Attribute-aware dynamic encryption access control method

Abstract

The invention relates to an attribute-aware dynamic encryption access control method, and belongs to the technical field of communication security. According to the technical scheme, the method comprises the steps that a multi-channel social attribute tensor is constructed by collecting a static social relation, a dynamic behavior relation, a bidirectional interaction relation and a deep social relation of a user, and a social attribute vector is obtained through vector embedding; a linear secret sharing structure access matrix is generated based on the vector, an access strategy is adjusted in real time according to a behavior deviation score and a neighborhood consistency index, a ring learning error lattice password is adopted to generate a user private key bound with a social state, and attribute-based encryption is performed according to an access structure to generate a ciphertext; during decryption, verifying social attribute satisfaction and then recovering a plaintext, and realizing verifiable revocation and social state tracking by using a Merkle tree and a fingerprint chain; according to the invention, the dynamic adaptability and security of access control can be obviously improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of communication security, and relates to an attribute-aware dynamic encryption access control method. BACKGROUND

[0002] With the popularity of cloud computing, data sharing platforms and multi-agent collaboration systems, access control and data encryption technology have become a key issue in the field of information security. As one of the mainstream solutions, attribute-based encryption (ABE) can control decryption permissions based on attribute satisfaction, but existing technologies generally have the following shortcomings:

[0003] In traditional ABE, user attributes are usually fixed information, such as roles, departments or labels, which cannot reflect the dynamic changes of user behavior. For example, recent abnormal behavior, decreased trust, changed social relationships or significantly reduced interaction frequency of users cannot be captured by static attributes. This static nature may lead to attackers abusing their privileges for a long time after being authorized, and the system cannot respond in a timely manner.

[0004] Access permissions based on real-world scenarios often rely heavily on complex social relationships. For example, whether members have long-term cooperation, whether individual behavior conforms to group consistency, whether there is mutual trust between the two parties, and whether the behavior sequence is abnormal, etc. Dynamic social relationship information cannot be effectively integrated and quantified by existing ABE schemes, resulting in overly rigid access control strategies that are difficult to adapt to real-world social application scenarios.

[0005] The revocation mechanism of traditional ABE has efficiency and credibility problems. Common revocation methods such as re-issuing keys, re-encrypting data or broadcasting revocation lists often have high operational costs and can cause service disruptions. More importantly, existing schemes generally lack verifiable revocation proofs, state non-repudiation mechanisms and fine-grained attribute-level revocation capabilities, which makes the revocation process lack transparency, attackers can evade tracking, and the system security state is difficult to audit.

[0006] With the rapid development of quantum computing technology, traditional ABE schemes based on the mathematical difficulty of bilinear pairings are gradually facing potential security threats. Lattice cryptography is considered an important candidate for the next generation of cryptography due to its resistance to quantum attacks, but in existing lattice cryptography ABE schemes, it is still difficult to effectively integrate dynamically changing social attributes. In addition, these schemes lack a flexible access structure suitable for social behavior-driven, do not support complex state binding, behavior tracking and verifiable revocation mechanisms. Therefore, the industry urgently needs a new method that can intelligently perceive the dynamics of social attributes and achieve precise access control under the framework of post-quantum cryptography. SUMMARY

[0007] In view of this, the purpose of the present invention is to provide an attribute-aware dynamic encrypted access control method.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] An attribute-aware dynamic encrypted access control method, the method comprising:

[0010] The social attribute modeling steps involve collecting users' static social relationships, dynamic behavioral relationships, two-way interaction relationships, and deep social relationships, and constructing a multi-channel social attribute tensor based on these relationships. Vector embedding is then performed on the tensor to obtain a social attribute vector for access control.

[0011] The access structure dynamic generation step involves constructing an access matrix for a Linear Secret Sharing Structure (LSSS) based on the social attribute vector, and performing real-time reinforcement and adjustment on the access structure according to behavioral deviation scores, neighborhood consistency indicators, and adaptive thresholds.

[0012] The key generation step involves constructing a public key and a master key based on Ring Learning With Errors (RLWE) for lattice cryptography, and generating a user private key bound to the social state based on the user's social attribute vector and access structure.

[0013] The encryption process involves performing attribute-based encryption on the plaintext based on the access structure matrix to generate ciphertext components related to social attributes and recording the social attribute binding vector.

[0014] The decryption process involves verifying whether the user's social attribute vector satisfies the access structure when the user initiates a decryption request. If so, the plaintext is recovered based on the user's private key and the access structure matrix.

[0015] The revocation and verification steps generate revocation proofs based on the Merkle tree structure, perform verifiable revocation on invalid social attributes or users, and use the fingerprint chain structure to verify whether the social state has been tampered with.

[0016] Furthermore, the social attributes in the social attribute modeling step include:

[0017] Static social relationship (SSR) is based on the static attribute vector s of any two nodes i and j. i ,s j Calculate mutual information MI(s) i ,s j and historical co-occurrence frequency ρ ij And define weighted fusion index

[0018] Dynamic behavior relationship SDR, which adopts linear regression model to utilize determination coefficient R 2 defined weight where d j represents the dynamic behavior data of node j, s i represents the static attribute vector of node i, R 2 (d j , s i ) represents the determination coefficient of linear regression model with s i as independent variable and d j as dependent variable.

[0019] Bidirectional dependency relationship DDR, which adopts dynamic time warping to calculate the optimal matching distance between sequences and define weight where d i and d j represent the dynamic behavior sequences of node i and node j respectively, DTW(d i , d j ) represents the dynamic time warping distance between sequences d i and d j , μ and σ represent the mean and standard deviation of all node pair distances respectively.

[0020] and deep social relationship DSR, which adopts Granger causality test to evaluate the prediction contribution of dynamic sequences to static attribute changes and define weight where p-value ij represents the significance p-value of Granger causality test.

[0021] Further, the social attribute modeling step adopts social relationship tensor and obtains embedding vector through time series graph convolution network, wherein the social relationship tensor is represented as: where represents the fused social relationship weight between node i and node j at time t, respectively represent the normalized static social relationship weight, dynamic behavior relationship weight, bidirectional dependency relationship weight and deep social relationship weight.

[0022] Further, the behavior deviation score adopts Z-score calculation and is represented as: where represents the behavior deviation score of user u at time t, represents the dynamic behavior vector of user u at time t, μ u represents the historical behavior mean of user u, σ u represents the historical behavior standard deviation of user u.

[0023] Furthermore, the neighborhood consistency index is used to measure the similarity between a user's social behavior and the behavior of users in its neighborhood, and when the consistency index drops below a preset threshold, it triggers the adaptive tightening of the access structure.

[0024] Furthermore, the dynamic adjustment in the access structure dynamic generation step includes at least one of adding constraint attributes, increasing attribute weights, reducing authorization thresholds, merging or splitting strategy nodes.

[0025] Furthermore, the key generation step employs an RLWE lattice cipher, satisfying:

[0026] b = a·s + e mod q

[0027] The social attribute vector is then embedded into the trapdoor generation stage of the private key, binding the user's private key to the user's social state, where a and b are public key components, s is the private key, e is the small error term, and q is the modulus.

[0028] Furthermore, the ciphertext generated by the encryption process includes an attribute vector, binding information, an access structure matrix, an RLWE random vector, and social state labels, represented as follows:

[0029] C = {M, ρ, attribute binding vector, RLWE random vector, social state label}

[0030] Where M represents the LSSS access matrix, and ρ is the function that maps matrix rows to attributes.

[0031] Furthermore, the revocation and verification steps include:

[0032] Revocation proofs are generated based on Merkle trees, where the user state is committed to by the leaf nodes:

[0033]

[0034] The verifier reconstructs the root hash using path proof to confirm that the user has not been revoked, and the session key is reconstructed using the LSSS structure as follows:

[0035]

[0036] UID u A unique identifier representing user u. Let S represent the social state of user u at time t, and let C be the set of attributes that satisfy the access policy. i and SK i These are the ciphertext component and the user's private key component related to attribute i, respectively, and λ is the LSSS reconstruction coefficient;

[0037] And based on fingerprint chains, it tracks changes in social state, including generating immutable log entries as follows:

[0038]

[0039] and write the block audit ledger, and generate social fingerprint hash chain as:

[0040]

[0041] Wherein CT represents the ciphertext, represents the attribute set of user u at time t, OpType represents the operation type, and t represents the timestamp, represents the social fingerprint information of user u at time t, represents the hash chain value of the previous time.

[0042] The beneficial effects of the present application are:

[0043] (1) By introducing multi-dimensional social attribute modeling and real-time behavior analysis, the system can perceive the dynamic changes of user social state and behavior patterns, and automatically adjust the access strategy accordingly. This makes the access control no longer rely on static attributes, and can effectively deal with internal threats, abuse of authority and other security risks, and timely curb security threats caused by abnormal user state.

[0044] (2) By integrating static social relationships, dynamic behavior, two-way interaction and deep causal relationship, a model for comprehensively quantifying user social trust is constructed. This makes the access control decision no longer based on a single attribute in isolation, but considers multiple factors such as the user's long-term role in the group, recent behavior consistency, mutual trust relationship, etc., so that the authorization decision is more suitable for complex social collaboration scenarios, and is more fine-grained and intelligent.

[0045] (3) The Merkle tree is used to generate a verifiable revocation proof, without the need to re-encrypt the entire ciphertext or broadcast a lengthy revocation list, significantly reducing the overhead of revocation operations. Combined with the fingerprint chain structure based on the blockchain idea, the non-tamperability and traceability of all social state changes and access operation records are ensured, providing a reliable basis for security audit and enhancing the credibility of the system.

[0046] (4) Based on the ring learning with errors problem in lattice cryptography, the encryption core is constructed, replacing the traditional bilinear pair-based cryptography system, so that the method can resist attacks by future quantum computers. At the same time, the dynamic social attribute management is successfully combined with post-quantum cryptography technology, providing a feasible technical path for solving the problem of dynamic access control in the post-quantum era.

[0047] (5) By calculating the behavior deviation score and neighborhood consistency index in real time, the system can actively detect the abnormal deviation of user behavior and the difference from the behavior mode of the social group to which the user belongs. Once suspicious signs are found, the tightening of access strategies can be automatically triggered, realizing the transition from passive defense to active early warning, and significantly improving the system's control ability over internal risks.

[0048] Other advantages, objects, and features of the application will be apparent to those skilled in the art from the following specification, which is to be taken in an illustrative rather than in a limiting sense. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to make the purposes, technical solutions and advantages of the present application clearer, the preferred detailed description of the present application will be combined with the drawings as follows, in which:

[0050] Figure 1 A schematic diagram of the overall flow of the attribute-aware dynamic encryption access control method of the present application. DETAILED DESCRIPTION

[0051] The embodiments of the present application are described below through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the disclosure. The present application can also be implemented or applied through other different embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be noted that the diagrams provided in the following examples only illustrate the basic concept of the present application in a schematic manner, and the following examples and features in the examples can be combined with each other without conflict.

[0052] The drawings are only used for illustrative explanation, and the representation is only a schematic diagram, not a physical diagram, and cannot be understood as a limitation of the present application; in order to better illustrate the embodiments of the present application, some components in the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product; for those skilled in the art, it is understandable that some known structures and their descriptions in the drawings may be omitted.

[0053] The same or similar reference numerals in the drawings of the embodiments of the present application correspond to the same or similar components; in the description of the present application, it is understood that if the orientations or positional relationships indicated by the terms "upper", "lower", "left", "right", "front", "back" and the like are based on the orientations or positional relationships shown in the drawings, they are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore the terms describing the positional relationship in the drawings are only used for exemplary illustration, and cannot be understood as a limitation on the present application, for those skilled in the art, the specific meanings of the above terms can be understood according to the specific circumstances.

[0054] Figure 1 The present application is a kind of attribute-aware dynamic encryption access control method overall flow schematic diagram.

[0055] 1. Social attribute modeling

[0056] The system collects multi-channel social relationship data of the user, and constructs a social attribute tensor for describing the social state of the user. In order to quantitatively describe the social dependence relationship between entities, four types of attributes are proposed - behavior interaction channels, which constitute a complete relationship set: static-static channel (SSR), static-dynamic channel (SDR), dynamic-dynamic channel (DDR) and dynamic-static channel (DSR). These four types of channels cover identity coupling, identity constraint on behavior, behavior cooperation and behavior reverse driving, which ensure the global description of the social dependence relationship between nodes.

[0057] Firstly, the SSR channel is used to measure the cooperation of long-term identity characteristics of entities. For any two nodes i, j, based on their static attribute vectors s i ,s j The mutual information MI(s i ,s j ) and the historical co-occurrence frequency p ij are calculated, and a weighted fusion index is defined:

[0058]

[0059] The numerator is used to fuse the statistical dependence and frequency evidence, and the denominator suppresses the weight bias caused by too large difference between the two, and finally reflects the coupling strength between long-term identities.

[0060] Secondly, the static-dynamic relationship (SDR) measures the explanatory power of identity to behavior. A linear regression model is used and the determination coefficient R2 Define weights, mainly because R 2 can reflect the degree of explanation of static properties to the variance of dynamic behavior, with clear interpretability and high robustness. R 2 The higher the value, the stronger the constraint of identity attributes on behavior characteristics, which can be used to quantify the influence of identity stability on behavior changes.

[0061]

[0062] Third, the DDR channel is used to measure the timing similarity and co-evolution between behavior sequences. Since the node behavior sequence often has time offset and nonlinear change, dynamic time warping (DTW) is used to calculate the optimal matching distance between sequences. DTW can flexibly align sequences on the time axis, thus accurately reflecting the synchronization of different node behaviors in the time dimension. Finally, the distance result is exponentially normalized to measure the co-evolution strength of behavior.

[0063]

[0064] where μ,σ are the mean and standard deviation of the distance of all nodes. This channel can reveal potential co-evolution behavior patterns or abnormal diffusion paths, providing a timing reference for dynamic strategy adjustment.

[0065] Finally, the DSR channel is used to depict the causal influence of dynamic behavior on static state (such as role change, permission upgrade). Granger causality test is used to evaluate the prediction contribution of dynamic sequence to static attribute change. The smaller the significance p-value, the stronger the driving effect of dynamic behavior on static state. Therefore, the p-value is used as an inverse indicator of causal strength to define the weight of behavior on identity evolution.

[0066]

[0067] After completing the quantitative modeling of the four types of channels, they are considered as independent and complementary social weights, and a unified relationship representation is constructed by directly using linear superposition. To ensure the comparability between different channels, first, the weight matrix W (·) is row-normalized to map its value range to [0,1] and can be interpreted as a conditional influence probability, denoted as Subsequently, the four types of normalized matrices are directly summed up to obtain the unified fusion tensor:

[0068]

[0069] After construction, temporal graph embedding is performed on the tensor. The embedding model can be any temporal graph representation learning model, such as a temporal graph convolutional network or a temporal attention network. The embedding output is a social attribute vector V corresponding to the user's social state. social Its dimensions can be set according to system requirements.

[0070] 2. Access Structure Generation

[0071] Based on social attribute vector V social The system automatically generates a Linear Secret Share (LSSS) access structure. The system executes:

[0072] 2.1 Map social attributes to an attribute vector space and assign weights based on indicators such as salience, activity, and credibility of the attributes;

[0073] 2.2 Construct the LSSS matrix M and its corresponding attribute set ρ so that access control can be determined by the satisfaction of social attributes;

[0074] 2.3 Establish an adjustable parameter set Θ for attribute weights and thresholds, which will be used to dynamically adjust the access strategy in the future.

[0075] The access structure can be represented as a triple (M, ρ, Θ).

[0076] 3. Dynamic reinforcement and adjustment

[0077] To achieve real-time and adaptive access control, this invention dynamically strengthens the access structure. This includes:

[0078] 3.1 Calculation of Behavioral Deviation Scores

[0079] The dynamic behavior vector of the user at time t Calculate the standardized deviation:

[0080]

[0081] when When the access threshold is exceeded, a tightening operation is performed on the access structure, which may include raising the threshold or increasing the number of required attributes.

[0082] 3.2 Calculation of Neighborhood Consistency Index

[0083] Consistency is assessed based on the similarity of features among neighboring nodes. When consistency falls below a lower threshold, the weights of key attributes in the access structure are increased, or an access threshold increase operation is performed.

[0084] 3.3 Local Adjustment of Access Structure

[0085] Based on real-time risk assessment, nodes in the access structure are merged, split, or weighted and redistributed to achieve adaptive updates of security policies.

[0086] 4. Key generation (Ring Learning with Errors (RLWE) lattice cryptography framework)

[0087] The system generates a master key and public key using the RLWE lattice cryptography framework:

[0088] 4.1 Choose parameters q, ring dimension n, and error distribution χ;

[0089] 4.2 Randomly generate vector a, sample error vector e;

[0090] 4.3 Generate public key (a, b) according to the RLWE relation b = a·s + e mod q.

[0091] In the trapdoor construction process, the user's social attribute vector V social is embedded into the private key generation constraints, resulting in a user private key that is bound to its social state. When the social state changes, the private key will be invalid or need to be regenerated.

[0092] 5. Encryption process

[0093] The encryption side performs attribute-based encryption under the LSSS access structure. The execution steps include:

[0094] 5.1 According to the access structure matrix M, the plaintext is divided into fragments or a shared vector is constructed;

[0095] 5.2 Generate a corresponding attribute binding vector for each attribute;

[0096] 5.3 Sample an RLWE random vector for constructing lattice encryption components;

[0097] 5.4 Write the social state label (e.g. timestamp, state hash) into the ciphertext.

[0098] The ciphertext components can be represented as:

[0099] C = {M, ρ, attribute binding vector, RLWE random vector, social state label}

[0100] 6. Decryption and verification

[0101] When the user submits a decryption request, the system performs:

[0102] 6.1 Check whether the current social attribute vector satisfies the access structure (M, ρ, Θ);

[0103] 6.2 If it satisfies, use the user's private key to perform LSSS inverse operation to recover the shared value and decrypt the plaintext;

[0104] 6.3 If the user's social state does not satisfy the access structure, decryption is refused.

[0105] Since the private key is bound to the social state, even if the private key is leaked, as long as the social state is not consistent, the attacker cannot decrypt.

[0106] 7. Revocation with verifiable proof

[0107] To support real-time revocation, the present application adopts:

[0108] 7.1 Merkle tree revocation proof

[0109] The revocation state is stored in a structured manner through a Merkle hash tree, and the leaf node commits the user state:

[0110]

[0111] The verifier proves the path Reconstruct the root hash to confirm that the user is not listed in the revocation set, achieving verifiability and anti-forgery of the revocation state. Among them is a discrete Gaussian noise that ensures the indistinguishability and privacy protection of the key. The user needs to satisfy the current social policy tree Its session key is reconstructed through the LSSS structure:

[0112]

[0113] If the user's attribute set no longer satisfies the policy condition or is marked by the revocation hash chain, the key recovery fails, ensuring the synchronization of access control and social state.

[0114] 7.2 Fingerprint chain records social state evolution

[0115] In addition, to achieve continuous traceable security audit, the system generates an unforgeable log item for each operation:

[0116]

[0117] The log items are written into the blockized audit ledger in turn, achieving verifiable backtracking of access history and responsibility tracking. The social fingerprint hash chain

[0118]

[0119] Further ensures the anti-replay and non-repudiation across time windows, making the key, access, and audit form a closed loop in logic. When the user state is detected to be invalid or the attribute is revoked, the system automatically generates a revocation proof and refuses any decryption operation.

[0120] Embodiment one: dynamic access control for internal threat detection

[0121] This embodiment shows the application of the present application in the scenario of preventing internal personnel from abusing their rights.

[0122] Workflow:

[0123] Initialization and attribute modeling: The system constructs a social attribute tensor for user A in the R&D department. Static social relationships are collected, such as the formal position "senior researcher" and the project team he belongs to; dynamic behavior relationships are collected, such as his recent code access frequency and log query frequency; bidirectional dependency relationships are analyzed, such as the code review interaction pattern between him and the project leader; and deep social relationships are evaluated, such as whether his code submission behavior significantly affects his project role weight. After all the relationships are quantified, they are fused into a social attribute tensor, and the current social attribute vector of A is obtained through a time series graph convolution network.

[0124] Policy generation and encryption: The data owner sets an access policy for a core design document. The policy requires that the user must simultaneously satisfy the two attributes "project team = Pioneer" and "behavior consistency score > 0.8". The system generates an LSSS access matrix based on this policy, and uses an RLWE lattice encryption algorithm to encrypt the document. The ciphertext contains the access policy matrix and the social state label.

[0125] Real-time monitoring and dynamic adjustment: The system continuously monitors the behavior of A. One day, it is detected that A has been accessing sensitive data modules unrelated to his current task at an abnormally high frequency outside working hours, resulting in a significant deviation of his behavior deviation score r_u^(t) from the threshold. At the same time, the neighborhood consistency index calculation shows that his behavior pattern deviates significantly from other members of the same team. The access control engine automatically triggers policy adjustment, dynamically strengthening the access policy to require simultaneous satisfaction of "project team = Pioneer", "behavior consistency score > 0.8", and "last security authentication status = valid".

[0126] Decryption attempt and access denial: Although A's static attributes still satisfy the original policy, his current social attribute vector has failed to meet the strengthened access structure with the added security authentication requirement due to his abnormal behavior. When A attempts to decrypt the design document, the system verification fails and the decryption request is denied. At the same time, this abnormal access behavior is recorded in the tamper-proof fingerprint chain.

[0127] Audit and traceability: The security administrator can clearly trace the time of A's abnormal access attempt, the type of operation, and the complete log of the system's automatic execution of policy adjustment by querying the fingerprint chain, providing a reliable basis for event analysis.

[0128] Example Two: Dynamic trust management in cross-domain data sharing

[0129] This example demonstrates how the present invention can be used for fine-grained access control based on dynamic trust in scenarios requiring cross-security domain collaboration.

[0130] Workflow:

[0131] Attribute mapping and key generation: External partner user B needs to access data within the domain. The system maps B's external identity attributes to temporary social attributes within the domain, such as "cooperation enterprise = Gamma Company" and "cooperation project = Sky Plan". At the same time, the system continuously collects B's dynamic interaction behavior through API interface, such as API call success rate and data request frequency, for calculating its dynamic behavior relationship weight. The system generates B's social attribute vector based on his static mapping attributes and dynamic behavior data, and accordingly generates a private key bound to his current social state using the RLWE trapdoor function.

[0132] Setting hierarchical access policy: The data owner sets a hierarchical access policy for shared data. The basic data requires the attribute "cooperation project = Sky Plan" to access. The core data requires a more stringent policy, such as requiring both "cooperation project = Sky Plan" and "recent interaction trustworthiness > 0.9".

[0133] Dynamic permission upgrade and downgrade: In the early stage of cooperation, B behaves well and the interaction trustworthiness index is good, and his social attribute vector meets the access policy of core data, so he can normally decrypt. During the cooperation process, the system detects that a security event occurs in B's domain, although it does not directly affect B, as a risk prevention measure, the system automatically adjusts B's "recent interaction trustworthiness" weight. This adjustment causes B's social attribute vector to no longer meet the access policy of core data, and his access permission is automatically reduced to only access basic data.

[0134] Verifiable revocation: After confirmation, the security event of B's enterprise has been properly handled, and the system restores his "recent interaction trustworthiness" weight. However, if the cooperation terminates, the data owner only needs to add B's identifier to the leaf node of the Merkle revocation tree and update the root hash. When B tries to access again, his state proof cannot pass the latest Merkle root verification, and the access will be denied, realizing efficient and verifiable revocation.

[0135] Finally, it should be pointed out that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and all should be covered in the scope of the claims of the present application.

Claims

1. An attribute-aware dynamic encrypted access control method, characterized in that: The method includes: The social attribute modeling steps involve collecting users' static social relationships, dynamic behavioral relationships, two-way interaction relationships, and deep social relationships, and constructing a multi-channel social attribute tensor based on these relationships. Vector embedding is then performed on the tensor to obtain a social attribute vector for access control. The access structure dynamic generation step involves constructing a linear secret sharing structure (LSSS) access matrix based on the social attribute vector, and performing real-time reinforcement and adjustment on the access structure according to behavioral deviation scores, neighborhood consistency indicators, and adaptive thresholds. The key generation steps involve constructing a public key and a master key based on lattice cryptographic ring learning with error (RLWE), and generating a user private key bound to the social state based on the user's social attribute vector and access structure. The encryption process involves performing attribute-based encryption on the plaintext based on the access structure matrix to generate ciphertext components related to social attributes and recording the social attribute binding vector. The decryption process involves verifying whether the user's social attribute vector satisfies the access structure when the user initiates a decryption request. If so, the plaintext is recovered based on the user's private key and the access structure matrix. The revocation and verification steps generate revocation proofs based on the Merkle tree structure, perform verifiable revocation on invalid social attributes or users, and use the fingerprint chain structure to verify whether the social state has been tampered with.

2. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The social attributes in the social attribute modeling step include: Static social relationship (SSR) is based on the static attribute vector s of any two nodes i and j. i ,s j Calculate mutual information MI(s) i ,s j and historical co-occurrence frequency ρ ij And define weighted fusion index Dynamic behavior relationship SDR, which uses a linear regression model with the coefficient of determination R 2 Define weights Where d j s represents the dynamic behavior data of node j. i R represents the static attribute vector of node i. 2 (d j ~s i ) indicates s i For independent variable, d j The coefficient of determination for a linear regression model of the dependent variable; Bidirectional dependency relation DDR uses dynamic time warping to calculate the optimal matching distance between sequences and defines weights. Where d i and d j Let DTW(d) represent the dynamic behavior sequences of nodes i and j, respectively. i ,d j ) represents the sequence d i and d j The dynamic time-warped distance between them, where μ and σ are the mean and standard deviation of the distances between all node pairs, respectively; And Deep Social Relationships (DSR), which uses Granger causality tests to assess the predictive contribution of dynamic sequences to changes in static attributes and defines weights. Where p-value ij The p-value represents the significance of the Granger causality test.

3. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The social attribute modeling step uses a social relation tensor and obtains embedding vectors through a temporal graph convolutional network, wherein the social relation tensor is represented as: in This represents the weight of the fused social relationship between node i and node j at time t. These represent the normalized weights of static social relationships, dynamic behavioral relationships, bidirectional dependency relationships, and deep social relationships, respectively.

4. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The behavioral deviation score is calculated using the Z-score and expressed as follows: in This represents the behavioral deviation score of user u at time t. μ represents the dynamic behavior vector of user u at time t. u σ represents the historical mean of user u's behavior. u This represents the standard deviation of user u's historical behavior.

5. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The neighborhood consistency index is used to measure the similarity between a user's social behavior and the behavior of users in its neighborhood. When the consistency index drops below a preset threshold, the access structure is triggered to tighten adaptively.

6. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The dynamic adjustment in the access structure dynamic generation step includes at least one of adding constraint attributes, increasing attribute weights, lowering authorization thresholds, merging or splitting strategy nodes.

7. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The key generation step uses an RLWE lattice cipher, satisfying the following: b = a·s + e mod q The social attribute vector is then embedded into the trapdoor generation stage of the private key, binding the user's private key to the user's social state, where a and b are public key components, s is the private key, e is the small error term, and q is the modulus.

8. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The ciphertext generated by the encryption process includes an attribute vector, binding information, an access structure matrix, an RLWE random vector, and social state labels, represented as follows: C = {M, ρ, attribute binding vector, RLWE random vector, social state label} Where M represents the LSSS access matrix, and ρ is the function that maps matrix rows to attributes.

9. The attribute-aware dynamic encrypted access control method according to claim 1, characterized in that: The revocation and verification steps include: Revocation proofs are generated based on Merkle trees, where the user state is committed to by the leaf nodes: The verifier reconstructs the root hash using path proof to confirm that the user has not been revoked, and the session key is reconstructed using the LSSS structure as follows: UID u A unique identifier representing user u. Let S represent the social state of user u at time t, and let C be the set of attributes that satisfy the access policy. i and SK i These are the ciphertext component and the user's private key component related to attribute i, respectively, and λ is the LSSS reconstruction coefficient; And based on fingerprint chains, it tracks changes in social state, including generating immutable log entries as follows: And write it into the blockchain audit ledger, and generate a social fingerprint hash chain as follows: CT represents encrypted text. This represents the attribute set of user u at time t, where OpType represents the operation type and t represents the timestamp. This represents the social fingerprint information of user u at time t. This represents the hash chain value at the previous moment.

Images