according to the data classification and grading labels and the target role strategy, rather than using fixed desensitization rules, to adapt to the security requirements of different scenarios.

[0107] Specifically, the data relay parses the decrypted plaintext data requests, identifies the category-level tags corresponding to each field, reads the preset role-differentiated de-identification strategy, and determines the de-identification method corresponding to each level field based on the access role of the target edge node, such as operation and maintenance role or query role; performs corresponding de-identification processing on sensitive and core level fields, such as full masking of core fields and partial masking of sensitive fields, while ordinary level fields are not subject to additional de-identification, ensuring that the de-identified data meets security requirements without affecting the business parsing of the target node.

[0108] For example, the decrypted plaintext request includes the fields: request content, device identifier, and asset number; the target edge node is a query role, and the preset strategy is: partially masking sensitive fields and fully masking core fields; therefore, the device identifier is anonymized as HQ-ZC-, and the asset number is anonymized as... The requested content remains unchanged, thus completing the dynamic desensitization process.

[0109] Finally, based on the security policy of the target edge node, the dynamically anonymized result is re-encrypted and forwarded to the target edge node via the route decision. Secondary re-encryption refers to encrypting the dynamically anonymized request data again using a pre-negotiated key and encryption algorithm according to the target edge node's security policy, ensuring the security of data transmission from the data transfer point to the target edge node. Route decision refers to selecting the optimal transmission path from the preset routing path based on the target edge node's location, load status, and network link quality, ensuring efficient and stable data forwarding.

[0110] Specifically, the data relay obtains the security policy of the target edge node, determines the symmetric encryption algorithm and pre-negotiated session key used for secondary encryption; performs secondary encryption on the dynamically de-identified request data, using field-level encryption to preserve the data format and field structure, ensuring that the target edge node can decrypt it normally; calls the routing decision module to collect parameters such as load rate and transmission delay of each candidate routing path, and selects the optimal routing path; forwards the secondary-encrypted request data to the target edge node through the selected routing path, and records the forwarding log for subsequent auditing and traceability.

[0111] For example, the security policy of the target edge node specifies the use of the SM4 symmetric encryption algorithm. The data relay calls the session key pre-negotiated with the target node to perform secondary field-level encryption on the dynamically de-identified asset ledger query request. The routing decision module detects two candidate paths: path 1 has a load rate of 30% and a delay of 10ms, and path 2 has a load rate of 60% and a delay of 25ms. Path 1 is selected as the optimal route, and the encrypted request data is forwarded to the target edge node through path 1.

[0112] In this embodiment of the invention, secure, compatible, and accurate data transmission in heterogeneous server networks is achieved through multi-stage secure processing and adaptive forwarding of data relay: digital signature verification and hash value comparison provide dual protection for the legitimacy of data sources and the integrity of transmission, preventing the forwarding of tampered or forged data; a protocol adapter library enables unified conversion of different communication protocols, solving the protocol compatibility problem of heterogeneous nodes; dynamic desensitization is performed based on classification-level labeling and role-differentiation strategies to adapt to the security requirements of different target nodes, balancing data security and business availability; secondary re-encryption and optimal routing decisions further enhance data transmission security and forwarding efficiency, providing secure, standardized, and efficient data support for subsequent processing by target edge nodes, and improving the end-to-end data security protection system.

[0113] S400: The target edge node receives the data forwarded by the data relay, responds to the request, generates a digital signature, and returns it to the source edge node through the data relay.

[0114] In this embodiment of the invention, the target edge node receives the data forwarded by the data relay, performs a request response, generates a digital signature, and returns it to the source edge node through the data relay. In a multi-node heterogeneous server network environment, the target edge node, as the final processing end of the data request, receives data that has undergone secondary encryption and verification through data relay. However, it still needs to confirm the legality and usability of the data through decryption and integrity verification to avoid business anomalies caused by illegal or tampered data. At the same time, the response data may contain sensitive information such as business results and device status. Direct transmission back poses a risk of leakage, and the legality of the response source needs to be confirmed through digital signature to prevent the response data from being forged. In addition, direct transmission back between the source and target edge nodes has problems such as protocol incompatibility and transmission insecurity. It is necessary to use data relay to realize response transmission and ensure a closed loop of data interaction throughout the entire link. Therefore, it is necessary to complete the secure closed-loop transmission of response data through receiving verification, request response, encryption signature, and relay transmission.

[0115] In this context, the target edge node refers to the edge node that receives and forwards request data, performs business processing, and generates response data; it is the final processing terminal for data requests. Request-response refers to the process by which the target edge node parses the received request data, executes the corresponding business logic, and generates feedback data containing information such as the business processing result and response status. Digital signature refers to the characteristic value generated by the target edge node using its own private key to encrypt the response data; this signature is used to verify the legitimacy and integrity of the response data's source and prevent forgery or tampering.

[0116] Specifically, the target edge node receives the re-encrypted request data forwarded by the data relay through a secure transmission channel established based on a cryptographic key negotiation protocol; it calls the decryption key pre-negotiated with the data relay and uses the same symmetric encryption algorithm as the re-encryption to perform field-level decryption on the encrypted data, obtaining the dynamically anonymized request plaintext; it recalculates the cryptographic hash value of the decrypted request plaintext and compares it with the original hash value carried in the request to confirm the integrity and legitimacy of the data; if the verification passes, it parses the request type, parameters, and other core information of the request plaintext, executes the corresponding business processing, and generates response data containing business results, response status, and response time; it classifies and classifies the response data, performs field-level encryption on sensitive fields using the session key pre-negotiated with the data relay, and then uses its own private key to digitally sign the encrypted response data, generating a complete response message containing the response data, digital signature, and response hash value; it sends the complete response message to the data relay through the secure transmission channel, and after the data relay verifies the validity of the response signature, it forwards it to the source edge node through a preset routing path to complete the response return.

[0117] For example, the target edge node is an asset ledger management node. It receives a secondary SM4 encrypted request forwarded by the data relay through a secure channel, calls the pre-negotiated SM4 decryption key, decrypts the encrypted fields one by one, and obtains a dynamically anonymized asset ledger query request. It recalculates the SHA-256 hash value of the plaintext request and compares it with the original hash value carried in the request to confirm that the data is legal and complete. It parses the request as an office equipment ledger query request, executes the query business, obtains the results such as the quantity and inventory status of office equipment, and generates response data in a unified JSON format. It performs SM4 field-level encryption on the inventory quantity field in the response data, and then uses its own private key to digitally sign the encrypted response data to generate a response message containing the ciphertext of the response, the digital signature, and the response hash value. It sends the response message to the data relay through a secure channel. After the data relay verifies the signature, it forwards the response message to the source edge node, i.e., the logistics asset management server, through the optimal routing path to complete the request response and return.

[0118] In this embodiment of the invention, a closed-loop interaction of data requests and responses in a multi-node heterogeneous server network is realized. Decryption and integrity verification upon receipt ensure that the target edge node processes legitimate and complete data, preventing business anomalies. Encryption and digital signature of response data provide dual protection for the security and legitimacy of the response data's source, preventing forgery, tampering, or leakage. Response return is achieved through data relay, resolving protocol compatibility and transmission security issues in direct return between source and target edge nodes, ensuring a closed-loop link. Simultaneously, encryption, signing, and logging of the response process further improve the end-to-end data security protection system, providing support for traceability and auditability of data interaction, and enhancing the security, reliability, and closed-loop nature of data interaction in heterogeneous networks.

[0119] Through the specific implementation methods described above, the embodiments of the present invention achieve the following technical effects: This invention discloses a method and system for secure data interoperability in a multi-node heterogeneous server network. By verifying the identity and data texture of the source edge node, it achieves trusted access and content security detection for requests, and transmits data after de-identification and encryption. Data relay completes signature and integrity verification, protocol adaptation, dynamic de-identification, and secondary re-encryption, adapting to heterogeneous nodes and ensuring transmission security. The target edge node decrypts and verifies the data, generates a response, adds a digital signature, and sends it back to the source node via data relay, forming a closed-loop protection throughout the entire process. This invention effectively solves problems such as identity trust, content masquerading, protocol compatibility, and data leakage in heterogeneous server networks, achieving secure, compliant, and efficient end-to-end data interaction across the entire chain.

[0120] Example 2, as Figure 2 As shown, this invention provides a secure data interconnection system for multi-node heterogeneous server networking, the system comprising: The identity verification module 11 is used to intercept real-time data requests through an embedded edge proxy based on the source edge node of the heterogeneous server, and to perform request identity verification on the real-time data requests based on multi-dimensional trust assessment. The texture verification module 12 is used to collect the context information of the real-time data request based on the edge agent if the identity verification is successful, merge it into a set of information to be verified, perform data texture verification on the set of information to be verified, and send the real-time data request to the data relay according to the data texture verification result. The relay encryption forwarding module 13 is used to verify the signature validity and data integrity of the real-time data request during data relay, and to perform secondary re-encryption on the real-time data request. After secondary re-encryption, the request is forwarded to the target edge node through routing decision. The response signature module 14 is used for the target edge node to receive the data forwarded by the data relay, make a request response, generate a digital signature, and return it to the source edge node through the data relay.

[0121] In one embodiment, the identity verification module 11 is further configured to: Based on the certificate issuance level, certificate key length, and revocation status check results, the real-time data request is scored for authentication strength. Based on the device firmware version, the operating status of security protection components, and system integrity metrics, a device health score is performed on the real-time data request. Based on the degree of deviation between the current request frequency, requested data volume, and accessed resource type and the historical behavior baseline, the real-time data request is scored for behavioral baseline deviation. The normalized weighted identity authentication strength score, device health score, and behavioral baseline deviation score are used to obtain a comprehensive trust score. When the comprehensive trust score meets the preset trust threshold, the identity verification is deemed to have passed. The normalized weighted weights are selected from preset weight combinations based on the data security level of the target scenario.

[0122] In one embodiment, the texture verification module 12 is further configured to: If the identity verification passes, the context information of the real-time data request is collected based on the edge proxy, merged into a set of information to be verified, and data texture verification is performed on the set of information to be verified, including: Obtain the context information of the real-time data request, and merge the real-time data request and the context information into a set of information to be verified, wherein each information item includes an information category label, an information value, and a timestamp; Based on local historical records, a priori information set marked as normal state is extracted, wherein the priori information set includes prior data requests and prior context information; By using preset encoding rules, the prior information set and the information set to be verified are uniformly encoded into an encoded data sequence, and the encoded data sequence is sorted according to time series labels and organized according to information categories to generate a two-dimensional encoded data matrix.

[0123] The texture feature difference analysis of the two-dimensional encoded data matrix is ​​performed using a multi-scale sliding window method, and the corresponding data texture verification results are output.

[0124] Specifically, by using preset encoding rules, the prior information set and the information set to be verified are uniformly encoded into an encoded data sequence. This encoded data sequence is then sorted according to time-series labels and organized according to information categories to generate a two-dimensional encoded data matrix, including: The numerical information values ​​in the prior information set and the information set to be verified are normalized. Based on preset encoding rules, categorical information values ​​are numerically encoded and converted into numerical vector representations; The normalization result and the numerical vector representation are time-aligned according to the timestamp to form the two-dimensional encoded data matrix. The horizontal axis of the encoded data matrix is ​​the time dimension, the vertical axis is the information category dimension, and the value of each matrix element is the encoded value of the corresponding information category at the corresponding time point.

[0125] Specifically, a multi-scale sliding window method is used to perform texture feature difference analysis on the two-dimensional encoded data matrix, and the corresponding data texture verification result is output. Prior to this, the process includes: Based on the prior information set, periodic analysis is performed using information categories as indexes to obtain information fluctuation characteristics corresponding to multiple information categories; Based on the preset encoding rules, the position range corresponding to each information category is marked in the two-dimensional encoded data matrix; Based on the information fluctuation characteristics, information categories whose ratio of information fluctuation characteristics belongs to the integer neighborhood are combined, and multiple position interval combinations are formed accordingly. Based on the item with the largest information fluctuation feature of the information category contained in each of the aforementioned location interval combinations, the corresponding sliding window size is determined, and the sliding window size for different location interval combinations is configured differently.

[0126] Specifically, a multi-scale sliding window method is used to perform texture feature difference analysis on the two-dimensional encoded data matrix, and the corresponding output data texture verification results include: Based on the differentiated configuration of the sliding window size, combined with the preset backtracking window number constraint and backtracking step size ratio, a sliding scan is performed on the combination of multiple position intervals on the encoded data matrix. In each sliding scan, local texture feature values ​​are calculated for the window position, and the number of backtracking windows constrained by the local texture feature values ​​are arranged in chronological order to obtain multiple data texture sequences; Based on the data texture sequence, a priori region texture subsequences are extracted, and combined with statistical feature extraction and trend regression analysis, the difference between the priori region texture subsequences and the corresponding data texture sequences is calculated. The difference between multiple data texture sequences is weighted to obtain a global anomaly index. When the global anomaly index exceeds a preset anomaly threshold, the data texture verification result is output as abnormal; otherwise, the data texture verification result is output as normal. The weighting weights are determined based on the information category importance preset by the target manufacturer.

[0127] In each sliding scan, local texture feature values ​​are calculated for the window position, including: For each window position, the value at the center of the window is defined as a threshold. Based on the relationship between the threshold and the values ​​of the neighboring positions at the center position within the window, the neighboring positions are binarized to generate binary codes. The merged output contains multiple binary codes representing positional relationships, which are then used to define the local texture feature values.

[0128] The process of sending the real-time data request to the data relay based on the data texture verification result includes: When the data texture verification result is normal, the edge proxy performs edge lightweight desensitization on the sensitive data fields in the real-time data request; The de-identified real-time data request is encrypted at the field level using a pre-negotiated session key and a symmetric encryption algorithm. The real-time data request, encrypted at the field level, is sent to the data relay through a secure transmission channel established based on a cryptographic key negotiation protocol.

[0129] In one embodiment, the relay encryption forwarding module 13 is further configured to: Verify the validity of the digital signature of the real-time data request, and verify the data integrity by comparing the cryptographic hash value of the data; Based on the protocol adapter library, the real-time data requests of different communication protocols are parsed and converted into a unified data format, and the field-level decryption of the real-time data requests is performed using a pre-negotiated decryption key; Based on the decryption results, identify data classification and hierarchical labels, and perform dynamic desensitization based on a preset role-differentiated desensitization strategy; Based on the security policy of the target edge node, the dynamic de-identification result is re-encrypted and then forwarded to the target edge node through the path determined by the routing decision.

[0130] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for secure data interoperability in a multi-node heterogeneous server network, characterized in that, include: The source edge node based on heterogeneous servers intercepts real-time data requests through an embedded edge proxy and performs request identity verification based on multi-dimensional trust assessment on the real-time data requests. If the identity verification passes, the context information of the real-time data request is collected based on the edge agent, merged into a set of information to be verified, the set of information to be verified is subjected to data texture verification, and the real-time data request is sent to the data relay according to the data texture verification result. The data relay verifies the signature validity and data integrity of the real-time data request, and performs a second re-encryption on the real-time data request. After the second re-encryption, the request is forwarded to the target edge node via routing decision. The target edge node receives the data forwarded by the data relay, responds to the request, generates a digital signature, and returns it to the source edge node through the data relay.

2. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 1, characterized in that, The source edge node based on heterogeneous servers intercepts real-time data requests through an embedded edge proxy and performs request identity verification based on multi-dimensional trust assessment on the real-time data requests, including: Based on the certificate issuance level, certificate key length, and revocation status check results, the real-time data request is scored for authentication strength. Based on the device firmware version, the operating status of security protection components, and system integrity metrics, a device health score is performed on the real-time data request. Based on the degree of deviation between the current request frequency, requested data volume, and accessed resource type and the historical behavior baseline, the real-time data request is scored for behavioral baseline deviation. The normalized weighted identity authentication strength score, device health score, and behavioral baseline deviation score are used to obtain a comprehensive trust score. When the comprehensive trust score meets the preset trust threshold, the identity verification is deemed to have passed. The normalized weighted weights are selected from preset weight combinations based on the data security level of the target scenario.

3. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 1, characterized in that, If the identity verification passes, the context information of the real-time data request is collected based on the edge proxy, merged into a set of information to be verified, and data texture verification is performed on the set of information to be verified, including: Obtain the context information of the real-time data request, and merge the real-time data request and the context information into a set of information to be verified, wherein each information item includes an information category label, an information value, and a timestamp; Based on local historical records, a priori information set marked as normal state is extracted, wherein the priori information set includes prior data requests and prior context information; By using preset encoding rules, the prior information set and the information set to be verified are uniformly encoded into an encoded data sequence, and the encoded data sequence is sorted according to time series labels and organized according to information categories to generate a two-dimensional encoded data matrix; The texture feature difference analysis of the two-dimensional encoded data matrix is ​​performed using a multi-scale sliding window method, and the corresponding data texture verification results are output.

4. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 3, characterized in that, By using preset encoding rules, the prior information set and the information set to be verified are uniformly encoded into an encoded data sequence. The encoded data sequence is then sorted according to time-series labels and organized according to information categories to generate a two-dimensional encoded data matrix, including: The numerical information values ​​in the prior information set and the information set to be verified are normalized. Based on preset encoding rules, categorical information values ​​are numerically encoded and converted into numerical vector representations; The normalization result and the numerical vector representation are time-aligned according to the timestamp to form the two-dimensional encoded data matrix. The horizontal axis of the encoded data matrix is ​​the time dimension, the vertical axis is the information category dimension, and the value of each matrix element is the encoded value of the corresponding information category at the corresponding time point.

5. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 3, characterized in that, The texture feature difference analysis of the two-dimensional coded data matrix is ​​performed using a multi-scale sliding window method, and the corresponding data texture verification result is output. Prior to this, the analysis includes: Based on the prior information set, periodic analysis is performed using information categories as indexes to obtain information fluctuation characteristics corresponding to multiple information categories; Based on the preset encoding rules, the position range corresponding to each information category is marked in the two-dimensional encoded data matrix; Based on the information fluctuation characteristics, information categories whose ratio of information fluctuation characteristics belongs to the integer neighborhood are combined, and multiple position interval combinations are formed accordingly. Based on the item with the largest information fluctuation feature of the information category contained in each of the aforementioned location interval combinations, the corresponding sliding window size is determined, and the sliding window size for different location interval combinations is configured differently.

6. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 5, characterized in that, The texture feature difference analysis of the two-dimensional coded data matrix is ​​performed using a multi-scale sliding window method, and the corresponding output data texture verification results include: Based on the differentiated configuration of the sliding window size, combined with the preset backtracking window number constraint and backtracking step size ratio, a sliding scan is performed on the combination of multiple position intervals on the encoded data matrix. In each sliding scan, local texture feature values ​​are calculated for the window position, and the number of backtracking windows constrained by the local texture feature values ​​are arranged in chronological order to obtain multiple data texture sequences; Based on the data texture sequence, a priori region texture subsequences are extracted, and combined with statistical feature extraction and trend regression analysis, the difference between the priori region texture subsequences and the corresponding data texture sequences is calculated. The difference between multiple data texture sequences is weighted to obtain a global anomaly index. When the global anomaly index exceeds a preset anomaly threshold, the data texture verification result is output as abnormal; otherwise, the data texture verification result is output as normal. The weighting weights are determined based on the information category importance preset by the target manufacturer.

7. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 1, characterized in that, During each sliding scan, local texture feature values ​​are calculated for the window position, including: For each window position, the value at the center of the window is defined as a threshold. Based on the relationship between the threshold and the values ​​of the neighboring positions at the center position within the window, the neighboring positions are binarized to generate binary codes. The merged output contains multiple binary codes representing positional relationships, which are then used to define the local texture feature values.

8. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 1, characterized in that, Based on the data texture verification result, the real-time data request is sent to the data relay, including: When the data texture verification result is normal, the edge proxy performs edge lightweight desensitization on the sensitive data fields in the real-time data request; The de-identified real-time data request is encrypted at the field level using a pre-negotiated session key and a symmetric encryption algorithm. The real-time data request, encrypted at the field level, is sent to the data relay through a secure transmission channel established based on a cryptographic key negotiation protocol.

9. The method for secure data interoperability in a multi-node heterogeneous server network as described in claim 1, characterized in that, The data relay verifies the signature validity and data integrity of the real-time data request, and performs a second re-encryption on the real-time data request. After the second re-encryption, the request is forwarded to the target edge node via routing decision, including: Verify the validity of the digital signature of the real-time data request, and verify the data integrity by comparing the cryptographic hash value of the data; Based on the protocol adapter library, the real-time data requests of different communication protocols are parsed and converted into a unified data format, and the field-level decryption of the real-time data requests is performed using a pre-negotiated decryption key; Based on the decryption results, identify data classification and hierarchical labels, and perform dynamic desensitization based on a preset role-differentiated desensitization strategy; Based on the security policy of the target edge node, the dynamic de-identification result is re-encrypted and then forwarded to the target edge node through the path determined by the routing decision.

10. A multi-node heterogeneous server networking data security interoperability system, characterized in that, The method for secure data interoperability in a multi-node heterogeneous server network as described in any one of claims 1-9 includes: The identity verification module is used to intercept real-time data requests from the source edge nodes of heterogeneous servers through an embedded edge proxy, and to perform request identity verification on the real-time data requests based on multi-dimensional trust assessment. The texture verification module is used to collect the context information of the real-time data request based on the edge agent if the identity verification is successful, merge it into a set of information to be verified, perform data texture verification on the set of information to be verified, and send the real-time data request to the data relay according to the data texture verification result. The relay encryption forwarding module is used to verify the signature validity and data integrity of the real-time data request during data relay, and to perform secondary re-encryption on the real-time data request. After secondary re-encryption, the request is forwarded to the target edge node through routing decision. The response signature module is used by the target edge node to receive the data forwarded by the data relay, make a request response, generate a digital signature, and return it to the source edge node through the data relay.

Images