A Blockchain-Based Method for Cross-Domain Trusted Data Sharing in the Internet of Things

Abstract

This invention discloses a blockchain-based method for cross-domain trusted data sharing in the Internet of Things (IoT), comprising the following steps: registering and verifying the identities of entities and devices in the providing and using domains; writing only data fingerprints, timestamps, unified identifiers, and directory indexes into the consortium blockchain; deploying an authorization contract on the blockchain and enabling an improved GraphRAG policy channel to retrieve the directory index and align the clause subgraph; submitting a request containing the unified identifier and directory index using the domain, which is matched by the adjudication channel and automatically adjudicated by the contract, generating a one-time token or recording the rejection reason hash; performing off-chain encrypted transmission based on the token, verifying integrity and freshness according to the on-chain fingerprint and timestamp, and uploading the results to the blockchain; completing settlement and reconciliation based on on-chain metering, and arbitrating in case of disputes. This invention achieves highly trusted sharing with no data leaving the domain, automatic compliance, end-to-end verifiable and auditable settlement.

Description

Technical Field

[0001] This invention relates to the field of blockchain and Internet of Things (IoT) data governance technology, and in particular to a blockchain-based method for cross-domain trusted data sharing in the Internet of Things. Background Technology

[0002] With the widespread adoption of the Internet of Things (IoT) in industry, cities, and cross-institutional collaboration, cross-domain sharing of device data and operational logs to support operations, maintenance, risk control, and compliance has become a necessity. Existing solutions mostly adopt an architecture of "on-chain registration and off-chain transmission": a centralized data platform or a simplified consortium blockchain is used for directory indexing, and authorization and location are completed by combining access control lists, role permissions, and keyword searches. Only summaries or fingerprints are stored on the blockchain, and access is mostly based on fixed thresholds or manual review. Settlement and reconciliation are mainly completed offline by comparing tables.

[0003] However, in practice, the following problems are common: cross-domain identity and device binding are fragile, and the updates of identifiers and public key states are not synchronized; data standardization standards are inconsistent, the fingerprint generation process is not verifiable, and versions are difficult to replay; policy expressions are crude and lack interpretable constraints at the semantic and relational levels, making it easy for standards to conflict and policies to drift; authorization thresholds do not consider adaptation, and the misjudgment of critical samples is high; off-chain transmission lacks strong consistency verification and freshness proof with on-chain digests, making it easy to be replayed; the binding between tokens and policy versions and scope of effectiveness is not strong, and the audit context is broken; measurement and settlement standards are scattered, evidence is fragmented, and dispute arbitration is difficult to form a closed loop traceability from authorization to settlement.

[0004] Therefore, how to provide a blockchain-based method for cross-domain trusted data sharing in the Internet of Things is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a blockchain-based method for cross-domain trusted data sharing in the Internet of Things (IoT). This invention achieves secure and controllable cross-domain data sharing through decentralized identity registration, on-chain data fingerprinting, generation of clause subgraphs and adaptive thresholds in policy channels, automatic authorization and issuance of one-time tokens in adjudication channels, off-chain encrypted transmission and on-chain integrity and freshness verification, and a closed loop of measurement settlement and arbitration. It has the advantages of high trustworthiness, strong traceability, anti-replay, anti-tampering, and audit-friendly features.

[0006] A blockchain-based method for cross-domain trusted data sharing in the Internet of Things (IoT) according to an embodiment of the present invention includes the following steps:

[0007] To provide domain registration and identity verification for entities and devices using the domain, decentralized identifiers and public key status are written into the consortium blockchain to generate a unified identifier and record the block height and block timestamp.

[0008] Provides domain preservation of the original text, only writing data fingerprints, timestamps, unified identifiers, and directory indexes to the consortium blockchain;

[0009] Deploy authorization contracts on the consortium blockchain, build an improved GraphRAG model and enable policy channels, retrieve and align clause subgraphs in the catalog index, and output clause subgraph fingerprints, policy version numbers, adjudication thresholds and constraint sets;

[0010] The domain submits a request containing a unified identifier and directory index, calls the adjudication channel matching clause subgraph of the improved GraphRAG model, and authorizes the contract to automatically adjudicate based on the constraint set, policy version number and adjudication threshold, generating a one-time token or writing the rejection reason hash.

[0011] Provides a domain with a one-time token for off-chain encrypted transmission. The domain is used to verify the integrity and freshness of the data based on the data fingerprint and timestamp on the consortium blockchain, and the verification result is written to the consortium blockchain.

[0012] Based on the measurement on the consortium blockchain, the settlement and reconciliation of the calls corresponding to the one-time tokens are completed, and the settlement summary and exception record are written into the consortium blockchain. In case of disputes, arbitration and tracing are carried out.

[0013] Optionally, the step of registering and verifying the identities of entities and devices providing and using the domain, writing decentralized identifiers and public key states into the consortium blockchain, generating a unified identifier, and recording block height and block timestamps specifically includes:

[0014] Develop and publish rules for registering and verifying the identities of entities and devices that provide and use domains, and open a registration portal on the consortium blockchain;

[0015] Collect registration information of entities and devices that provide and use the domain, and generate the basic data required for decentralized identification;

[0016] Complete registration and verify identity according to the rules, generate decentralized identifiers for the subject and device, and generate corresponding public key states;

[0017] Write the decentralized identifier and public key status into the consortium blockchain, and record the block height and block timestamp at the time of writing;

[0018] After registration and identity verification are completed, a unified identifier is generated for the subject and the device, and the unified identifier and the decentralized identifier are bound together on the consortium blockchain.

[0019] When the public key status or decentralized identifier of the subject and device changes, it is synchronously updated to the consortium blockchain, and the block height and block timestamp are recorded again.

[0020] Optionally, the provided domain retains the original text, only writing the data fingerprint, timestamp, unified identifier, and directory index into the consortium blockchain, specifically including:

[0021] Provide the domain to retain the original text in an off-chain controlled storage environment, determine the set of original texts that need to be written to the consortium blockchain, the original texts are real data content generated or hosted by the domain and not desensitized or pruned, including sensor sampling sequences, equipment logs, control command records, image and audio / video files, industrial documents and reports and interface return messages and archived files;

[0022] Perform normalization processing on each original text to unify the character set, newline character, time zone, and byte order, and obtain a normalized byte sequence;

[0023] Generate a timestamp for each original text, generate a table of contents index for each original text, and establish a one-to-one binding relationship between the table of contents index and the unified identifier;

[0024] The unified identifier, directory index, normalized byte sequence length and timestamp encoding value, and normalized byte sequence are concatenated in a fixed order to form a byte stream to be fingerprinted;

[0025] A collision-resistant cryptographic hash algorithm is used to calculate the data fingerprint from the byte stream to be fingerprinted.

[0026] Entries are written to the consortium blockchain using data fingerprints, timestamps, unified identifiers, and directory indexes, and the block height and timestamp at the time of writing are recorded.

[0027] When the original text is added, modified, or deleted, normalization processing is re-executed, timestamps and directory indexes are regenerated and maintained in a one-to-one binding relationship with the unified identifier, the byte stream to be fingerprinted is reassembled and the data fingerprint is calculated again, the updated entry is written to the consortium blockchain and the block height and block timestamp are recorded, and the previous target is marked as historical state.

[0028] Optionally, the step of deploying the authorization contract on the consortium blockchain to build the improved GraphRAG model specifically includes:

[0029] Deploy the authorization contract and initialize the registration interface on the consortium blockchain;

[0030] An improved GraphRAG model is constructed, comprising a common base layer, a policy channel, and an adjudication channel. The common base layer includes a data input and indexing layer, a semantic normalization layer, an index retrieval layer, a clause subgraph representation normalization layer, an alignment scoring function family layer, an adjudication threshold, a policy version parameter management layer, and a registration interface layer for the consortium blockchain.

[0031] The data input and indexing layer receives the directory index and its corresponding unified identifier, performs word segmentation, field mapping, case unification and stop character cleanup on the relevant fields of the directory index, establishes searchable index items and inverted indexes, and outputs normalized field sequences and index item identifiers.

[0032] The semantic normalization layer receives a normalized field sequence, a unified identifier, and a directory index, and embeds the unified identifier code and directory index code in a fixed order to generate a standardized semantic vector and a position code sequence.

[0033] The index retrieval layer receives the directory index, standardized semantic vectors and index item identifiers, performs vector similarity retrieval and keyword matching in parallel, and performs weighted sorting based on similarity score, time freshness and source credibility, and outputs a candidate node list, hit order and scoring trajectory identifier.

[0034] The clause subgraph representation specification layer receives the candidate node list and the hit order, extracts nodes and edges according to the type consistency and key adjacency retention rules, records the node sequence, edge sequence and attribute sequence in a fixed order, and generates the representation sequence and representation sequence fingerprint;

[0035] The alignment scoring function family layer receives node sequences, edge sequences, attribute sequences and scoring trajectory identifiers, calculates name consistency score, relation pattern consistency score and adjacency retention score, forms an alignment scoring vector, and extracts the minimum interpretation path as alignment evidence;

[0036] The adjudication threshold and strategy version parameter management layer receives the aligned scoring vector and directory index, and calculates the adaptive value of the adjudication threshold. The adaptive calculation uses the historical pass percentile as the benchmark threshold, superimposed with positive corrections for topic risk score, negative corrections for freshness score, and negative corrections for credibility score. It is then pruned within the minimum and maximum boundaries and quantized by thousandths. At the same time, a strategy version number is assigned and a version lock is set. The historical pass percentile, topic risk score, freshness score, credibility score, and weighted reorganization are written into the strategy version parameter table to generate a threshold calculation summary.

[0037] The registration interface layer of the consortium blockchain receives the policy version number and version lock flag, generates a registration handle, and provides atomic write and read-only reference paths for the authorization contract.

[0038] Optionally, enabling the policy channel involves retrieving and aligning the clause subgraphs in the directory index, and outputting the clause subgraph fingerprint, policy version number, adjudication threshold, and constraint set, specifically including:

[0039] The strategy channel is enabled, and neighborhood expansion and pruning are performed according to the candidate node list and the hit order. Following the rules of type consistency and key adjacency retention, entities with the same name are merged and low-weight redundant edges and isolated nodes are deleted to form an ordered representation of the clause subgraph and obtain the node sequence, edge sequence, attribute sequence and representation sequence.

[0040] Alignment of clause subgraphs is confirmed based on alignment score vector and adjudication threshold. Thresholds are determined for name consistency score, relation pattern consistency score and adjacency retention score. If not satisfied, the search keywords and time window are expanded with the directory index and standardized semantic vector as prompts, and the neighborhood expansion and pruning are repeated until the threshold is satisfied, and a converged clause subgraph is obtained.

[0041] Extract the subject scope, data scope, field granularity, time window and usage restrictions from the converged clause subgraph to generate a constraint set. Construct an event state transition table to perform satisfiability checks on time windows, frequency and cooldown periods. Targeted supplementary checks and re-extraction are performed on items that fail to meet the consistency requirements until the constraint set meets the consistency requirements.

[0042] The node sequence, edge sequence, and attribute sequence are concatenated in a fixed order and combined with the representation sequence fingerprint to calculate the clause subgraph fingerprint, generating a check value used to uniquely identify the clause subgraph;

[0043] The threshold calculation summary, normalized rule fingerprint, representation sequence fingerprint, and directory index topic identifier are aggregated into a version tuple. The version lock value is calculated and the policy version number is fixed. At the same time, the clause subgraph fingerprint, policy version number, adjudication threshold, and constraint set are output.

[0044] Optionally, the process of submitting a request containing a unified identifier and directory index using a domain, invoking the adjudication channel matching clause subgraph of the improved GraphRAG model, and authorizing the contract to automatically adjudicate based on the constraint set, policy version number, and adjudication threshold, generating a one-time token or writing a rejection reason hash, specifically includes:

[0045] Receive access requests submitted using a domain, the access requests containing a unified identifier and a directory index, read the clause subgraph fingerprint, policy version number, adjudication threshold and constraint set from the policy channel output header, and verify the consistency between the version lock of the policy version number and the block height and block timestamp.

[0046] Based on the fingerprint of the clause subgraph, the clause subgraph is located and stored in the consortium blockchain. The representation digests of the node sequence, edge sequence and attribute sequence are reconstructed and the verification hash is calculated. The clause subgraph is then compared with the fingerprint of the clause subgraph to complete the clause subgraph location.

[0047] Initiate a two-level alignment decision. First, perform an initial screening in the fast alignment channel using approximate estimates of name consistency score and relation pattern consistency score. If the initial screening fails, switch to the full alignment channel to calculate the full score of name consistency score, relation pattern consistency score, and adjacency retention score on the ordered representation that retains key adjacencies, and generate an alignment score vector.

[0048] Perform item-by-item verification on the access request and constraint set, complete constraint verification according to subject scope, data scope, field granularity, time window and usage restrictions, construct event state automaton based on time window, frequency and cooldown period and complete the timing satisfiability verification, and generate a list of verification results;

[0049] Perform counterfactual detection on boundary samples, replace the evidence item with the highest contribution in the alignment score vector with the lowest weight item of the same type and the synonym item respectively, calculate the change in alignment score, mark the sample as unstable and request supplementary retrieval when the change exceeds the stability threshold, otherwise confirm the alignment is stable;

[0050] Perform double layer Figure 1 Consistency verification involves constructing an execution trajectory graph, using unified identifiers, directory indexes, access paths, and authorized contract call sequences as nodes and edges. This verifies the consistency between the execution trajectory graph and the subgraph corresponding to the policy channel output header in terms of subject, relationship, and time constraints. If there is a discrepancy, a rejection reason hash is generated.

[0051] Automatic adjudication is completed by combining the alignment score vector, adjudication threshold and verification result list. When the adjudication is successful, a one-time token is generated and written to the consortium blockchain. The one-time token is bound to a unified identifier, directory index, clause subgraph fingerprint, strategy version number, scope of use, effective block and invalid block and records transaction hash and writing sequence number.

[0052] If the request fails, the rejection reason hash is written to the consortium blockchain, and the unified identifier, directory index, clause subgraph fingerprint, policy version number, block height and block timestamp are recorded.

[0053] Optionally, the provision of a domain for off-chain encrypted transmission using a one-time token, the use of the domain to verify integrity and freshness based on data fingerprints and timestamps on the consortium blockchain, and the writing of the verification results into the consortium blockchain, specifically includes:

[0054] Provides the domain to verify the effective and invalid blocks and scope of use of one-time tokens in the consortium blockchain, reads the unified identifier and directory index, and confirms availability;

[0055] The providing domain and the using domain negotiate a session key by combining a one-time token with the public key states of both parties, and establish an off-chain encrypted channel.

[0056] The domain extracts the content to be shared from the original text based on the directory index, calculates the data fingerprint, generates a timestamp, and sends the unified identifier, directory index, data fingerprint, and timestamp along with the encrypted original text to the user domain.

[0057] Using the domain to decrypt off-chain, the data fingerprint of the decrypted original text is recalculated and compared with the data fingerprint under the corresponding unified identifier and directory index on the consortium blockchain for consistency.

[0058] The domain is used to perform a freshness check by combining the timestamp of the block recorded in the consortium blockchain with the timestamp of the time stamp to determine whether it is within a preset time window;

[0059] The domain is used to write the integrity verification result, the freshness verification result, and the one-time token number into the consortium blockchain, recording the block height and the block timestamp;

[0060] When any verification fails, write the rejection reason hash and mark the one-time token as redeemed and non-reusable. When it passes, mark the one-time token as redeemed and complete the confirmation record of off-chain encrypted transmission.

[0061] Optionally, based on the measurement on the consortium blockchain, the settlement and reconciliation of the calls corresponding to the one-time tokens are completed, and the settlement summary and exception records are written to the consortium blockchain. In case of disputes, arbitration and tracing are carried out, specifically including:

[0062] Aggregate the metering records on the consortium blockchain according to the settlement cycle, using one-time tokens, unified identifiers, directory indexes, clause subgraph fingerprints, strategy version numbers, transaction hashes, and write sequence numbers as keys to generate a metering list and total metering amount, and calculate the list hash;

[0063] The providing domain and the using domain submit the reconciliation list hash on the blockchain and reference the metering list, compare the amount, number of items and time window, and output the settlement net amount and the list of difference items and their hashes;

[0064] For discrepancies, trace back the authorization contract registration and cancellation sequence, verify the unified identifier, directory index, clause subgraph fingerprint, policy version number, effective block and invalid block, and generate anomaly reason hash;

[0065] The settlement summary is synthesized from the net settlement amount, total measurement amount, hash of difference entries and hash of abnormal reasons and put on the chain. The block height, block timestamp and write sequence number are recorded. When there is a difference, an abnormal record is written to form an abnormal ledger.

[0066] In the event of a dispute, on-chain evidence is located using a one-time token, the measurement, reconciliation and registration write-off sequence is reconstructed, an arbitration conclusion hash is generated, and a settlement adjustment instruction is issued by the authorized contract. The red-out, supplementary record and net amount correction are recorded on the chain and referenced with the settlement summary.

[0067] The beneficial effects of this invention are:

[0068] Compared to existing loosely structured "on-chain registration, off-chain transmission" schemes, this invention forms a closed loop from "identity—policy—adjudication—transmission—verification—measurement—settlement—arbitration": By uploading decentralized identifiers and public key states to the blockchain, a unified identifier and version replayable identity standard are established, reducing cross-domain impersonation; data fingerprints and timestamps are calculated and uploaded to the blockchain using standardized concatenation of the original text, ensuring consistent verification and tamper-proofing across systems; an improved GraphRAG policy channel is introduced to generate interpretable clause subgraphs and constraint sets, and adjudication thresholds are adaptively set based on historical percentiles, source credibility, and freshness, reducing critical misjudgments; the adjudication channel strongly binds one-time tokens to policy versions, scope of use, and effective blocks, improving the traceability of authorization and minimizing necessary disclosure; off-chain transmission uses session keys for encryption, while on-chain verification of integrity and timeliness is performed using data fingerprints and timestamps, effectively resisting replay and expired calls; token-based measurement and periodic reconciliation are solidified on the blockchain with anomaly and arbitration records, achieving verifiable settlement and traceable disputes. Overall, this invention is significantly superior to existing technologies in terms of credibility, explainability, compliance auditing, cross-domain interoperability, and operational costs. Attached Figure Description

[0069] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0070] Figure 1 This is a flowchart of a blockchain-based cross-domain trusted data sharing method for the Internet of Things proposed in this invention;

[0071] Figure 2 This is a schematic diagram of a blockchain-based cross-domain trusted data sharing method for the Internet of Things proposed in this invention;

[0072] Figure 3 This is a schematic diagram of the improved GraphRAG model in the blockchain-based IoT cross-domain trusted data sharing method proposed in this invention. Detailed Implementation

[0073] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0074] refer to Figure 1-3 A blockchain-based method for cross-domain trusted data sharing in the Internet of Things includes the following steps:

[0075] To provide domain registration and identity verification for entities and devices using the domain, decentralized identifiers and public key status are written into the consortium blockchain to generate a unified identifier and record the block height and block timestamp.

[0076] Provides domain preservation of the original text, only writing data fingerprints, timestamps, unified identifiers, and directory indexes to the consortium blockchain;

[0077] Deploy authorization contracts on the consortium blockchain, build an improved GraphRAG model and enable policy channels, retrieve and align clause subgraphs in the catalog index, and output clause subgraph fingerprints, policy version numbers, adjudication thresholds and constraint sets;

[0078] The domain submits a request containing a unified identifier and directory index, calls the adjudication channel matching clause subgraph of the improved GraphRAG model, and authorizes the contract to automatically adjudicate based on the constraint set, policy version number and adjudication threshold, generating a one-time token or writing the rejection reason hash.

[0079] Provides a domain with a one-time token for off-chain encrypted transmission. The domain is used to verify the integrity and freshness of the data based on the data fingerprint and timestamp on the consortium blockchain, and the verification result is written to the consortium blockchain.

[0080] Based on the measurement on the consortium blockchain, the settlement and reconciliation of the calls corresponding to the one-time tokens are completed, and the settlement summary and exception record are written into the consortium blockchain. In case of disputes, arbitration and tracing are carried out.

[0081] In this embodiment, the process of registering and verifying the identities of entities and devices providing and using the domain, writing decentralized identifiers and public key states into the consortium blockchain, generating unified identifiers, and recording block heights and block timestamps specifically includes:

[0082] Develop and publish rules for registering and verifying the identities of entities and devices that provide and use domains, and open a registration portal on the consortium blockchain;

[0083] Collect registration information of entities and devices that provide and use the domain, and generate the basic data required for decentralized identification;

[0084] Complete registration and verify identity according to the rules, generate decentralized identifiers for the subject and device, and generate corresponding public key states;

[0085] Write the decentralized identifier and public key status into the consortium blockchain, and record the block height and block timestamp at the time of writing;

[0086] After registration and identity verification are completed, a unified identifier is generated for the subject and the device, and the unified identifier and the decentralized identifier are bound together on the consortium blockchain.

[0087] When the public key status or decentralized identifier of the subject and device changes, it is synchronously updated to the consortium blockchain, and the block height and block timestamp are recorded again.

[0088] This implementation method establishes a replayable identity caliber by registering decentralized identifiers and public key statuses on the consortium blockchain and generating a unified identifier, thereby reducing the risk of cross-domain impersonation and mismatch and facilitating full-chain traceability.

[0089] In this embodiment, the provision of the domain preserves the original text, only writing the data fingerprint, time stamp, unified identifier, and directory index into the consortium blockchain, specifically including:

[0090] Provide the domain to retain the original text in an off-chain controlled storage environment, determine the set of original texts that need to be written to the consortium blockchain, the original texts are real data content generated or hosted by the domain and not desensitized or pruned, including sensor sampling sequences, equipment logs, control command records, image and audio / video files, industrial documents and reports and interface return messages and archived files;

[0091] Perform normalization processing on each original text to unify the character set, newline character, time zone, and byte order, and obtain a normalized byte sequence;

[0092] Generate a timestamp for each original text, generate a table of contents index for each original text, and establish a one-to-one binding relationship between the table of contents index and the unified identifier;

[0093] The unified identifier, directory index, normalized byte sequence length and timestamp encoding value, and normalized byte sequence are concatenated in a fixed order to form a byte stream to be fingerprinted;

[0094] A collision-resistant cryptographic hash algorithm is used to calculate the data fingerprint from the byte stream to be fingerprinted.

[0095] Entries are written to the consortium blockchain using data fingerprints, timestamps, unified identifiers, and directory indexes, and the block height and timestamp at the time of writing are recorded.

[0096] When the original text is added, modified, or deleted, normalization processing is re-executed, timestamps and directory indexes are regenerated and maintained in a one-to-one binding relationship with the unified identifier, the byte stream to be fingerprinted is reassembled and the data fingerprint is calculated again, the updated entry is written to the consortium blockchain and the block height and block timestamp are recorded, and the previous target is marked as historical state.

[0097] This implementation method standardizes and concatenates the original text, calculates data fingerprints and timestamps, and puts only the fingerprints and indexes on the chain to achieve cross-system consistent verification and anti-tampering, supporting version playback and data source verification.

[0098] In this embodiment, the step of deploying the authorization contract on the consortium blockchain to build the improved GraphRAG model specifically includes:

[0099] Deploy an authorization contract on the consortium blockchain and initialize the registration interface. The authorization contract provides an identity verification interface, an atomic write interface, a read-only reference interface and a version lock interface for registering the clause subgraph fingerprint, policy version number, adjudication threshold and constraint set. It also records the block height, block timestamp and write sequence number each time registration is performed, and returns the registration handle.

[0100] An improved GraphRAG model is constructed, comprising a common base layer, a policy channel, and an adjudication channel. The common base layer includes a data input and indexing layer, a semantic normalization layer, an index retrieval layer, a clause subgraph representation normalization layer, an alignment scoring function family layer, an adjudication threshold, a policy version parameter management layer, and a registration interface layer for the consortium blockchain.

[0101] The data input and indexing layer receives the directory index and its corresponding unified identifier, performs word segmentation, field mapping, case unification and stop character cleanup on the relevant fields of the directory index, establishes searchable index items and inverted indexes, and outputs normalized field sequences and index item identifiers.

[0102] The semantic normalization layer receives a normalized field sequence, a unified identifier, and a directory index, and embeds the unified identifier code and directory index code in a fixed order to generate a standardized semantic vector and a position code sequence.

[0103] The index retrieval layer receives the directory index, standardized semantic vectors and index item identifiers, performs vector similarity retrieval and keyword matching in parallel, and performs weighted sorting based on similarity score, time freshness and source credibility, and outputs a candidate node list, hit order and scoring trajectory identifier.

[0104] The clause subgraph representation specification layer receives the candidate node list and the hit order, extracts nodes and edges according to the type consistency and key adjacency retention rules, records the node sequence, edge sequence and attribute sequence in a fixed order, and generates the representation sequence and representation sequence fingerprint;

[0105] The alignment scoring function family layer receives node sequences, edge sequences, attribute sequences and scoring trajectory identifiers, calculates name consistency score, relation pattern consistency score and adjacency retention score, forms an alignment scoring vector, and extracts the minimum interpretation path as alignment evidence;

[0106] The adjudication threshold and strategy version parameter management layer receives the aligned scoring vector and directory index, and calculates the adaptive value of the adjudication threshold. The adaptive calculation uses the historical pass percentile as the benchmark threshold, superimposed with positive corrections for topic risk score, negative corrections for freshness score, and negative corrections for credibility score. It is then pruned within the minimum and maximum boundaries and quantized by thousandths. At the same time, a strategy version number is assigned and a version lock is set. The historical pass percentile, topic risk score, freshness score, credibility score, and weighted reorganization are written into the strategy version parameter table to generate a threshold calculation summary.

[0107] The registration interface layer of the consortium blockchain receives the policy version number and version lock flag, generates a registration handle, and provides atomic write and read-only reference paths for the authorization contract.

[0108] This implementation provides interpretable clause subgraphs and adaptive thresholds through an improved GraphRAG public base and registration interface, threshold and version management, combined with version locks and registration handles, to ensure policy consistency and parameter traceability.

[0109] In this embodiment, enabling the policy channel involves retrieving and aligning the clause subgraphs in the directory index, and outputting the clause subgraph fingerprint, policy version number, adjudication threshold, and constraint set. Specifically, this includes:

[0110] The strategy channel is enabled, and neighborhood expansion and pruning are performed according to the candidate node list and the hit order. Following the rules of type consistency and key adjacency retention, entities with the same name are merged and low-weight redundant edges and isolated nodes are deleted to form an ordered representation of the clause subgraph and obtain the node sequence, edge sequence, attribute sequence and representation sequence.

[0111] Alignment of clause subgraphs is confirmed based on alignment score vector and adjudication threshold. Thresholds are determined for name consistency score, relation pattern consistency score and adjacency retention score. If not satisfied, the search keywords and time window are expanded with the directory index and standardized semantic vector as prompts, and the neighborhood expansion and pruning are repeated until the threshold is satisfied, and a converged clause subgraph is obtained.

[0112] Extract the subject scope, data scope, field granularity, time window and usage restrictions from the converged clause subgraph to generate a constraint set. Construct an event state transition table to perform satisfiability checks on time windows, frequency and cooldown periods. Targeted supplementary checks and re-extraction are performed on items that fail to meet the consistency requirements until the constraint set meets the consistency requirements.

[0113] The node sequence, edge sequence, and attribute sequence are concatenated in a fixed order and combined with the representation sequence fingerprint to calculate the clause subgraph fingerprint, generating a check value used to uniquely identify the clause subgraph;

[0114] The threshold calculation summary, normalized rule fingerprint, representation sequence fingerprint, and directory index topic identifier are aggregated into a version tuple. The version lock value is calculated and the policy version number is fixed. At the same time, the clause subgraph fingerprint, policy version number, adjudication threshold, and constraint set are output.

[0115] This implementation method generates clause subgraph fingerprints, policy version numbers, adjudication thresholds, and constraint sets by expanding and pruning the neighborhood of the policy channel, aligning the decision, and performing targeted supplementary inspections, thereby improving release accuracy and reducing policy conflicts.

[0116] In this embodiment, the process of submitting a request containing a unified identifier and directory index using a domain, invoking the adjudication channel matching clause subgraph of the improved GraphRAG model, and having the authorization contract automatically adjudicate based on the constraint set, policy version number, and adjudication threshold, generating a one-time token or writing a rejection reason hash, specifically includes:

[0117] Receive access requests submitted using a domain, the access requests containing a unified identifier and a directory index, read the clause subgraph fingerprint, policy version number, adjudication threshold and constraint set from the policy channel output header, and verify the consistency between the version lock of the policy version number and the block height and block timestamp.

[0118] Based on the fingerprint of the clause subgraph, the clause subgraph is located and stored in the consortium blockchain. The representation digests of the node sequence, edge sequence and attribute sequence are reconstructed and the verification hash is calculated. The clause subgraph is then compared with the fingerprint of the clause subgraph to complete the clause subgraph location.

[0119] Initiate a two-level alignment decision. First, perform an initial screening in the fast alignment channel using approximate estimates of name consistency score and relation pattern consistency score. If the initial screening fails, switch to the full alignment channel to calculate the full score of name consistency score, relation pattern consistency score, and adjacency retention score on the ordered representation that retains key adjacencies, and generate an alignment score vector.

[0120] Perform item-by-item verification on the access request and constraint set, complete constraint verification according to subject scope, data scope, field granularity, time window and usage restrictions, construct event state automaton based on time window, frequency and cooldown period and complete the timing satisfiability verification, and generate a list of verification results;

[0121] Perform counterfactual detection on boundary samples, replace the evidence item with the highest contribution in the alignment score vector with the lowest weight item of the same type and the synonym item respectively, calculate the change in alignment score, mark the sample as unstable and request supplementary retrieval when the change exceeds the stability threshold, otherwise confirm the alignment is stable;

[0122] Perform double layer Figure 1 Consistency verification involves constructing an execution trajectory graph, using unified identifiers, directory indexes, access paths, and authorized contract call sequences as nodes and edges. This verifies the consistency between the execution trajectory graph and the subgraph corresponding to the policy channel output header in terms of subject, relationship, and time constraints. If there is a discrepancy, a rejection reason hash is generated.

[0123] Automatic adjudication is completed by combining the alignment score vector, adjudication threshold and verification result list. When the adjudication is successful, a one-time token is generated and written to the consortium blockchain. The one-time token is bound to a unified identifier, directory index, clause subgraph fingerprint, strategy version number, scope of use, effective block and invalid block and records transaction hash and writing sequence number.

[0124] If the request fails, the rejection reason hash is written to the consortium blockchain, and the unified identifier, directory index, clause subgraph fingerprint, policy version number, block height and block timestamp are recorded.

[0125] This implementation enhances authorization traceability and auditability by using two-level alignment and item-by-item constraint verification of the adjudication channel, combined with automatic adjudication based on adaptive thresholds and issuing one-time tokens or writing rejection reason hashes.

[0126] In this embodiment, the provision of a domain using a one-time token for off-chain encrypted transmission, the use of the domain to verify integrity and freshness based on data fingerprints and timestamps on the consortium blockchain, and the writing of the verification results into the consortium blockchain specifically include:

[0127] Provides the domain to verify the effective and invalid blocks and scope of use of one-time tokens in the consortium blockchain, reads the unified identifier and directory index, and confirms availability;

[0128] The providing domain and the using domain negotiate a session key by combining a one-time token with the public key states of both parties, and establish an off-chain encrypted channel.

[0129] The domain extracts the content to be shared from the original text based on the directory index, calculates the data fingerprint, generates a timestamp, and sends the unified identifier, directory index, data fingerprint, and timestamp along with the encrypted original text to the user domain.

[0130] Using the domain to decrypt off-chain, the data fingerprint of the decrypted original text is recalculated and compared with the data fingerprint under the corresponding unified identifier and directory index on the consortium blockchain for consistency.

[0131] The domain is used to perform a freshness check by combining the timestamp of the block recorded in the consortium blockchain with the timestamp of the time stamp to determine whether it is within a preset time window;

[0132] The domain is used to write the integrity verification result, the freshness verification result, and the one-time token number into the consortium blockchain, recording the block height and the block timestamp;

[0133] When any verification fails, write the rejection reason hash and mark the one-time token as redeemed and non-reusable. When it passes, mark the one-time token as redeemed and complete the confirmation record of off-chain encrypted transmission.

[0134] This implementation method uses a one-time token to negotiate the session key for off-chain encrypted transmission, and verifies the integrity and freshness of the data with on-chain data fingerprints and timestamps, effectively resisting the risks of replay and expired data.

[0135] In this embodiment, the settlement and reconciliation of the calls corresponding to the one-time tokens are completed based on the measurement on the consortium blockchain, and the settlement summary and exception record are written to the consortium blockchain. Arbitration and tracing are carried out in case of disputes. Specifically, this includes:

[0136] Aggregate the metering records on the consortium blockchain according to the settlement cycle, using one-time tokens, unified identifiers, directory indexes, clause subgraph fingerprints, strategy version numbers, transaction hashes, and write sequence numbers as keys to generate a metering list and total metering amount, and calculate the list hash;

[0137] The providing domain and the using domain submit the reconciliation list hash on the blockchain and reference the metering list, compare the amount, number of items and time window, and output the settlement net amount and the list of difference items and their hashes;

[0138] For discrepancies, trace back the authorization contract registration and cancellation sequence, verify the unified identifier, directory index, clause subgraph fingerprint, policy version number, effective block and invalid block, and generate anomaly reason hash;

[0139] The settlement summary is synthesized from the net settlement amount, total measurement amount, hash of difference entries and hash of abnormal reasons and put on the chain. The block height, block timestamp and write sequence number are recorded. When there is a difference, an abnormal record is written to form an abnormal ledger.

[0140] In the event of a dispute, on-chain evidence is located using a one-time token, the measurement, reconciliation and registration write-off sequence is reconstructed, an arbitration conclusion hash is generated, and a settlement adjustment instruction is issued by the authorized contract. The red-out, supplementary record and net amount correction are recorded on the chain and referenced with the settlement summary.

[0141] This implementation method uses on-chain measurement, reconciliation, and anomaly ledgers aggregated by tokens, and reconstructs the authorization and write-off sequence to complete arbitration and settlement adjustments in the event of a dispute, forming a closed-loop evidence chain from invocation to settlement, thereby reducing dispute costs.

[0142] Example 1:

[0143] To verify the feasibility of this invention in practice, it was applied to a data sharing project between a provincial power company's cross-regional wind farms and equipment manufacturers. The wind farms have long needed to provide high-frequency sensor data and maintenance logs to the manufacturers for remote diagnostics. However, traditional methods rely on manual approval, FTP retrieval, and email communication, resulting in untraceable processes, delayed data arrival, and inconsistent data reporting across different regional subsidiaries, easily leading to issues such as unauthorized fields and duplicate data retrieval. The pilot project covered three wind farms and two maintenance service providers, involving 126 wind turbines and approximately 600 edge sensors. Data types included temperature and vibration monitoring, gearbox oil temperature, control command logs, and alarm snapshots.

[0144] On the consortium blockchain, entities and devices providing and using domains are registered and their identities verified. Decentralized identifiers and public key states are written, a unified identifier is generated, and block height and timestamps are recorded. The providing domain retains its original text off-chain, with only the directory index, timestamp, and data fingerprint uploaded to the blockchain. An authorization contract is deployed on-chain, and an improved GraphRAG policy channel and adjudication channel are built. The policy channel aligns the clause subgraph from the directory index and regulations and outputs the policy version, adjudication threshold, and constraint set. The adjudication channel automatically adjudicates and generates a one-time token when a specific request arrives, matching the clause subgraph. Off-chain transmission uses session key encryption. The using domain completes integrity and freshness verification based on the on-chain fingerprint and timestamp and writes the results back to the blockchain. Metering uses the one-time token as the key to complete settlement and reconciliation. In case of disputes, arbitration can be traced back according to the chain of evidence.

[0145] The pilot program first standardized the directory index at the edge gateways of three wind farms providing the domain, unifying the character set, time, and byte order to ensure that fields extracted from logs from different manufacturers could be mapped to a common template. GraphRAG's policy channel uses a policy library, confidentiality agreements, and regulatory requirements as its corpus, automatically extracting clause subgraphs such as subject scope, data range, field granularity, time windows, and usage restrictions. Examples include: "The original vibration waveform is only allowed to provide one quantile feature every fifteen minutes; the accuracy of equipment location information must be less than 100 meters; overseas maintenance teams cannot query personal attendance logs between midnight and 6 AM." Policy outputs are registered on-chain as version 102, with two levels of thresholds: pass and partial. When the maintenance party initiates a request, the adjudication channel performs a local search based on the unified identifier and directory index. It calculates the name consistency score, relational schema consistency score, and adjacency retention score on the ordered representation of key adjacency retention. It then combines the source credibility and the freshness of the terms to obtain a compliance score. If the threshold is exceeded, the request is automatically released and a one-time token is issued. The token is bound to the scope of use, the effective block, and the invalid block. It is then cancelled after the transmission is completed.

[0146] Table 1 Summary of Comprehensive Indicators

[0147]

[0148] As shown in Table 1, after adopting this invention, the request approval rate increased from 70% to over 90%. This is because the policy channel's clause subgraph unified and aligned the policy interpretations of different regions, reducing rejections caused by inconsistencies in format and field meaning. The average automatic adjudication latency decreased from the five-second range to less than one second. The fast alignment sub-channel of the adjudication channel can complete the approval for most common requests without switching to the fully aligned channel. The integrity inconsistency rate and freshness timeout rate decreased by approximately four-fifths and three-fifths, respectively. This is because fingerprints and timestamps are used to verify off-chain transmissions before and after transmission, and requests exceeding the preset window are intercepted and the reason hash is written back. The one-time token redemption rate stabilized at over 99%, eliminating the problem of repeated link reuse. The end-to-end arrival latency decreased from nearly 100 seconds to approximately 20 seconds, mainly benefiting from automated approval, directory indexing and positioning, and instant token issuance and use. The rate of duplicate data retrieval decreased from 12.7% to approximately 2%. GraphRAG merged entities with the same name and aliases in the ordered representation of the directory index and used tokens and directory indexes as keys for measurement, preventing the same request from triggering multiple times. The number of unauthorized field interceptions decreased from over eighty times per week to zero because the constraint set solidified the granularity and usage restrictions of fields into contract verification rules. The accuracy of measurement and settlement is close to 100%, and the dispute rate decreased from 2.3% to around 0.3%. When disputes occur, the data retrieval and reconciliation sequence can be reconstructed based on the evidence chain of one-time tokens and directory indexes to quickly locate the cause of discrepancies.

[0149] In real-world business processes, the implementation of this invention does not alter the original text storage strategy of the providing domain. Sensitive data remains in a controlled environment, with only fingerprints, timestamps, unified identifiers, and directory indexes stored on-chain, eliminating concerns about external leakage of the original text. The policy channel expresses policy terms in a graph structure, allowing rules from different sources to naturally merge under consistent type and key adjacency retention rules, forming an interpretable minimal interpretation path. When laws or contracts are updated, only the policy version and thresholds need to be recalculated, and the adjudication channel can take effect with the new version on the next request. Counterfactual probing is performed on boundary samples during the adjudication process. If the compliance score is extremely sensitive to a single evidence item, targeted supplementary checks will be requested to improve stability and robustness. After the settlement cycle ends, both parties only need to submit the reconciliation list hash on-chain and reference the measurement list to complete the reconciliation. Discrepancies can be traced through the execution trajectory graph of the authorized contract. If necessary, the on-chain arbitration contract issues settlement adjustment instructions and retains an anomaly ledger.

[0150] Based on the quantitative results and operational feedback from the pilot project, this invention achieves cross-domain trusted data sharing without copying the original data, significantly improving the efficiency of automated adjudication, reducing delays and disputes, and strengthening traceability and compliance control. For industries outside the power sector, such as oil and gas, rail transit, and intelligent manufacturing, as long as they have catalog indexes, rule corpora, and consortium blockchain infrastructure, they can be deployed in the same way. Rapid expansion and smooth upgrades are achieved through the decoupling of policy channels and adjudication channels, thus achieving a better engineering balance between data security and sharing efficiency.

[0151] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A blockchain-based method for cross-domain trusted data sharing in the Internet of Things, characterized in that, Includes the following steps: To provide domain registration and identity verification for entities and devices using the domain, decentralized identifiers and public key status are written into the consortium blockchain to generate a unified identifier and record the block height and block timestamp. Provides domain preservation of the original text, only writing data fingerprints, timestamps, unified identifiers, and directory indexes to the consortium blockchain; Deploy authorization contracts on the consortium blockchain, build an improved GraphRAG model and enable policy channels, retrieve and align clause subgraphs in the catalog index, and output clause subgraph fingerprints, policy version numbers, adjudication thresholds and constraint sets; The domain submits a request containing a unified identifier and directory index, calls the adjudication channel matching clause subgraph of the improved GraphRAG model, and authorizes the contract to automatically adjudicate based on the constraint set, policy version number and adjudication threshold, generating a one-time token or writing the rejection reason hash. The domain provides off-chain encrypted transmission based on one-time tokens, and uses the domain to verify the integrity and freshness of data based on the data fingerprint and timestamp on the consortium blockchain, and writes the verification results into the consortium blockchain. Based on the measurement on the consortium blockchain, the settlement and reconciliation of the calls corresponding to the one-time tokens are completed, and the settlement summary and exception record are written into the consortium blockchain. In case of disputes, arbitration and tracing are carried out.

2. The blockchain-based cross-domain trusted data sharing method for the Internet of Things according to claim 1, characterized in that, The process of registering and verifying the identities of entities and devices using the domain, writing decentralized identifiers and public key states into the consortium blockchain, generating a unified identifier, and recording block height and block timestamps specifically includes: Develop and publish rules for registering and verifying the identities of entities and devices that provide and use domains, and open a registration portal on the consortium blockchain; Collect registration information of entities and devices that provide and use the domain, and generate the basic data required for decentralized identification; Complete registration and verify identity according to the rules, generate decentralized identifiers for the subject and device, and generate corresponding public key states; Write the decentralized identifier and public key status into the consortium blockchain, and record the block height and block timestamp at the time of writing; After registration and identity verification are completed, a unified identifier is generated for the subject and the device, and the unified identifier and the decentralized identifier are bound together on the consortium blockchain. When the public key status or decentralized identifier of the subject and device changes, it is synchronously updated to the consortium blockchain, and the block height and block timestamp are recorded again.

3. The blockchain-based cross-domain trusted data sharing method for the Internet of Things according to claim 1, characterized in that, The provision of the domain preserves the original text, only writing the data fingerprint, timestamp, unified identifier, and directory index into the consortium blockchain, specifically including: Provide the domain to retain the original text in an off-chain controlled storage environment, determine the set of original texts that need to be written to the consortium blockchain, the original texts are real data content generated or hosted by the domain and not desensitized or pruned, including sensor sampling sequences, equipment logs, control command records, image and audio / video files, industrial documents and reports and interface return messages and archived files; Perform normalization processing on each original text to unify the character set, newline character, time zone, and byte order, and obtain a normalized byte sequence; Generate a timestamp for each original text, generate a table of contents index for each original text, and establish a one-to-one binding relationship between the table of contents index and the unified identifier; The unified identifier, directory index, normalized byte sequence length and timestamp encoding value, and normalized byte sequence are concatenated in a fixed order to form a byte stream to be fingerprinted; A collision-resistant cryptographic hash algorithm is used to calculate the data fingerprint from the byte stream to be fingerprinted. Entries are written to the consortium blockchain using data fingerprints, timestamps, unified identifiers, and directory indexes, and the block height and timestamp at the time of writing are recorded. When the original text is added, modified, or deleted, normalization processing is re-executed, timestamps and directory indexes are regenerated and maintained in a one-to-one binding relationship with the unified identifier, the byte stream to be fingerprinted is reassembled and the data fingerprint is calculated again, the updated entry is written to the consortium blockchain and the block height and block timestamp are recorded, and the previous target is marked as historical state.

4. The blockchain-based cross-domain trusted data sharing method for the Internet of Things according to claim 1, characterized in that, The deployment of authorization contracts on the consortium blockchain to build the improved GraphRAG model specifically includes: Deploy the authorization contract and initialize the registration interface on the consortium blockchain; An improved GraphRAG model is constructed, comprising a common base layer, a policy channel, and an adjudication channel. The common base layer includes a data input and indexing layer, a semantic normalization layer, an index retrieval layer, a clause subgraph representation normalization layer, an alignment scoring function family layer, an adjudication threshold, a policy version parameter management layer, and a registration interface layer for the consortium blockchain. The data input and indexing layer receives the directory index and its corresponding unified identifier, performs word segmentation, field mapping, case unification and stop character cleanup on the relevant fields of the directory index, establishes searchable index items and inverted indexes, and outputs normalized field sequences and index item identifiers. The semantic normalization layer receives a normalized field sequence, a unified identifier, and a directory index, and embeds the unified identifier code and directory index code in a fixed order to generate a standardized semantic vector and a position code sequence. The index retrieval layer receives the directory index, standardized semantic vectors and index item identifiers, performs vector similarity retrieval and keyword matching in parallel, and performs weighted sorting based on similarity score, time freshness and source credibility, and outputs a candidate node list, hit order and scoring trajectory identifier. The clause subgraph representation specification layer receives the candidate node list and the hit order, extracts nodes and edges according to the type consistency and key adjacency retention rules, records the node sequence, edge sequence and attribute sequence in a fixed order, and generates the representation sequence and representation sequence fingerprint; The alignment scoring function family layer receives node sequences, edge sequences, attribute sequences and scoring trajectory identifiers, calculates name consistency score, relation pattern consistency score and adjacency retention score, forms an alignment scoring vector, and extracts the minimum interpretation path as alignment evidence; The adjudication threshold and strategy version parameter management layer receives the aligned scoring vector and directory index, and calculates the adaptive value of the adjudication threshold. The adaptive calculation uses the historical pass percentile as the benchmark threshold, superimposed with positive corrections for topic risk score, negative corrections for freshness score, and negative corrections for credibility score. It is then pruned within the minimum and maximum boundaries and quantized by thousandths. At the same time, a strategy version number is assigned and a version lock is set. The historical pass percentile, topic risk score, freshness score, credibility score, and weighted reorganization are written into the strategy version parameter table to generate a threshold calculation summary. The registration interface layer of the consortium blockchain receives the policy version number and version lock flag, generates a registration handle, and provides atomic write and read-only reference paths for the authorization contract.

5. The blockchain-based cross-domain trusted data sharing method for the Internet of Things according to claim 1, characterized in that, The enabled policy channel retrieves and aligns clause subgraphs in the directory index, outputting clause subgraph fingerprints, policy version numbers, adjudication thresholds, and constraint sets. Specifically, this includes: The strategy channel is enabled, and neighborhood expansion and pruning are performed according to the candidate node list and the hit order. Following the rules of type consistency and key adjacency retention, entities with the same name are merged and low-weight redundant edges and isolated nodes are deleted to form an ordered representation of the clause subgraph and obtain the node sequence, edge sequence, attribute sequence and representation sequence. Alignment of clause subgraphs is confirmed based on alignment score vector and adjudication threshold. Thresholds are determined for name consistency score, relation pattern consistency score and adjacency retention score. If not satisfied, the search keywords and time window are expanded with the directory index and standardized semantic vector as prompts, and the neighborhood expansion and pruning are repeated until the threshold is satisfied, and a converged clause subgraph is obtained. Extract the subject scope, data scope, field granularity, time window and usage restrictions from the converged clause subgraph to generate a constraint set. Construct an event state transition table to perform satisfiability checks on time windows, frequency and cooldown periods. Targeted supplementary checks and re-extraction are performed on items that fail to meet the consistency requirements until the constraint set meets the consistency requirements. The node sequence, edge sequence, and attribute sequence are concatenated in a fixed order and combined with the representation sequence fingerprint to calculate the clause subgraph fingerprint, generating a verification value used to uniquely identify the clause subgraph; The threshold calculation summary, normalized rule fingerprint, representation sequence fingerprint, and directory index topic identifier are aggregated into a version tuple. The version lock value is calculated and the policy version number is fixed. At the same time, the clause subgraph fingerprint, policy version number, adjudication threshold, and constraint set are output.

6. The blockchain-based cross-domain trusted data sharing method for the Internet of Things according to claim 1, characterized in that, The method of submitting a request containing a unified identifier and directory index using a domain calls the adjudication channel matching clause subgraph of the improved GraphRAG model. The authorization contract automatically adjudicates based on the constraint set, policy version number, and adjudication threshold, generating a one-time token or writing a rejection reason hash, specifically including: Receive access requests submitted using a domain, the access requests containing a unified identifier and a directory index, read the clause subgraph fingerprint, policy version number, adjudication threshold and constraint set from the policy channel output header, and verify the consistency between the version lock of the policy version number and the block height and block timestamp. Based on the fingerprint of the clause subgraph, the clause subgraph is located and stored in the consortium blockchain. The representation digests of the node sequence, edge sequence and attribute sequence are reconstructed and the verification hash is calculated. The clause subgraph is then compared with the fingerprint of the clause subgraph to complete the clause subgraph location. Initiate a two-level alignment decision. First, perform an initial screening in the fast alignment channel using approximate estimates of name consistency score and relation pattern consistency score. If the initial screening fails, switch to the full alignment channel to calculate the full score of name consistency score, relation pattern consistency score, and adjacency retention score on the ordered representation that retains key adjacencies, and generate an alignment score vector. Perform item-by-item verification on the access request and constraint set, complete constraint verification according to subject scope, data scope, field granularity, time window and usage restrictions, construct event state automaton based on time window, frequency and cooldown period and complete the timing satisfiability verification, and generate a list of verification results; Perform counterfactual detection on boundary samples, replace the evidence item with the highest contribution in the alignment score vector with the lowest weight item of the same type and the synonym item respectively, calculate the change in alignment score, mark the sample as unstable and request supplementary retrieval when the change exceeds the stability threshold, otherwise confirm the alignment is stable; Perform a two-layer graph consistency check, construct an execution trajectory graph, and use unified identifiers, directory indexes, access paths and authorized contract call sequences as nodes and edges to verify the consistency between the execution trajectory graph and the subgraph corresponding to the policy channel output header in terms of subject, relationship and time constraints. If there is a discrepancy, generate a rejection reason hash. Automatic adjudication is completed by combining the alignment score vector, adjudication threshold and verification result list. When the adjudication is successful, a one-time token is generated and written to the consortium blockchain. The one-time token is bound to a unified identifier, directory index, clause subgraph fingerprint, strategy version number, scope of use, effective block and invalid block and records transaction hash and writing sequence number. If the request fails, the rejection reason hash is written to the consortium blockchain, and the unified identifier, directory index, clause subgraph fingerprint, policy version number, block height and block timestamp are recorded.

7. A blockchain-based method for cross-domain trusted data sharing in the Internet of Things, as described in claim 1, is characterized in that... The provided domain uses off-chain encrypted transmission based on a one-time token. The domain verifies integrity and freshness based on data fingerprints and timestamps on the consortium blockchain, and writes the verification results into the consortium blockchain. Specifically, this includes: Provides the domain to verify the effective and invalid blocks and scope of use of one-time tokens in the consortium blockchain, reads the unified identifier and directory index, and confirms availability; The providing domain and the using domain negotiate a session key by combining a one-time token with the public key states of both parties, and establish an off-chain encrypted channel. The domain extracts the content to be shared from the original text based on the directory index, calculates the data fingerprint, generates a timestamp, and sends the unified identifier, directory index, data fingerprint, and timestamp along with the encrypted original text to the user domain. Using the domain to decrypt off-chain, the data fingerprint of the decrypted original text is recalculated and compared with the data fingerprint under the corresponding unified identifier and directory index on the consortium blockchain. The domain is used to perform a freshness check by combining the timestamp of the block recorded in the consortium blockchain with the timestamp of the time stamp to determine whether it is within a preset time window; The domain is used to write the integrity verification result, the freshness verification result, and the one-time token number into the consortium blockchain, recording the block height and the block timestamp; When any verification fails, write the rejection reason hash and mark the one-time token as redeemed and non-reusable. When it passes, mark the one-time token as redeemed and complete the confirmation record of off-chain encrypted transmission.

8. A blockchain-based method for cross-domain trusted data sharing in the Internet of Things, as described in claim 1, is characterized in that... Based on the measurement on the consortium blockchain, settlement and reconciliation are completed for the calls corresponding to one-time tokens. Settlement summaries and exception records are written to the consortium blockchain. In case of disputes, arbitration and tracing are carried out. Specifically, this includes: Aggregate the metering records on the consortium blockchain according to the settlement cycle, using one-time tokens, unified identifiers, directory indexes, clause subgraph fingerprints, strategy version numbers, transaction hashes, and write sequence numbers as keys to generate a metering list and total metering amount, and calculate the list hash; The providing domain and the using domain submit the reconciliation list hash on the blockchain and reference the metering list, compare the amount, number of items and time window, and output the settlement net amount and the list of difference items and their hashes; For discrepancies, trace back the authorization contract registration and cancellation sequence, verify the unified identifier, directory index, clause subgraph fingerprint, policy version number, effective block and invalid block, and generate an anomaly reason hash; The settlement summary is synthesized from the net settlement amount, total measurement amount, hash of difference entries and hash of abnormal reasons and put on the chain. The block height, block timestamp and write sequence number are recorded. When there is a difference, an abnormal record is written to form an abnormal ledger. In the event of a dispute, on-chain evidence is located using a one-time token, the measurement, reconciliation and registration write-off sequence is reconstructed, an arbitration conclusion hash is generated, and a settlement adjustment instruction is issued by the authorized contract. The red-out, supplementary record and net amount correction are recorded on the chain and referenced with the settlement summary.

Images