Enterprise Intranet Data Loss Prevention and Tracing Methods Based on Zero Trust Architecture

By using a zero-trust architecture-based method for preventing data leakage and tracing within an enterprise intranet, device trust anchors and data flow permission credentials are generated. The evidence chain of the flow trajectory is evaluated and constructed in real time, solving the problem of decoupling trust status and data protection level in traditional systems. This enables real-time response and accurate tracing for preventing data leakage and tracing.

CN123122873APending Publication Date: 2026-11-17JILIN TIANQIKAI NETWORK TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611121903.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-27
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In traditional enterprise intranet data leakage prevention and tracing systems, identity authentication and data security protection operate decoupled, resulting in the inability to synchronize trust status and data protection level in real time, and leakage tracing lacks technical credibility and complete evidence support.

Method used

Based on a zero-trust architecture, device trust anchors are generated by acquiring the hardware characteristics, operating environment, and user identity of terminal devices. Dynamic session tokens are derived and combined with the security level identifier of sensitive data and the target information of the data flow to generate data flow permission credentials. Trust assessment and comparison are performed in real time at intermediate nodes to form a controlled data encapsulation. A chain of evidence for the flow trajectory is constructed through forward hashing.

Benefits of technology

It achieves real-time coupling of trust status and security level during data flow, enabling it to detect trust degradation at critical stages and respond in real time, reducing the risk of leakage. Furthermore, it accurately pinpoints the nodes and temporal locations of trust chain breaks through cryptographically verifiable flow trajectory evidence chains, thereby improving the credibility and integrity of tracing conclusions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122873A_ABST
    Figure CN123122873A_ABST
Patent Text Reader

Abstract

This invention belongs to the field of data security technology. It discloses a method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture. The method includes: acquiring hardware characteristic information, operating environment status, and user identity credentials of terminal devices; generating device trust anchors and deriving dynamic session tokens through multi-dimensional feature fusion; combining the dynamic session tokens with the security level identifiers of sensitive data to generate data flow permission credentials, and cryptographically binding them with the sensitive data to form a controlled data encapsulation; comparing real-time trust assessment values ​​with trust access thresholds at each intermediate flow node, performing corresponding actions, and generating node flow evidence records; anchoring each evidence record through forward hashing to form a flow trajectory evidence chain and writing it into a distributed evidence node cluster; when a leakage event is detected, extracting the evidence chain, reversing the traversal to locate trust breakpoints, and outputting a tracing evidence report; significantly improving the effectiveness of preventing data leakage and accurate tracing within an enterprise intranet.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of data security technology, and more specifically, to a method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture. Background Technology

[0002] With the deepening of enterprise digital transformation and the widespread adoption of remote collaborative work models, the flow of sensitive data carried on enterprise intranets between terminal devices, business systems, and service nodes is becoming increasingly frequent and the paths are becoming more complex. Under this trend, the ability to prevent data leakage and trace leakage incidents within enterprise intranets has become a core technical requirement for enterprise information security defense-in-depth systems. To effectively prevent data leakage risks, enterprises are now generally deploying data leakage prevention systems based on network boundary protection. By setting content detection and behavior auditing strategies at key nodes such as gateways and terminals, combined with one-time identity authentication mechanisms, they monitor and intercept the outward transmission of sensitive data.

[0003] Traditional data loss prevention (DLP) architectures treat identity authentication and data security protection as two decoupled subsystems. The identity authentication system grants persistent session trust after a one-time verification upon user access, while the DLP independently performs data content inspection and transmission behavior review at pre-defined boundary nodes based on static rules. This architectural decoupling means that as data flows through multiple nodes within the intranet, intermediate processing stages cannot synchronously verify the real-time trust status of the accessing entity against the required security protection level. When an initially authenticated terminal device is compromised by malware during the session, or when internal personnel gradually aggregate sensitive data fragments using legitimate access permissions scattered across multiple business systems, the system... The inability to perceive that the trust foundation supporting the data access behavior has deteriorated or even completely failed makes it impossible to block ongoing leakage in real time at critical stages of data flow. More importantly, when conducting forensic investigation and tracing after a leakage incident, the lack of a cryptographically verifiable link between the static session logs generated during identity authentication and the discrete operation records generated during data flow makes it impossible for the tracing system to reconstruct the complete flow path of data from the source storage node to the leakage exit, or to accurately pinpoint the specific node and timing of the break in the trust chain. This results in leakage tracing conclusions lacking sufficient technical credibility and complete evidentiary support, ultimately limiting the effective application of data leakage prevention and accurate tracing technologies in enterprise intranet security practices.

[0004] In view of this, the present invention proposes a method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of existing technologies and achieve the above objectives, this invention provides the following technical solution: a method for preventing and tracing data leakage in enterprise intranets based on a zero-trust architecture, comprising: Step S1: Obtain the hardware feature information, operating environment status and user identity credentials of the terminal device, generate the device trust anchor point through multi-dimensional feature fusion, and derive the dynamic session token based on the device trust anchor point. Step S2: Obtain the security level identifier and target node information of the sensitive data, generate a data transfer permission certificate by combining the dynamic session token, and cryptographically bind the data transfer permission certificate to the sensitive data to form a controlled data encapsulation. Step S3: At each intermediate flow node through which the controlled data encapsulation passes, extract the real-time trust assessment value of the current node and compare it with the trust access threshold in the data flow license certificate. Based on the comparison result, perform release, block or downgrade actions and generate node flow evidence records. Step S4: Link the flow and storage records of each node in chronological order, form a flow trajectory evidence chain by forward hashing, and synchronously write the flow trajectory evidence chain into the distributed storage node cluster. Step S5: When a data breach is detected, extract the evidence chain of the flow trajectory from the distributed evidence storage node cluster, reverse traverse the flow evidence storage records of each node to locate the trust break node, and output a source tracing evidence report containing the complete flow path and break location.

[0006] The technical effects and advantages of this invention's enterprise intranet data leakage prevention and tracing method based on zero-trust architecture are as follows: This invention forms a controlled data encapsulation by cryptographically binding dynamic session tokens and the security level identifiers of sensitive data. This ensures that data carries a trust state and security level constraint throughout the entire flow, achieving architectural-level coupling between identity authentication and data security protection. This avoids the problem of decoupling in traditional architectures, which leads to a lack of synchronous verification of trust state and data protection level. At each intermediate flow node, a node trust assessment value is collected in real time and compared with a trust access threshold. Based on the comparison result, actions are taken to allow, downgrade, or block access. This allows the system to detect and respond in real time to deterioration of the trust foundation at critical stages of data flow, reducing the risk of authenticated terminals being compromised or internal personnel using legitimate privileges. The risk of leakage not being promptly stopped when gradually accumulating sensitive data fragments is addressed. By using forward hashing to anchor the flow records of each node, a cryptographically verifiable chain of evidence is formed and synchronously written into a distributed evidence node cluster. This establishes a cryptographic link between identity authentication information and data flow operation records throughout the entire chain. When a leakage event is detected, the chain of evidence is traced in reverse to locate the node where trust breaks. This allows for the reconstruction of the complete flow path of data from the source to the leakage exit and precise identification of the specific node and timing of the trust chain break. This improves the technical credibility and evidentiary integrity of the leakage tracing conclusions, enhancing the effectiveness of enterprise intranet data leakage prevention and precise tracing. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the enterprise intranet data leakage prevention and tracing method based on zero-trust architecture of the present invention. Detailed Implementation

[0008] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0009] Example 1 Please see Figure 1 As shown, this embodiment describes a method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture, including: Step S1: Obtain the hardware feature information, operating environment status and user identity credentials of the terminal device, generate the device trust anchor through multi-dimensional feature fusion, and derive the dynamic session token based on the device trust anchor.

[0010] In traditional enterprise intranet security systems, identity authentication typically employs a one-time verification model, where users gain continuous session trust after completing identity verification upon accessing the intranet. However, this static trust granting mechanism has significant security vulnerabilities: when a terminal device that has passed initial authentication is compromised by malicious programs during the session, or when the device's operating environment deteriorates, the system cannot detect that the trust foundation supporting data access has changed. Therefore, this solution uses multi-dimensional feature fusion technology to generate device trust anchors, cryptographically binding hardware characteristics, operating environment, and user identity, laying the foundation for real-time trust verification during subsequent data flow.

[0011] Preferably, in some possible implementations of the embodiments of the present invention, step S1 specifically includes: Hardware characteristic information collection and device hardware fingerprint generation involves collecting the terminal device's motherboard serial number, chip identifier, and network card physical address as hardware characteristic information. The motherboard serial number is obtained by reading the DMI information in the terminal device's BIOS, and is usually a 12 to 20-digit alphanumeric combination, such as "PF3KXYZ7890123"; the chip identifier is obtained by reading the CPU's CPUID instruction return value, which includes the processor model, family code, and extended characteristic identifier, such as "GenuineIntel-0x906EA-0x00050654"; the network card physical address, i.e., the MAC address, is obtained by querying the operating system's network interface and is a 48-bit hexadecimal number, such as "00:1A:2B:3C:4D:5E".

[0012] Hardware characteristic information is concatenated in a preset order to form a device hardware fingerprint string. The preset order is: motherboard serial number first, chip identifier in the middle, and network card physical address last, with each part connected by the separator "|". For example, the device hardware fingerprint string of a certain terminal device is: The fingerprint string is defined as “PF3KXYZ7890123|GenuineIntel-0x906EA-0x00050654|00:1A:2B:3C:4D:5E”. It should be noted that the generation of the device hardware fingerprint string employs deterministic concatenation rules to ensure that the device hardware fingerprint string generated by the same terminal device at different times remains consistent, thus providing a stable feature foundation for subsequent device identity binding. In other embodiments, other hardware features such as hard drive serial numbers and TPM chip identifiers can also be incorporated into the generation process of the device hardware fingerprint string.

[0013] The runtime environment status is collected by obtaining the operating system kernel version currently running on the terminal device and the running status of the security software. The operating system kernel version is obtained through system call interfaces, such as the kernel version "10.0.19045" for Windows or "5.15.0-76-generic" for Linux. The running status of the security software includes: whether the endpoint protection agent deployed by the enterprise is running normally (a value of 1 indicates normal operation, and a value of 0 indicates abnormality or not running), the interval between the virus database update time and the current time (in hours), and the result code of the most recent security scan (a value of 0 indicates no threat detected, and a non-zero value indicates the number of threats detected). The above information is encoded into a runtime environment status description and represented in JSON format.

[0014] The collection of operating environment status enables device trust anchors to reflect the real-time security status of terminal devices. When the device operating environment deteriorates (such as when the protection software is turned off or the virus database has not been updated for a long time), the trust status can be updated by regenerating the device trust anchors, thereby achieving dynamic trust assessment.

[0015] User identity credential processing receives user identity credentials input by the user, including a user identifier and an authentication key. The user identifier is a unique user number assigned by the enterprise unified identity management system, such as "EMP20231025001"; the authentication key is the user's login password or digital certificate private key.

[0016] A one-way hash operation is performed on the authentication key to obtain the credential digest value. This embodiment uses the SHA-256 hash algorithm to convert an authentication key of arbitrary length into a fixed 256-bit (32-byte) credential digest value. The irreversible nature of the one-way hash operation ensures that even if the credential digest value is intercepted, an attacker cannot recover the original authentication key, thereby protecting the security of user identity credentials.

[0017] The device hardware fingerprint string, operating environment status, and credential digest value are used as inputs to generate a fixed-length device trust anchor through a key derivation function. In this embodiment, PBKDF2 is used as the key derivation function, with 10,000 iterations and an output length of 256 bits.

[0018] The formula for generating device trust anchors is expressed as follows:

[0019] In the formula, As a trusted anchor point for the device; For device hardware fingerprinting; Description of the operating environment status; This represents the voucher summary value; "||" indicates a string concatenation operation. The system-preset random salt value is 128 bits long and is used to enhance the security of key derivation; 10000 is the number of iterations; and 256 is the number of output bits.

[0020] Setting the number of iterations to 10,000 effectively resists brute-force attacks while ensuring efficient key derivation calculations. Based on the computing power of current mainstream computing devices, 10,000 iterations can control the time for a single key derivation calculation to within 50 milliseconds, while simultaneously increasing the cost of brute-force attacks to an unacceptable level. The number of iterations can range from 5,000 to 50,000, and implementers can adjust it according to the computing power and security requirements of the terminal device.

[0021] Device trust anchors cryptographically fuse information from three dimensions: hardware characteristics, operating environment, and user identity. This ensures that any change in any dimension will lead to a change in the device trust anchor, thereby achieving a comprehensive characterization of the trusted state of the terminal device.

[0022] A dynamic session token is generated by performing an XOR obfuscation operation between the device trust anchor and the current timestamp, and then appending an expiration date field. The Unix timestamp of the current system time (accurate to milliseconds), such as "1698234567890", is obtained. This timestamp is converted to a 256-bit binary representation (padded with zeros at the high bits) and then XORed with the device trust anchor to obtain the obfuscation result. The dynamic session token structure includes three parts: the obfuscation result (256 bits), the expiration date field (32 bits), and the token version number (8 bits). The expiration date field records the token's expiration time, represented as a Unix timestamp. In this embodiment, the default expiration date of the dynamic session token is set to 30 minutes, meaning the value of the expiration date field is the current timestamp plus 1,800,000 milliseconds. Setting the expiration date to 30 minutes limits the usable time window after the token is stolen while ensuring the continuity of user operations. The expiration date can range from 15 minutes to 120 minutes, and implementers can adjust it according to the security sensitivity of the business scenario.

[0023] The formula for generating dynamic session tokens is expressed as follows:

[0024] In the formula, For dynamic session tokens; This indicates a bitwise XOR operation; This is the 256-bit binary representation of the current timestamp; For validity period field; This is the token version number, with an initial value of 0x01.

[0025] It's important to note that the time-limited design of dynamic session tokens ensures that even if a token is intercepted, attackers can only utilize it within a limited time window, significantly reducing the security risks associated with token leakage. Furthermore, the token version number field supports subsequent token update mechanisms; the version number increments with each token update, making it easy to distinguish tokens generated by the same device at different times.

[0026] Step S2: Obtain the security level identifier and target node information of the sensitive data, generate a data transfer permission certificate by combining the dynamic session token, and cryptographically bind the data transfer permission certificate to the sensitive data to form a controlled data encapsulation.

[0027] In traditional data loss prevention architectures, data security and identity authentication operate as two decoupled subsystems. This results in sensitive data failing to carry control information associated with the trust status of the currently accessing entity when it flows through the internal network. This step cryptographically binds dynamic session tokens, data security level information, and flow path information, ensuring that sensitive data always carries verifiable authorization credentials during the flow process, providing a basis for real-time trust verification at subsequent intermediate nodes.

[0028] Preferably, in some possible implementations of the embodiments of the present invention, step S2 specifically includes: Read the metadata header of the sensitive data and extract the security classification identifier field. Sensitive data is marked in the enterprise intranet according to a unified data classification and grading standard. The metadata header is a fixed-length data block (64 bytes in this embodiment) appended to the beginning of the sensitive data body, in which the security classification identifier field occupies the positions of bytes 17 to 32.

[0029] The security classification field includes two subfields: a data protection level code and a minimum trust threshold. The data protection level code is a 4-digit numeric code, ranging from 0001 to 0004, corresponding to the four protection levels of "Public," "Internal," "Confidential," and "Top Secret," respectively. The minimum trust threshold is a 2-digit hexadecimal number, ranging from 0x00 to 0xFF (corresponding to decimal 0 to 255), representing the minimum trust assessment value required to access the data.

[0030] For example, the security classification field of a confidential R&D design document is "0003-0xC8", which means that the data protection level is confidential (level code 0003). Accessing this data requires the real-time trust assessment value of the intermediate flow node to be no less than 200 (hexadecimal 0xC8 corresponds to decimal 200).

[0031] The correspondence between data protection levels and minimum trust access thresholds is shown in the table below: Data protection level Level Code Minimum Trust Access Threshold Range Typical values Public 0001 0x00-0x3F 0x20(32) Internal level 0002 0x40-0x7F 0x60(96) Confidential 0003 0x80-0xBF 0xA0 (160) Top Secret 0004 0xC0-0xFF 0xE0 (224)

[0032] The data transfer path information is encoded to obtain the target node address and expected transfer path sequence specified in the data transfer request. The target node address is the network identifier of the business system or terminal device that will ultimately receive the sensitive data, and is represented in the format of "node type:node identifier", for example, "SERVER:BIZ-SYS-001" represents the business system server node.

[0033] The expected flow path sequence records a list of intermediate nodes that sensitive data passes through from the source node to the target node, arranged in the flow order. For example, the expected flow path sequence for a sensitive document from the R&D department server to the finance department terminal is: ["GW-RD-001", "CORE-SW-002", "FW-FIN-001", "GW-FIN-001"], representing the R&D department gateway, core switch, finance department firewall, and finance department gateway, respectively.

[0034] The target node address and the expected flow path sequence are encoded into a path description block. The path description block adopts the TLV encoding format and has the following structure: Type field (1 byte): 0x01 represents the target node address, 0x02 represents the expected flow path sequence; Length field (2 bytes): records the byte length of the value field; Value field (variable length): the specific node address or path sequence data.

[0035] The design of the path description block enables the data flow permission certificate to clearly define the legal flow range of sensitive data, and any flow behavior that deviates from the expected path will be identified and processed by the intermediate nodes.

[0036] The dynamic session token, security level identifier field, and path description block are concatenated, and an asymmetric encryption algorithm is used to sign the data owner's private key to generate a data transfer permission certificate.

[0037] This embodiment uses the ECDSA signature algorithm with a secp256r1 elliptic curve. The data owner's private key is the private key held by the original owner of the sensitive data (usually the data creator or the department to which it belongs) in the enterprise PKI system.

[0038] The structure of the data flow license certificate includes: a certificate header (8 bytes, containing the certificate version number and signature algorithm identifier), a dynamic session token (37 bytes), a security level identifier field (16 bytes), a path description block (variable length), and a digital signature (64 bytes).

[0039] The formula for generating a data transfer license certificate is as follows: In the formula, For data transfer license certificate; This is the header of the voucher; This is a security classification field. For path description blocks; For ECDSA signature functions; The private key of the data owner; This is the SHA-256 hash function. Digital signatures ensure the integrity and non-repudiation of data transfer authorization credentials; any tampering with the credential content will cause signature verification to fail, thus protecting the authorization validity of the credential.

[0040] The sensitive data ontology and data transfer license are segmented and interleaved. The purpose of segmented interleaving encoding is to deeply integrate the license information with the data ontology, preventing attackers from separating the license and data through simple data interception. The specific encoding method is as follows: the sensitive data ontology is divided into several data segments of a fixed size (1024 bytes in this embodiment), and the data transfer license is divided into several license segments of the same size (padding is done if the last segment is less than 1024 bytes); the segments are then concatenated in an alternating order of "data segment-license segment-data segment-license segment" to form the interleaved encoding result.

[0041] The segment size is set to 1024 bytes, which aligns with the block size of mainstream storage systems, ensuring encoding and decoding efficiency. Simultaneously, the smaller segment granularity allows for a more even distribution of voucher information within the data, enhancing the encoding's resistance to attacks. The segment size ranges from 512 bytes to 4096 bytes and can be adjusted according to data size and performance requirements.

[0042] An encapsulation header is appended to the front of the interleaving encoding result to form a controlled data encapsulation body. The encapsulation header is a fixed 32-byte data block, containing: a magic number (4 bytes, fixed value 0x53444550, i.e., ASCII code "SDEP" indicating Secure Data Encapsulation Protocol), encapsulation version number (2 bytes), original data length (8 bytes), credential length (4 bytes), segment size (4 bytes), number of data segments (4 bytes), and reserved fields (6 bytes).

[0043] An integrity check code is appended to the end of the controlled data encapsulation. This integrity check code is obtained by hashing the encapsulation header and the interleaving encoding result. In this embodiment, the SHA-256 algorithm is used to calculate the integrity check code, generating a 32-byte check value. The integrity check code is appended to the end of the controlled data encapsulation, enabling the receiving node to verify the integrity of the data before parsing it.

[0044] It should be noted that the integrity check code is designed to effectively detect any tampering or damage to data during transmission, ensuring that the controlled data package remains in its original state when it reaches intermediate nodes.

[0045] Step S3: At each intermediate flow node through which the controlled data encapsulation passes, extract the real-time trust assessment value of the current node and compare it with the trust access threshold in the data flow license certificate. Based on the comparison result, perform release, block or downgrade actions and generate node flow evidence records.

[0046] Traditional data loss prevention systems independently perform data detection based on static rules at preset boundary nodes, failing to synchronously verify the real-time trust status of the currently accessing entity against data security protection level requirements. This step implements real-time trust assessment at each intermediate node along the sensitive data flow path, dynamically comparing the node's trust status with the data's security requirements, and taking appropriate action based on the comparison results, thereby achieving continuous trust verification during data flow.

[0047] Preferably, in some possible implementations of the embodiments of the present invention, step S3 specifically includes: The controlled data encapsulation reception and integrity verification process begins when the controlled data encapsulation reaches an intermediate transfer node. The node first verifies the integrity checksum. It extracts a 32-byte integrity checksum from the end of the controlled data encapsulation, recalculates the SHA-256 hash value based on the encapsulation header and interleaving encoding result, and compares the calculated result byte-by-byte with the extracted integrity checksum. If verification fails (i.e., the calculated result does not match the extracted checksum), it indicates that the controlled data encapsulation has been tampered with or damaged during transmission. The node then discards the controlled data encapsulation and records an anomaly. The anomaly record includes: event type (integrity check failure), event timestamp, node identifier, data source node address, and specific information about the verification failure (the first 8 bytes of the expected checksum and the actual calculated value). After successful verification, the encapsulation header is parsed to obtain parameters such as the original data length, voucher length, segment size, and number of data segments, providing a basis for subsequent voucher extraction and data separation.

[0048] The data transfer license is extracted from the interleaved encoding result. Based on the segment size and number of data segments recorded in the encapsulation header, deinterleaving is performed in the reverse order of encoding to reassemble the license segments into a complete data transfer license. The signature validity of the data transfer license is verified using the data owner's public key. The data owner's public key is obtained from the enterprise PKI certificate server and indexed using the data owner identifier in the credential header.

[0049] The signature verification process is as follows: The digital signature field is extracted from the data transfer license credential; the SHA-256 hash value is calculated for all other content in the credential except the signature; and the signature is verified using the data owner's public key and the ECDSA verification algorithm. If verification fails, it indicates that the credential has been tampered with or illegally forged, and the node will execute blocking measures and trigger a high-priority security alarm.

[0050] The trust access threshold is extracted from the data transfer license certificate, specifically the minimum trust access threshold subfield in the security classification field. Simultaneously, device health indicators and network security indicators of the current intermediate transfer nodes are collected, and a logical AND operation is performed on these two indicators to obtain a real-time trust assessment value.

[0051] The equipment health index reflects the safety status of node equipment, with a value range of 0 to 255, and is calculated by weighting the following factors:

[0052] Operating system patch update status (weight 0.25): 255 points for the latest patch installed, 0 points for any missing high-risk patches; Security software running status (weight 0.25): 255 points for running normally, 0 points for abnormal or not running; Recent security event records (weight 0.20): 255 points for no security events in the past 30 days, 25 points deducted for each security event; System resource usage status (weight 0.15): 255 points for both CPU and memory usage below 80%, deducted proportionally for usage exceeding 80%; Peripheral device access status (weight 0.15): 255 points for no unauthorized peripheral devices connected, 0 points for any unauthorized peripheral devices connected.

[0053] Taking patch update status as an example, its numerical mapping rules are as follows: Patch installation status Rating value All emergency and high-risk patches have been installed, and the installation rate of medium-risk patches is ≥90%. 255 All emergency and high-risk patches have been installed, but the installation rate for medium-risk patches is between 70% and 89%. 220 All emergency and high-risk patches have been installed, but the installation rate for medium-risk patches is less than 70%. 180 There are 1 to 2 missing high-risk patches, and no missing emergency patches. 120 There are 3 to 5 missing high-risk patches and no missing emergency patches. 60 There are 6 or more missing high-risk patches, or any number of missing emergency patches. 0

[0054] Network security indicators reflect the security status of the network environment in which a node is located. Values ​​range from 0 to 255 and are calculated by weighting the following factors: Network partition compliance (weight 0.30): A node located in a legally authorized network partition receives a full score of 255, while a node located in an unauthorized partition receives 0 points; Network traffic anomaly level (weight 0.30): No abnormal traffic receives a full score of 255, and points are deducted based on the severity of detected abnormal traffic; Network protection device status (weight 0.20): Upstream firewalls and intrusion detection systems operating normally receive a full score of 255, and points are deducted for anomalies; Network connection source trustworthiness (weight 0.20): Data originating from nodes in the trusted node list receives a full score of 255, otherwise, the score is calculated proportionally based on the node's trust level. Taking network partition compliance as an example, its numerical mapping rules are as follows: Network partition type Authorization status Rating value Core security zone Has a long-term valid license for this partition (remaining validity > 7 days). 255 Core security zone The license for this partition has a remaining validity period of between 24 hours and 7 days. 230 Core security zone The partition has an authorization but the remaining validity period is less than 24 hours. 180 General office area Has valid authorization for this partition 220 Research and development testing area Has valid authorization for this partition 200 Visitor area Obtain a valid temporary authorization through the temporary authorization process. 140 Visitor area The temporary license has expired but is still in the network. 60 Arbitrary partition Node IP address does not match authorization record 30 Unauthorized partition The node is located in an unauthorized network partition. 0

[0055] The real-time trust assessment value is obtained by performing a logical AND operation between device health indicators and network environment security indicators. This real-time trust assessment value is then compared with the trust access threshold, and appropriate actions are taken based on the comparison result, as follows: Scenario 1: Sufficient Trust, Allow Access; When the real-time trust assessment value is greater than or equal to the trust admission threshold, the node is marked as having sufficient trust and allows access. The node forwards the controlled data encapsulation to downstream nodes without modifying the data content.

[0056] Scenario 2: Mild Trust Insufficiency, Perform Downgrade and Anonymization Before Release; When the real-time trust assessment value is less than the trust admission threshold and the difference is within the preset tolerance range, perform data downgrade and anonymization before release. The preset tolerance range is 15% of the trust admission threshold, that is, when (the absolute value of the trust admission threshold minus the real-time trust assessment value) ≤ 15% of the trust admission threshold, downgrade and anonymization are performed. Setting the tolerance range to 15% provides a moderate margin of error for temporary trust fluctuations that may occur in normal business processes while ensuring data security. The tolerance range is 10% to 20%, and implementers can adjust it according to business continuity requirements.

[0057] Scenario 3: Insufficient Trust, Blocking Execution; When the difference exceeds the preset tolerance range, blocking is executed and an alarm is triggered. The node discards the controlled data encapsulation and sends a high-priority alarm notification to the security management center. The alarm content includes: alarm type (insufficient trust blocking), controlled data identifier, trust access threshold, real-time trust assessment value, difference, node identifier, and alarm timestamp.

[0058] The specific methods for data degradation and desensitization include: When the decision is to downgrade and desensitize, the sensitive data ontology is extracted from the controlled data encapsulation. Based on the segment size and number of data segments recorded in the encapsulation header, a deinterleaving operation is performed on the interleaving encoding result, reassembling the data segments into a complete sensitive data ontology. The data type of the sensitive data ontology is determined based on the security classification field in the data transfer license. The data type determination is based on the data protection level code in the security classification field and the type label attached by the data owner when classifying the data. Common data types include: identity information, financial data, technical documents, business contracts, etc. Desensitization rules matching this data type are retrieved from the desensitization rule base. The desensitization rule base is a pre-configured set of rules for the enterprise, stored in the local rule cache of intermediate nodes, and periodically updated from the security management center.

[0059] The typical desensitization rules for some data types are shown in the table below: Data types Sensitive field examples Desensitization methods Example after desensitization Identity information ID number Middle bit mask 110101********1234 Financial data bank account Tail mask 6222************ Technical Documentation Core algorithm parameters Range truncation [Desensitized] Business Contract Contract amount Magnitude blurring ¥***10,000 yuan

[0060] According to the de-identification rules, sensitive fields in the sensitive data ontology are masked or truncated to generate a de-identified data ontology. The masking operation replaces some characters in the sensitive fields with mask characters, preserving some field characteristics to meet the basic processing needs of downstream businesses. The truncation operation completely removes the content of the sensitive fields and replaces it with the de-identification marker string "[De-identified]".

[0061] The de-identification process records a de-identification log, which includes: the SHA-256 hash value of the original field (excluding the original content), the de-identification method, the de-identification execution time, and the execution node identifier, for subsequent de-identification auditing and data recovery authorization.

[0062] The security classification field in the updated data transfer license is changed to the downgraded protection level code. The downgrade rules are as follows: Top Secret to Confidential, Confidential to Internal, Internal to Public, and Public remains unchanged. Simultaneously, the minimum trust threshold is adjusted to the typical value corresponding to the downgraded protection level. A desensitization flag and a desensitization execution node identifier are added to the license. The desensitization flag is a fixed byte sequence 0xDE (representing Desensitized), and the desensitization execution node identifier records the intermediate node number that performed the desensitization operation. The updated data transfer license is re-signed using the current node's temporary signing key, while retaining the original data owner's signature as a historical credential to ensure traceability of license changes. The desensitized data body and the updated data transfer license are re-segmented and interleaved to generate a new controlled data package. The original data length field in the package header is updated (the data length may change after desensitization), and the integrity check code is recalculated before continuing the transfer to downstream nodes.

[0063] The downgrade and de-identification mechanism maintains business process continuity while ensuring data security, avoiding business interruptions caused by temporary trust fluctuations. Meanwhile, complete de-identification logs and credential change records ensure the auditability of de-identification actions, supporting subsequent compliance checks and incident investigations.

[0064] After performing release, downgrade, or blocking actions, a node flow record is generated to support subsequent leak tracing. The node identifier of the current intermediate flow node and the precise timestamp of the action are obtained; the node identifier is a unique number for that node in the enterprise intranet asset management system. A fragment of the dynamic session token in the data flow license is extracted as a session association identifier. In this embodiment, the first 16 bytes of the dynamic session token are extracted as the session association identifier, which is sufficient to uniquely identify a data flow session within the enterprise intranet. The session association identifier and the node identifier are concatenated to form a node session binding code, with the concatenation format "session association identifier@node identifier".

[0065] Record the handling type field, where the handling type field can take three status codes: release (status code 0x01), downgrade and desensitize (status code 0x02), and block (status code 0x03).

[0066] The node session binding code, disposal type field, precise timestamp, and real-time trust assessment value are encapsulated into a structured data block. The structured data block uses a fixed format with a total length of 128 bytes. The fields are distributed as follows: node session binding code (64 bytes), disposal type field (1 byte), precise timestamp (27 bytes), real-time trust assessment value (2 bytes), and reserved extension field (34 bytes).

[0067] The structured data block is hashed using SHA-256 to obtain a proof digest. The structured data block and the proof digest are then combined to form a node flow proof record.

[0068] The records of evidence transfer between nodes are temporarily stored in the local evidence cache of the current intermediate transfer node, awaiting subsequent evidence chain construction and distributed evidence storage synchronization. The local evidence cache adopts a circular buffer design with a capacity of 10,000 records. When the capacity is exceeded, the oldest record is overwritten, and an emergency synchronization is triggered to push the overwritten record to the distributed evidence storage node cluster.

[0069] Step S4: Link the flow and storage records of each node in chronological order, form a flow trajectory evidence chain by forward hashing, and synchronously write the flow trajectory evidence chain into the distributed storage node cluster.

[0070] In the process of evidence collection and tracing after a data breach, traditional architectures lack a cryptographically verifiable link between the static session logs generated during the authentication process and the discrete operation records generated during data flow. This prevents the tracing system from reconstructing the complete data flow path. This step uses forward hashing anchoring technology to construct a cryptographically linked chain of evidence from the flow records of each node, and ensures the immutability and high availability of the evidence chain through a distributed storage mechanism.

[0071] After the controlled data encapsulation completes one node transfer, the newly generated node transfer evidence record is read from the local evidence cache of the current node. If the current node is the first intermediate node in the data transfer, the forward anchor value is set to all zeros; if the current node is not the first intermediate node, the evidence summary of the previous node's transfer evidence record is obtained as the forward anchor value. The forward anchor value is obtained by receiving the attached previous evidence summary from the upstream node when forwarding the controlled data encapsulation, or by extracting it from the temporary extended field of the controlled data encapsulation. The forward anchor value is written to the associated field of the current node's transfer evidence record. The associated field is located in the reserved extended field of the structured data block and occupies 32 bytes.

[0072] For the current node's transferred evidence record containing the forward anchor value, recalculate the SHA-256 hash and update the evidence digest field so that the current evidence digest contains the cryptographic association of the previous evidence digest. The updated evidence digest calculation formula is: In the formula, This is the updated evidence summary; For structured data blocks; This is the forward anchoring value.

[0073] The updated node transfer records are appended to the end of the transfer trajectory evidence chain in chronological order, forming a unidirectional chain structure. This unidirectional chain structure ensures that any tampering with historical records will cause the digest verification of all subsequent records to fail, thus guaranteeing the integrity and immutability of the evidence chain.

[0074] The distributed evidence storage synchronization process is triggered when the length of the evidence chain of the data transfer trajectory reaches a preset synchronization threshold or when the controlled data package reaches the final target node. The preset synchronization threshold is set to 5 records, meaning that a batch synchronization is performed every time 5 nodes of data transfer evidence records are accumulated or when the data transfer ends.

[0075] Setting the synchronization threshold to 5 records strikes a balance between ensuring timely evidence storage and reducing network synchronization overhead. The range of the synchronization threshold is 3 to 10 records, which implementers can adjust according to network bandwidth and evidence storage timeliness requirements.

[0076] The complete evidence chain of the circulation trajectory is pushed in batches to the distributed evidence storage node cluster. The distributed evidence storage node cluster consists of multiple independent evidence storage nodes deployed in the enterprise intranet. In this embodiment, the number of evidence storage nodes is set to 5, distributed in different physical locations and network partitions to ensure that a single point of failure does not affect the overall availability of the evidence storage system.

[0077] After receiving the evidence chain of the transfer trajectory, the distributed evidence storage node cluster first verifies the consistency between the forward anchor value and the evidence digest, starting from the first node's transfer evidence storage record. The verification method is as follows: for each evidence storage record, the forward anchor value in its associated field is extracted, and it is verified whether it is equal to the evidence digest of the preceding evidence storage record (for the first record, the forward anchor value is verified to be all zeros); at the same time, the hash of the structured data block and associated field is recalculated, and it is verified whether it is equal to the stored evidence digest.

[0078] If any inconsistency is found during the verification process, the evidence storage node refuses to store the evidence chain and sends an evidence chain integrity anomaly alert to the security management center. A globally unique evidence chain identifier is assigned to each verified flow trajectory evidence chain. The evidence chain identifier is generated using UUID v4 format.

[0079] An index mapping is established between the evidence chain identifier and the session association identifier of the first record in the flow trajectory evidence chain. The index mapping is stored in the cluster index service node to support fast retrieval during subsequent source tracing queries.

[0080] The evidence chain of the transfer trajectory is fragmented and replicated to multiple independent evidence storage nodes in a distributed evidence storage node cluster. Each independent evidence storage node stores a complete copy of the evidence chain of the transfer trajectory. This embodiment adopts an N / 2+1 redundancy strategy, where N represents the number of nodes, that is, at least 3 of the 5 evidence storage nodes store the same copy of the evidence chain, ensuring that the evidence chain data can still be completely recovered even if any 2 nodes fail.

[0081] Step S5: When a data breach is detected, extract the evidence chain of the flow trajectory from the distributed evidence storage node cluster, reverse traverse the flow evidence storage records of each node to locate the trust break node, and output a source tracing evidence report containing the complete flow path and break location.

[0082] Traditional architectures, when conducting forensic investigations and tracing after a data breach, lack a cryptographically verifiable link throughout the entire chain. This makes it impossible for the tracing system to reconstruct the complete data flow path from the source storage node to the breach point, and also makes it difficult to accurately pinpoint the specific node and timing of the break in the trust chain. This step, by reversing the flow trajectory evidence chain and combining cryptographic verification with handling status analysis, precisely locates the node of trust breakage and generates a tracing report with sufficient technical credibility and complete evidentiary support.

[0083] Upon receiving a data breach alarm, the dynamic session token fragment involved in the alarm information is extracted as the source retrieval keyword. The data breach alarm is generated by the enterprise data loss prevention system when it detects the outflow of sensitive data. The alarm information includes a dynamic session token fragment extracted from the controlled data encapsulation that caused the alarm. The cluster index service node is queried using the source retrieval keyword. Through the index mapping between session association identifiers and evidence chain identifiers, the corresponding evidence chain identifier is obtained. If the query does not find any index record, it indicates that the data flow process did not pass through a monitored intermediate node, or that the evidence storage system has missing data. The source retrieval process records the "evidence chain missing" status and notifies security operations personnel to conduct a manual investigation. Based on the evidence chain identifier, the complete flow trajectory evidence chain is pulled from any independent evidence storage node. The distributed evidence storage multi-copy mechanism ensures that even if some evidence storage nodes are unavailable, complete evidence chain data can still be obtained from other nodes.

[0084] Starting from the tail node of the evidence chain, the process is reversed and the cryptographic consistency between the evidence digest and the forward anchor value is verified one by one.

[0085] The reverse traversal verification process is as follows: For each evidence record, its evidence digest and structured data block are extracted, the hash value is recalculated and compared with the stored evidence digest; simultaneously, it is verified whether the evidence digest of this record is equal to the forward anchor value in the associated field of subsequent records. Breakpoints where verification fails are marked. If the evidence digest verification of a record fails, or its evidence digest is inconsistent with the forward anchor value of subsequent records, this position is marked as an "evidence chain breakpoint," indicating that the evidence association between this node and subsequent nodes has been broken, and there may be evidence tampering.

[0086] Extract the "Disposal Type" field from the data flow records of each node, and identify the node where the first instance of blocking, interception, or downgrading / de-identification occurs as the starting node of trust degradation. The identification logic for the starting node of trust degradation is as follows: traverse the data flow records of each node in chronological order. When the value of the "Disposal Type" field is not equal to "allow" for the first time (status code 0x01), that node is the starting node of trust degradation. The starting node of trust degradation indicates that from this node, the trust status on the data flow path begins to deteriorate.

[0087] The trust fracture node is determined by combining the fracture location with the trust degradation initiation node. The determination rules are as follows: If a break in the chain of evidence exists and this break is earlier than or equal to the trust degradation initiation node, then the break is a trust break node. If no break in the chain of evidence exists, then the trust degradation initiation node is a trust break node. If a break in the chain of evidence exists and this break is later than the trust degradation initiation node, then both are marked as key analysis nodes, with the trust degradation initiation node being the trust break node and the break being an evidence anomaly node.

[0088] Summarize all node identifiers, precise timestamps, disposal types, and trust breakpoint information in the evidence chain of the circulation trajectory, and organize them in chronological order to form a source tracing evidence report.

[0089] The structure of the source tracing evidence report includes: Report header: report generation time, report unique identifier, associated alarm number; Session summary: dynamic session token digest, data security level identifier, source node address, target node address; Complete flow path: a list of node information arranged chronologically, with each record containing node identifier, flow timestamp, handling type, and real-time trust assessment value; Trust analysis conclusion: trust break node identifier, break time, and the difference between the trust assessment value and the threshold at the time of break; Evidence integrity statement: evidence chain verification results, existence of a break point, and break location.

[0090] The source tracing evidence report is digitally signed before being output. The signature uses the service private key of the source tracing system and the ECDSA algorithm to ensure the authenticity and non-repudiation of the report. The signed source tracing evidence report is then sent to the security management center and relevant responsible persons through a secure channel.

[0091] Dynamic Session Token Update: A token validity countdown begins after the dynamic session token is generated. The endpoint security agent continuously monitors the remaining validity period of the dynamic session token. When the remaining countdown time is lower than a preset update trigger threshold, the token update process is initiated. The preset update trigger threshold is set to 20% of the total validity period; that is, if the token validity period is 30 minutes, the update is triggered when 6 minutes remain. Setting the update trigger threshold to 20% allows sufficient processing time for the token update process while avoiding premature updates that waste computational resources. The update trigger threshold can range from 10% to 30%, and implementers can adjust it based on network latency and update processing time.

[0092] The system re-collects the terminal device's operating environment status and generates a new operating environment status descriptor. This new descriptor is then compared with the one used when the device trust anchor was generated. Key security indicators for this comparison include: the running status of the protection software, the virus database update time, and the most recent security scan results. If any of these indicators changes, it is marked as a change in a critical security indicator; if only non-critical indicators such as the operating system kernel version change, it is marked as a general change. If the comparison result shows a change in a critical security indicator, the user is required to re-enter their credentials for re-authentication. The re-authentication process is the same as the initial authentication process: the user enters their user identifier and authentication key, and the system verifies the validity of the credentials. After successful re-authentication, the device trust anchor is updated with the new operating environment status descriptor. The new descriptor, along with the original device hardware fingerprint and the new credential digest value, are used as input to generate a new device trust anchor through a key derivation function. If the comparison result shows only a general change or no change, re-authentication is not required, and the new operating environment status descriptor is used directly to update the device trust anchor.

[0093] A new dynamic session token is generated based on the updated device trust anchor and the new timestamp. The new dynamic session token inherits the session association identifier prefix of the original dynamic session token and adds an update sequence number suffix. The update sequence number starts from 01 and increments, with the format "-Uxx". For example, if the original session association identifier is "SA1698234567890", after the first update it becomes "SA1698234567890-U01". The design of inheriting the session association identifier prefix ensures that multiple tokens generated by the same device in a complete business session can be associated and traced. At the same time, the update sequence number suffix makes the temporal relationship between the tokens clear. The original dynamic session token is marked as invalid and the invalidation time is recorded. The invalidation record is stored in the token history table in the terminal's local secure storage area, including: the original token digest, the invalidation time, the invalidation reason (expiration or active renewal), and the successor token digest. The new dynamic session token is written to the terminal's secure storage area, overwriting the original storage location, and at the same time, the token replacement evidence record is pushed to the distributed evidence storage node cluster. The token replacement record includes: original token digest, new token digest, replacement time, and replacement reason, which is used to support traceability analysis across token cycles.

[0094] This solution enables continuous trust verification and the construction of a complete evidence chain during the flow of sensitive data within an enterprise intranet. Device trust anchors cryptographically bind hardware characteristics, operating environment, and user identity; the timeliness design and update mechanism of dynamic session tokens ensure the real-time validity of the trust state. Controlled data encapsulation deeply integrates authorization credentials with sensitive data, ensuring that data always carries verifiable access control information during flow. Real-time trust assessment and tiered handling mechanisms at each intermediate node enable timely protective measures when the trust state deteriorates. Forward hashing and distributed storage of the flow trajectory evidence chain ensure the integrity and immutability of the evidence, providing reliable technical support for accurate tracing of data breaches.

[0095] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. 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 preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture, characterized in that: include: Step S1: Obtain the hardware feature information, operating environment status and user identity credentials of the terminal device, generate the device trust anchor point through multi-dimensional feature fusion, and derive the dynamic session token based on the device trust anchor point. Step S2: Obtain the security level identifier and target node information of the sensitive data, generate a data transfer permission certificate by combining the dynamic session token, and cryptographically bind the data transfer permission certificate to the sensitive data to form a controlled data encapsulation. Step S3: At each intermediate flow node through which the controlled data encapsulation passes, extract the real-time trust assessment value of the current node and compare it with the trust access threshold in the data flow license certificate. Based on the comparison result, perform release, block or downgrade actions and generate node flow evidence records. Step S4: Link the flow and storage records of each node in chronological order, form a flow trajectory evidence chain by forward hashing, and synchronously write the flow trajectory evidence chain into the distributed storage node cluster. Step S5: When a data breach is detected, extract the evidence chain of the flow trajectory from the distributed evidence storage node cluster, reverse traverse the flow evidence storage records of each node to locate the trust break node, and output a source tracing evidence report containing the complete flow path and break location.

2. The enterprise intranet data leakage prevention and tracing method based on zero-trust architecture according to claim 1, characterized in that, Step S1 includes: The system collects the motherboard serial number, chip identifier, and network card physical address of the terminal device as hardware feature information, and concatenates the hardware feature information in a preset order to form a device hardware fingerprint string; it obtains the operating system kernel version and protection software running status of the terminal device as the operating environment status; it receives user identity credentials input by the user, including user identifier and authentication key, and performs a one-way hash operation on the authentication key to obtain a credential digest value; it uses the device hardware fingerprint string, operating environment status, and credential digest value as input to generate a fixed-length device trust anchor point through a key derivation function; it generates a dynamic session token by performing an XOR obfuscation operation on the device trust anchor point and the current timestamp, and then appending a validity period field.

3. The enterprise intranet data leakage prevention and tracing method based on zero-trust architecture according to claim 2, characterized in that, Step S2 includes: The process involves: reading the metadata header of sensitive data and extracting the security classification identifier field, which contains the data protection level code and the minimum trust threshold; obtaining the target node address and expected transfer path sequence specified in the data transfer request, and encoding the target node address and expected transfer path sequence into a path description block; concatenating the dynamic session token, the security classification identifier field, and the path description block, and signing it with the data owner's private key using an asymmetric encryption algorithm to generate a data transfer permission certificate; segmenting and interleaving the sensitive data body and the data transfer permission certificate, and appending an encapsulation identifier header to the front of the interleaving encoding result to form a controlled data encapsulation; and appending an integrity check code to the end of the controlled data encapsulation, where the integrity check code is obtained by hashing the encapsulation identifier header and the interleaving encoding result.

4. The method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture as described in claim 3, characterized in that, Extract the real-time trust assessment value of the current node and compare it with the trust access threshold in the data flow license credential. Based on the comparison result, perform actions such as allowing, blocking, or downgrading, including: When the controlled data package arrives at the intermediate transfer node, the integrity check code is first verified. If the verification fails, the controlled data package is discarded and the abnormal event is recorded. After successful verification, the package header is parsed, and the data transfer license is extracted from the interleaved encoding result. The signature validity of the data transfer license is verified using the data owner's public key. The trust access threshold is extracted from the data transfer license. At the same time, the device health index and network environment security index of the current intermediate transfer node are collected. The two indexes are logically ANDed to obtain the real-time trust assessment value. The real-time trust assessment value is compared with the trust access threshold. When the real-time trust assessment value is greater than or equal to the trust access threshold, it is marked as trust sufficient and the data is allowed to pass. When the real-time trust assessment value is less than the trust access threshold and the difference is within the preset tolerance range, the data is downgraded and desensitized before being allowed to pass. When the difference exceeds the preset tolerance range, the data is blocked and an alarm is triggered.

5. The method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture as described in claim 4, characterized in that, The methods for generating node transfer and evidence storage records include: After performing release, downgrade, or blocking actions, obtain the node identifier of the current intermediate transfer node and the precise timestamp of the action time; extract the dynamic session token fragment from the data transfer license credential as a session association identifier, and concatenate the session association identifier with the node identifier to form a node session binding code; record the action type field, where the values ​​of the action type field include three status codes: release, downgrade and desensitization, and blocking; encapsulate the node session binding code, action type field, precise timestamp, and real-time trust assessment value into a structured data block, perform a hash operation on the structured data block to obtain a storage digest; combine the structured data block and the storage digest to form a node transfer storage record, and temporarily store it in the local storage cache area of ​​the current intermediate transfer node.

6. The method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture as described in claim 5, characterized in that, The ways to form a chain of evidence for the circulation trajectory include: After the controlled data encapsulation completes one node transfer, the newly generated node transfer evidence record is read from the local evidence cache of the current node; the evidence digest of the previous node transfer evidence record is obtained as the forward anchor value, and the forward anchor value is written into the association field of the current node transfer evidence record; the hash of the current node transfer evidence record containing the forward anchor value is recalculated, and the evidence digest field is updated so that the current evidence digest contains the cryptographic association of the previous evidence digest; the updated node transfer evidence record is appended to the end of the transfer trajectory evidence chain in chronological order to form a unidirectional chain structure.

7. The method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture as described in claim 6, characterized in that, The methods for synchronously writing the evidence chain of the transfer trajectory into the distributed evidence storage node cluster include: When the length of the flow trajectory evidence chain reaches the preset synchronization threshold or the controlled data encapsulation reaches the final target node, the complete flow trajectory evidence chain is pushed to the distributed evidence storage node cluster in batches. After receiving the flow trajectory evidence chain, the distributed evidence storage node cluster first verifies the consistency between the forward anchor value and the evidence digest one by one, starting from the flow storage record of the first node in the chain. A globally unique evidence chain identifier is assigned to the verified flow trajectory evidence chain, and an index mapping is established between the evidence chain identifier and the session association identifier of the first record in the flow trajectory evidence chain. The flow trajectory evidence chain is fragmented and copied to multiple independent evidence storage nodes in the distributed evidence storage node cluster, and each independent evidence storage node stores a complete copy of the flow trajectory evidence chain.

8. The method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture as described in claim 7, characterized in that, Obtaining the source tracing evidence report includes: Upon receiving a leak event alarm, the system extracts the dynamic session token fragments involved in the alarm information as keywords for tracing the source. It then queries the cluster index service node using these keywords to obtain the corresponding evidence chain identifier. Based on the evidence chain identifier, it retrieves the complete flow trajectory evidence chain from any independent evidence storage node. Starting from the tail node of the flow trajectory evidence chain, it traverses backwards, verifying the cryptographic consistency between the evidence digest and the forward anchor value, and marking the break points where verification failed. It extracts the disposal type field from the flow evidence storage records of each node, identifies the node where the first instance of blocking, interception, or downgrading / desensitization occurs as the trust degradation starting node, and determines the trust break point based on the break point. Finally, it summarizes all node identifiers, precise timestamps, disposal types, and trust break point information in the flow trajectory evidence chain, organizes them chronologically to form a tracing evidence report, and outputs the tracing evidence report after digitally signing it.

9. The method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture as described in claim 8, characterized in that, The methods for updating dynamic session tokens include: After the dynamic session token is generated, a token validity countdown is started. When the remaining countdown time is lower than the preset update trigger threshold, the token update process is initiated. The operating environment status of the terminal device is re-collected, and a new operating environment status descriptor is generated. The new operating environment status descriptor is compared with the operating environment status descriptor used when the device trust anchor was generated. If the comparison result shows that key security indicators have changed, the user is required to re-enter identity credentials for re-authentication. After successful re-authentication, the device trust anchor is updated with the new operating environment status descriptor. The dynamic session token is regenerated based on the updated device trust anchor and the new timestamp. The new dynamic session token inherits the session association identifier prefix of the original dynamic session token and adds an update sequence number suffix. The original dynamic session token is marked as invalid and the invalidation time is recorded. The new dynamic session token is written to the terminal's secure storage area, overwriting the original storage location. At the same time, the token replacement evidence record is pushed to the distributed evidence storage node cluster.

10. The method for preventing data leakage and tracing within an enterprise intranet based on a zero-trust architecture as described in claim 9, characterized in that, Data downgrading and desensitization methods include: When the decision to downgrade and de-identify the data is made, the sensitive data ontology is extracted from the controlled data encapsulation. The data type of the sensitive data ontology is determined based on the security classification field in the data transfer license certificate, and a de-identification rule matching the data type is retrieved from the de-identification rule base. Sensitive fields in the sensitive data ontology are masked or truncated according to the de-identification rules to generate the de-identified data ontology. The security classification field in the data transfer license certificate is updated to the downgraded protection level code, and a de-identification processing mark and a de-identification execution node identifier are added to the certificate. The de-identified data ontology and the updated data transfer license certificate are re-segmented and interleaved to generate a new controlled data encapsulation. After recalculating the integrity check code, the data continues to flow to downstream nodes.