Dynamic access control management method and system based on zero trust architecture

Abstract

The present invention proposes a dynamic access control management method and system based on a zero-trust architecture, which includes: obtaining original feature data, constructing feature data, and assigning labels to the feature data; generating an identity chain based on the feature data and labels, encrypting and signing and authenticating the identity chain, and obtaining an initial authentication result set; performing an adversarial sample attack on the initial authentication result set to increase the detection capability of the authentication result set and prevent forged identity attacks, and generating a final authentication result set; obtaining user identity credibility based on the authentication result set, obtaining resource information requested by the user, dynamically adjusting the access rights requested by the user based on the user identity credibility, and outputting user access rights; performing real-time identity chain monitoring, updating access rights through an identity credibility update model based on dynamically calculated user identity credibility and real-time access records, and triggering further security measures or permission adjustments through anomaly detection.

Description

Technical Field

[0001] The present invention belongs to the field of dynamic access control management, and in particular relates to a dynamic access control management method and system based on a zero-trust architecture. Background Art

[0002] In current network security architectures, access control is a crucial component of ensuring system security. Traditional access control methods primarily rely on role-based access control and attribute-based access control, which typically manage user access rights to resources through predefined rules. However, with the prevalence of emerging technologies such as cloud computing, the Internet of Things, and remote work, user access environments have become increasingly complex, and traditional access control methods have struggled to adapt to dynamically changing security requirements. To address this issue, zero-trust architecture has become a key development direction in the security field in recent years. Zero-trust architecture emphasizes strict identity authentication and permission verification for all access requests, regardless of whether they originate from within or outside the network, to minimize the attack surface. In a zero-trust architecture, access control typically relies on multi-factor authentication and AI-based anomaly detection methods. Multi-factor authentication verifies user identity through methods such as passwords, one-time verification codes, and biometrics, while AI-based anomaly detection methods utilize machine learning algorithms to analyze user access behavior and detect anomalous access patterns.

[0003] However, existing zero-trust access control methods still have many shortcomings.

[0004] First, role-based access control and attribute-based access control rely on static rules and are difficult to dynamically adjust access rights. In practical applications, they are prone to over-authorization or mistaken rejection of legitimate users.

[0005] Secondly, multi-factor authentication mainly relies on explicit credentials, such as passwords, fingerprints, facial recognition, etc., but these credentials are easy to steal, forge, or bypass. Attackers can steal user identity information through phishing attacks, malware, or forged biometrics, thereby bypassing the identity authentication mechanism.

[0006] Again, although AI-based access control systems can detect abnormal access behavior to a certain extent, these systems are vulnerable to adversarial sample attacks. Attackers can generate forged access features to cause the AI ​​system to mistakenly identify the attacker as a legitimate user, thereby bypassing the security defense mechanism.

[0007] Finally, traditional access control systems typically only perform authentication when an access request is initiated, and do not continuously monitor user behavior during the access process. Therefore, once an attacker gains access, they can remain inside the system for a long time, increasing the risk of data leakage.

[0008] To sum up, the existing zero-trust access control methods still have problems such as static rules being difficult to adapt to dynamic security environments, identity authentication relying on explicit credentials that are easily forged, AI-based access control systems being easily bypassed by adversarial attacks, and a lack of continuous identity monitoring and dynamic adjustment mechanisms. Therefore, a more secure, dynamic, difficult to forge, and attack-resistant access control method is needed to adapt to the ever-changing network security environment. Summary of the Invention

[0009] The purpose of this invention is to propose a dynamic access control management method and system based on a zero-trust architecture, which improves the security and reliability of the access control system by constructing unforgeable user identity information and combining artificial intelligence technology to combat the forgery behavior of attackers.

[0010] In order to achieve the above object, a first aspect of the present invention provides a dynamic access control management method based on a zero trust architecture, the method comprising the following steps:

[0011] Acquiring original feature data, constructing feature data, and assigning labels to the feature data; wherein the original feature data is obtained through preprocessing;

[0012] Generate an identity chain based on the feature data and the tag, encrypt and sign the identity chain, and authenticate it to obtain an initial authentication result set; wherein the authentication is used to reject any identity chain that cannot pass the match;

[0013] Performing an adversarial sample attack on the initial authentication result set to increase the detection capability of the authentication result set and prevent forged identity attacks, and generating a final authentication result set;

[0014] The adversarial sample attack achieves adversarial sample detection by calculating identity information perturbation metrics, weighted perturbation evaluation, and dynamic regularization terms.

[0015] Obtaining user identity credibility based on the authentication result set, obtaining resource information requested by the user, dynamically adjusting the access rights requested by the user based on the user identity credibility, and outputting the user access rights;

[0016] Real-time identity chain monitoring, based on dynamically calculated user identity credibility and real-time access records, updates access rights through the identity credibility update model, and triggers further security measures or permission adjustments through anomaly detection;

[0017] The identity credibility update model dynamically adjusts the user's identity credibility score according to the user's behavior history, current access record and system status, and controls the access rights of each user based on a preset identity credibility threshold.

[0018] Preferably, the preprocessing includes data cleaning, normalization, denoising and label generation.

[0019] Preferably, generating an identity chain based on the feature data and the tag, encrypting and signing the identity chain, and authenticating the identity chain to obtain an initial authentication result set specifically includes:

[0020] Build a corresponding identity chain based on each feature data through a mapping function as a unique identity identifier;

[0021] Encrypted identity chain to obtain encrypted identity information;

[0022] By signing the encrypted identity information through the signature algorithm, the corresponding signature result is obtained;

[0023] A weighted regularization term is designed to adjust the matching degree between the encrypted identity information and the signature, reducing overmatching of low-trust identity information.

[0024] The corresponding identity chain, the corresponding signature result and the weighted regularization item are combined into an authentication result to construct a final authentication result set.

[0025] Preferably, the design weighted regularization term includes:

[0026] Get the matching degree between the identity chain and its corresponding signature result;

[0027] Based on the matching degree and the validity indicator function of the identity information, a weighted regularization term is obtained by weighted summation;

[0028] The corresponding identity chain, the corresponding signature result and the weighted regularization item are passed through a verification function to generate an authentication result.

[0029] Preferably, performing an adversarial sample attack on the initial authentication result set to increase the detection capability of the authentication result set and prevent forged identity attacks, and generating a final authentication result set, specifically includes:

[0030] By comparing the difference between the predicted result of each identity information obtained in the preset authentication model and the original authentication result, it is determined whether the information is affected by the adversarial sample attack and the perturbation measure of the corresponding identity information is generated;

[0031] Performing weighted perturbation evaluation analysis on the perturbation metric corresponding to the identity information to generate a weighted perturbation metric;

[0032] Regularization terms are introduced to penalize abnormal disturbances during the perturbation identification process, thereby improving the ability to filter adversarial samples.

[0033] Combining the weighted perturbation measure and the regularization term, we generate an adversarial sample detection decision function to determine whether to accept a piece of identity information:

[0034] For identity information whose sum of the weighted perturbation metric and the regularization term exceeds a preset threshold, the identity information will be rejected; otherwise, the identity authentication process will continue.

[0035] Preferably, the determining whether the information is affected by the adversarial sample attack is specifically:

[0036] If the difference between the predicted result and the original authentication result exceeds the set threshold, it is considered that the identity information may be vulnerable to adversarial attacks;

[0037] The weighted perturbation estimation analysis weights the perturbation metric according to the sensitivity of the features.

[0038] Preferably, the obtaining of user identity credibility based on the authentication result set, obtaining resource information requested by the user, dynamically adjusting the access rights requested by the user based on the user identity credibility, and outputting the user access rights specifically include:

[0039] Use dynamic credibility update rules to update user identity credibility and generate dynamic credibility;

[0040] Based on the sensitivity of the resource request and the user's behavioral activity, the dynamic credibility is comprehensively scored to further adjust the identity credibility;

[0041] Adopting a preset dynamic access rights calculation model, it determines whether to allow a user to access a resource based on different dynamic credibility and resource request sensitivity, and generates dynamically calculated user access rights.

[0042] The user access rights include: a specific access rights level and an access rights vector.

[0043] Preferably, the dynamic credibility update rule calculates the updated credibility by combining the historical credibility and the currently newly evaluated credibility in a weighted average manner to obtain the dynamic credibility; the dynamic credibility changes according to the time of the identity verification, and the change of the identity verification is based on the updated version of the authentication result obtained by the adversarial sample detection;

[0044] The access rights calculation rules of the preset dynamic access rights calculation model are as follows:

[0045]

[0046] Where A(t) is the user access permission θ high and θ loware high confidence threshold and low confidence threshold, which are used to determine the upper and lower limits of access; scale(C(t),θ low ,θ high ) is mapped to the range [0,1] according to the dynamic identity credibility C(t), indicating partial access rights; C(t) is the dynamic identity credibility.

[0047] Preferably, the real-time identity chain monitoring, based on the dynamically calculated user identity credibility and real-time access records, updates access rights through the identity credibility update model, and triggers further security measures or permission adjustments through anomaly detection, specifically includes:

[0048] Adjust the user's identity credibility score based on the user's historical behavior, real-time request records, and current access rights; if the user's behavior is normal, the identity credibility score will increase; if abnormal behavior occurs, the score will decrease;

[0049] Based on the real-time monitored identity credibility score, the access permission adjustment threshold is dynamically calculated. If the identity credibility score exceeds the adjusted access permission adjustment threshold, the user is allowed to access the corresponding resources. At the same time, during the real-time security adjustment process, anomaly detection is performed based on the current identity credibility score and the average of the identity credibility scores. If the difference between the user's identity credibility score and the average of the identity credibility scores exceeds the preset threshold, the user behavior is considered abnormal, triggering further security monitoring or permission adjustment.

[0050] In a second aspect of the present invention, a dynamic access control management system based on a zero-trust architecture is provided, the system comprising:

[0051] An implicit identity feature acquisition unit, configured to acquire original feature data, construct feature data, and assign labels to the feature data; wherein the original feature data is obtained through preprocessing;

[0052] An identity chain generation unit is configured to generate an identity chain based on the feature data and the tag, perform cryptographic signing and authentication on the identity chain, and obtain an initial authentication result set; wherein the authentication is used to reject any identity chain that fails to pass the match;

[0053] An identity chain enhancement unit, configured to perform an adversarial sample attack on the initial authentication result set, thereby increasing the detection capability of the authentication result set and preventing forged identity attacks, and generating a final authentication result set;

[0054] The adversarial sample attack achieves adversarial sample detection by calculating identity information perturbation metrics, weighted perturbation evaluation, and dynamic regularization terms.

[0055] A user access permission setting unit is configured to obtain user identity credibility based on the authentication result set, obtain resource information requested by the user, dynamically adjust the access permission requested by the user based on the user identity credibility, and output the user access permission;

[0056] The access rights optimization unit is used to monitor the identity chain in real time. Based on the dynamically calculated user identity credibility and real-time access records, it updates access rights through the identity credibility update model and triggers further security measures or permission adjustments through anomaly detection.

[0057] The identity credibility update model dynamically adjusts the user's identity credibility score according to the user's behavior history, current access record and system status, and controls the access rights of each user based on a preset identity credibility threshold.

[0058] The beneficial technical effects of the present invention are at least as follows:

[0059] In response to the shortcomings of existing zero-trust access control methods, the present invention proposes a zero-trust access control method based on implicit identity chain and adversarial sample detection. By constructing unforgeable user identity information and combining artificial intelligence technology to combat the attacker's forgery behavior, the security and reliability of the access control system are improved.

[0060] The first innovation of this invention is its implicit identity chain authentication method. This method constructs a unique identity by collecting information such as the user's device fingerprint, behavioral characteristics, environmental characteristics, and access sequence. This method then uses a hash algorithm to calculate the integrity of the identity chain, ensuring that the identity information cannot be tampered with. Because the implicit identity chain authenticates based on the user's multi-dimensional characteristics, even if an attacker steals the user's password, one-time verification code, or forges biometric information, they cannot simulate the user's complete identity characteristics, effectively preventing identity forgery attacks.

[0061] The second innovation of this invention is an access control method based on adversarial sample detection. This method enhances the AI ​​access control model's ability to identify forged identity features by introducing adversarial training and robustness detection techniques. By generating adversarial samples during the training process, the robustness of the access control system is improved, enabling it to identify maliciously forged identity data. It then combines methods such as feature compression and entropy analysis to detect whether access requests contain adversarial sample features, ensuring that attackers cannot bypass the authentication mechanism by forging identity information.

[0062] In addition, the present invention also introduces a continuous identity monitoring and dynamic permission adjustment mechanism, which continuously monitors the user's behavior patterns during the access process. If abnormal behavior is detected, such as frequent device changes in a short period of time or an abnormal increase in access resources, the system will dynamically adjust the user's access rights or even block access requests, thereby effectively reducing security risks. By combining implicit identity chains and adversarial sample detection technology, the present invention realizes a more secure and reliable zero-trust access control method, which can effectively solve the shortcomings of existing access control technologies in dynamic security environments and provide more complete security protection for access control in scenarios such as cloud computing, remote office, and the Internet of Things. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative effort.

[0064] Figure 1 This is a flow chart of the dynamic access control management method based on zero trust architecture disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The following describes embodiments of the present invention in detail. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present invention and are not to be construed as limiting the present invention.

[0066] Example 1

[0067] like Figure 1 As shown, the embodiment of the present invention provides a dynamic access control management method based on a zero trust architecture, the method comprising the following steps:

[0068] S1. Obtain original feature data, construct feature data, and assign labels to the feature data; wherein the original feature data is obtained through preprocessing.

[0069] Specifically, input: original feature dataset where x i It is the raw feature data collected from sensors or devices. These data usually contain noise, missing values, outliers, etc., and may exist in different data formats and scales.

[0070] Furthermore, for missing values ​​in the data, the present invention uses a weighted average method based on neighborhood data to fill in the missing values. The missing data is filled by calculating the weighted average value within the neighborhood of the data point, avoiding the errors that may be caused by simple mean filling. The calculation formula is:

[0071]

[0072] in, is feature x j Neighborhood dataset, w k is the weight of each neighborhood point, x jk is the value of the neighborhood data point.

[0073] Furthermore, to ensure the consistency of different feature scales, a normalization method based on adaptive weights is adopted. Feature normalization eliminates the scale differences between different features and preserves their relative relationships by linearly mapping the feature values ​​to the interval [0,1]. The specific formula is as follows:

[0074]

[0075] Among them, w j is feature x j The importance weight of the feature. This weight can be set based on the training effect or domain knowledge of the feature.

[0076] Furthermore, the noise in the data may come from sensor errors or environmental interference, so denoising is required. The present invention uses a Gaussian process regression model to estimate noise, obtaining a more accurate numerical estimate. Through this method, the noise can be removed, preserving the true information of the data. Mathematically, it can be expressed as:

[0077]

[0078] in, is the cleaned dataset, is the data after denoising.

[0079] Furthermore, the label y i Typically, labels are not directly obtained from the data collection system, but rather through a subsequent label generation process. For example, in some scenarios, labels can be obtained through expert annotation, where after collecting the raw data, human experts assign a label to each piece of data based on domain knowledge.

[0080] Output: preprocessed dataset The dataset after data cleaning, normalization, denoising and label generation.

[0081] S2. Generate an identity chain based on the feature data and the tag, encrypt and sign the identity chain, and authenticate it to obtain an initial authentication result set; wherein the authentication is used to reject any identity chain that cannot pass the match.

[0082] Specifically, input: the data preprocessing result of the previous step in is the characteristic data, y i For label.

[0083] operate:

[0084] Furthermore, we build an identity chain: Based on a mapping function Building an identity chain i The identity chain is used to represent the unique identity of a device or sensor.

[0085] Among them, the generation formula of the identity chain is:

[0086]

[0087] This step uses feature data and label y i Create a unique identity, which is very important for subsequent authentication.

[0088] Encryption and Signature: To ensure the security of identity chain information, first i The encrypted identity information is At the same time, the encrypted identity information is signed by the signature algorithm to obtain the signature s i The process adopts an improved encryption-signature joint strategy to ensure the security of identity information during transmission.

[0089] The combined formula for encryption and signature:

[0090]

[0091] Among them, ε1 is the identity chain encryption function, is the signature function, s i This process avoids the redundant use of multiple encryption operations and ensures the integrity and tamper-proofness of identity information through a single signature function.

[0092] Furthermore, during the identity chain verification process, this paper proposes an innovative regularization strategy. This regularization term is specifically used to adjust the influence of different identity information on the authentication result. A weighting term is used to increase the weight of high-trusted identity information, thereby reducing overmatching of low-trusted identity information. This regularization term is designed as follows:

[0093]

[0094] Where: α is the regularization coefficient, which controls the matching degree between the encrypted identity information and the signature; β is the penalty coefficient for invalid identity information, is the validity indicator function of identity information. i When it is valid identity information, Otherwise, it is 0. This regularization term ensures that only highly reliable identity information is strictly matched during the authentication process, while invalid or suspicious identity information is suppressed, thereby improving the security and accuracy of the overall authentication.

[0095] Furthermore, identity chain verification and authentication result output: After the identity information is passed to the authentication system, the system uses the verification function Encrypted identity information and signatures i Compare and generate the final authentication result The authentication process is judged by the regularized matching function and any identity information that cannot pass the match is rejected.

[0096] The formula for the authentication process is:

[0097]

[0098] in, is the validation function, based on the weighted regularization term Compare and authenticate the identity information. Finally, the authentication result Indicates whether the identity information is valid.

[0099] Output: Authentication result set Indicates the authentication result of each piece of identity information.

[0100] S3. Perform an adversarial sample attack on the initial authentication result set to increase the detection capability of the authentication result set and prevent forged identity attacks, and generate a final authentication result set.

[0101] Specifically, in identity authentication systems such as the Internet of Things and biometrics, attackers may forge identity information through adversarial samples, thereby breaking through the identity authentication mechanism and obtaining illegal access rights. Adversarial examples are created by applying small, carefully designed perturbations to the original data, causing errors in the model's prediction results, which in turn leads to the failure of the identity authentication system. Therefore, preventing adversarial examples from forging identities is one of the core tasks of the present invention. In order to improve the robustness of the authentication system, the present invention proposes an adversarial sample detection method based on weighted perturbation measurement and dynamic regularization technology. This method significantly improves the system's ability to detect adversarial samples by utilizing an adaptive weighting mechanism, dynamically adjusting the sensitivity of the perturbation measurement, and combining regularization terms to suppress abnormal perturbations.

[0102] Input: Authentication result set from the second step output Among them Ci is the authentication result of identity information i, is the predicted value, This is the original identity authentication information.

[0103] Further, the operation:

[0104] Identity information perturbation detection: In this step, the present invention first performs perturbation detection on each piece of authentication information. Specifically, by comparing the difference between the predicted result obtained by the authentication model and the original authentication result of each piece of identity information, it is determined whether the information has been affected by the adversarial sample attack. Perturbation measurement The calculation is as follows:

[0105]

[0106] in, is the identity authentication result predicted by the model; is the original authentication result of the identity information; ∥·∥2 represents the L2 norm, which is used to measure the size of the disturbance.

[0107] if Exceeding the set threshold It is considered that the identity information may be vulnerable to adversarial attacks. is a pre-set maximum perturbation threshold. This threshold can be dynamically adjusted based on the specific application scenario of the model. The specific adjustment strategy will be described later.

[0108] Weighted perturbation evaluation: To further improve the detection capability of adversarial examples, this paper proposes a weighted perturbation evaluation method. This method combines multiple features in the identity information and assigns different weights to different parts of the perturbation. Specifically, different features in the identity information have different degrees of influence on the final authentication result, so it is necessary to weight the perturbation measurement based on the sensitivity of the feature. The weighted perturbation measurement formula is:

[0109]

[0110] in: and Represent the predicted value and original value of the jth feature in the i-th identity information respectively; w j is the weighting factor of feature j, reflecting the importance of this feature to the final authentication result. For features with greater influence, w j Take a larger value; for features with less influence, w j Then take the smaller value.

[0111] The weighted perturbation measure can improve the sensitivity to perturbations of important features and suppress overreaction to perturbations of unimportant features by adjusting the feature importance weights.

[0112] Dynamic Regularization: To further improve the detection accuracy of adversarial examples, we propose a dynamic regularization technique to penalize anomalies in perturbation detection. By introducing a regularization term, we can penalize abnormal perturbations during the perturbation detection process, further improving the ability to filter adversarial examples.

[0113] Specific regularization terms The form is:

[0114]

[0115] Among them, λ is the regularization coefficient, which is used to control the strength of the regularization term; γ j is the perturbation tolerance threshold for the jth feature, representing the maximum perturbation allowed for that feature in adversarial examples. This regularization term penalizes the squared error of the perturbation, ensuring that identity information that does not conform to the expected perturbation range is considered abnormal. The introduction of the dynamic regularization term effectively limits the magnitude of the perturbation, allowing the system to consider the degree of abnormality of the perturbation rather than just the perturbation size during detection, further improving its ability to resist adversarial examples.

[0116] Adversarial sample detection and identity verification: Combining the above perturbation metrics, weighted perturbation evaluation, and dynamic regularization terms, this paper proposes a final adversarial sample detection decision function to determine whether to accept a piece of identity information. and regularization term Exceeding the threshold If the identity information is not provided, the information will be rejected; otherwise, continue with the identity verification process.

[0117] The final adversarial sample detection decision function is:

[0118]

[0119] in: The final certification result; is a comprehensive threshold that combines the perturbation measure and the regularization term.

[0120] Further, output: final authentication result set Indicates the final authentication result of each piece of identity information. If an identity information is detected as an adversarial example, its authentication result is "rejected", otherwise the original authentication result is maintained.

[0121] S4. Obtain the user identity credibility based on the authentication result set, obtain the resource information requested by the user, dynamically adjust the access rights requested by the user based on the user identity credibility, and output the user access rights.

[0122] Specifically, from the third step: the authentication result set C in the adversarial sample detection = {C i}, where C i is the authentication result of each identity information i. At this point, represents the authentication result predicted by the model, and Represents the original identity authentication information. These authentication results C i Used to assess the credibility of a user's identity.

[0123] Resource information R requested by the user request , indicating the resource the user wants to access. This includes the resource type, sensitivity level, access frequency, etc.

[0124] Furthermore, the dynamic credibility change is calculated: using the dynamic credibility update rule, the historical credibility C(t-1) and the current newly evaluated credibility C are combined in a weighted average manner. new (t) to calculate the updated dynamic identity credibility C(t). Specifically, C t represents the dynamic identity credibility of the user at time t, which is obtained based on the adversarial sample detection C i Certification results evaluation. i Represents the authentication result of a single identity information, while C t is the overall credibility of the user. Here we need to explain that C(t) is only the i It is a dynamic representation that changes according to the time of identity authentication. The present invention can be simply understood as the dynamic identity credibility C(t) is the C obtained based on the adversarial sample detection. i The updated version of reflects the user's identity credibility at a specific time t. The formula is as follows:

[0125] C(t)=λ·C(t-1)+(1-λ)·C new (t) (12)

[0126] Where: C(t) is the dynamic identity credibility; λ is the weighting factor, which controls the historical credibility C(t-1) and the current evaluation result C new The relative weight of (t). C new (t) is the new credibility assessed based on real-time data such as the user's current behavior, operating mode, and external environment. The time decay factor δ(t) is used to further adjust the changing trend of credibility: over time, the impact of long-term behavior on credibility will gradually decrease, reducing the impact of outdated behavior on the current identity's credibility.

[0127]

[0128] Where: T is the time constant that controls the decay rate. A smaller T indicates a faster decay rate, and a larger T indicates that long-term behavior has a greater impact on credibility.

[0129] The updated dynamic credibility change rules are as follows:

[0130] C(t)=λ·C(t-1)+(1-λ)·C new (t)·δ(t) (14)

[0131] Furthermore, after calculating the dynamic identity credibility C(t), the present invention introduces a comprehensive scoring mechanism S(t) that combines multiple factors (such as user behavior, resource requests, etc.) to further adjust the identity credibility. The comprehensive score is used to provide a basis for the final identity credibility adjustment. The calculation formula for the comprehensive score S(t) is:

[0132] S(t)=α·C(t)+β·R request_level +γ·activity_frequency (15)

[0133] Among them, α, β, and γ are weight coefficients, which represent the influence of different dimensions on the comprehensive score. request_level : Indicates the sensitivity of the resource request. For example, requests for sensitive resources require higher credibility. activity_frequency: Indicates the user's behavioral activity. Frequent user activity may increase their credibility. The comprehensive score S(t) provides a basis for dynamically adjusting identity credibility, helping the system optimize access control based on actual conditions.

[0134] Furthermore, based on the updated dynamic identity credibility C(t), the system adopts a dynamic access rights calculation model. This model will determine whether to allow a user to access a resource based on different identity credibility and the sensitivity of the resource request. The specific access rights calculation rules are as follows:

[0135]

[0136] Among them, θ high and θ low are the high confidence threshold and the low confidence threshold, which are used to determine the upper and lower limits of access. scale(C(t),θ low ,θ high ): Based on the dynamic identity trustworthiness C(t), it is mapped to the range [0,1], representing partial access rights. This model enables the system to flexibly control the access rights of users based on their dynamic trustworthiness, ensuring that high-risk behaviors are strictly restricted.

[0137] Furthermore, in order to avoid excessive fluctuations in access rights, a regularization term is designed It is used to smooth the changes in access rights and ensure that the adjustment of permissions is not too drastic, thereby improving system stability. The calculation method of the regularization term is:

[0138]

[0139] Among them, λ reg is the regularization factor, which controls the smoothness. i represents the user behavior record at the i-th moment, A(t i ) is the access permission at time i. The regularization term ensures that the change in access permissions is not too drastic, thus ensuring a stable user experience.

[0140] Furthermore, the output is the dynamically calculated user access rights (A(t)), which is determined by the identity credibility and resource request information. The output access rights A(t) can be:

[0141] Specific access permission level, such as 1 for full access and 0 for denied access;

[0142] Permission vectors represent access permissions for different resource categories, such as data access permissions, system operation permissions, etc.

[0143] Through these steps, the system can dynamically adjust access rights based on the user's identity credibility and behavior, thereby ensuring the security and flexibility of the system while avoiding the adverse impact of frequent permission changes on user experience.

[0144] S5. Real-time identity chain monitoring. Based on the dynamically calculated user identity credibility and real-time access records, access rights are updated through the identity credibility update model. Further security measures or permission adjustments are triggered through anomaly detection.

[0145] Specifically, the input for this step comes from the output of step 4—the user's access rights (A(t)). This value is determined by the identity credibility score and resource request information, and is a dynamic representation of the user's current access rights. Other input information includes the user's historical behavior information (B(t)) and real-time access records (R(t)). This data will be used in the next step to update the identity score and adjust access rights.

[0146] Furthermore, the identity monitoring model is designed: The goal of this step is to continuously monitor the user's identity status and dynamically adjust access rights based on real-time information. To achieve this goal, an identity credibility update model (f id ), which dynamically adjusts the user's identity credibility score based on the user's behavior history, current access records, and system status. The core function of this model is as follows:

[0147] C id(t+1)=f id (C id (t),B(t),R(t),A(t)) (18)

[0148] Among them, C id (t): The identity credibility score at the current time point. B(t): The historical behavior characteristics of the current sample, including access behavior patterns, operation types, and other information. R(t): Real-time access records, indicating the sample's most recent request or operation. A(t): System environment factors, such as current load, task type, etc. C id (t+1): Updated identity credibility score.

[0149] This function adjusts a user's identity score based on their historical behavior (B(t)), real-time request history (R(t)), and current access permissions (A(t)). If the user's behavior is normal, the identity credibility score increases; if there is abnormal behavior, the score decreases.

[0150] Furthermore, according to the identity credibility score monitored in real time, the present invention further designs a dynamic access right adjustment mechanism. This mechanism uses the identity credibility threshold adjustment function (T adjust ) to determine the access rights of each user. Specifically, the present invention uses a linear adjustment formula based on credibility score:

[0151] T adjust (C id (t+1))=α·C id (t+1)+β (19)

[0152] Among them, T adjust : Dynamically calculated access rights adjustment threshold. α: Weight coefficient that controls the impact of trustworthiness score on access rights. β: Basic threshold for permissions, ensuring that users still have basic permissions even when the trust score is low. C id (t+1): The updated identity credibility score. This function is used to determine whether the user is qualified to access certain high-authority resources through a linear adjustment formula. If the identity credibility score exceeds the adjusted threshold (T adjust ), the user is allowed to access the corresponding resource.

[0153] Furthermore, during the real-time security adjustment process, the present invention also needs to implement an anomaly detection mechanism to respond promptly when abnormal behavior occurs. The present invention defines an anomaly scoring function to measure the degree of abnormality of user behavior. The specific anomaly scoring formula is:

[0154]

[0155] Where, ΔCid (t): Abnormal score of the sample. C id (t): Current identity credibility score. μ: Mean identity credibility score. σ: Standard deviation of identity credibility score. This function calculates the degree of abnormality of user behavior based on the difference between the current user's identity credibility score and the mean. If the difference between the user's identity score and the mean exceeds a preset threshold, the user's behavior is considered abnormal, triggering further security monitoring or permission adjustments.

[0156] Output:

[0157] The output of this step is the dynamically calculated user access rights (A(t)), which is determined by the above formula based on the identity credibility score and resource request information. The output may be:

[0158] A specific access permission level, such as 1 for full access and 0 for denied access; or

[0159] Permission vectors represent access permissions for different resource categories (e.g., data access permissions, system operation permissions, etc.).

[0160] This step of the present invention fully embodies the concept of zero trust architecture, ensuring that each access request must be authorized based on real-time identity authentication, rather than relying on previous authentication information or permission models. At each request, the system needs to be based on the dynamic identity credibility score (C id Access requests are verified in real time using B(t)) and historical behavior data (B(t)), ensuring that each request complies with current security policies. This zero-trust architecture means that even after a user has been authenticated, their system permissions are continuously adjusted based on their real-time behavior. This real-time assessment and adjustment mechanism prevents security risks caused by identity verification bypass or tampering by attackers.

[0161] Example 2

[0162] In a second embodiment of the present invention, a dynamic access control management system based on a zero-trust architecture is provided, the system comprising:

[0163] An implicit identity feature acquisition unit, configured to acquire original feature data, construct feature data, and assign labels to the feature data; wherein the original feature data is obtained through preprocessing;

[0164] An identity chain generation unit is configured to generate an identity chain based on the feature data and the tag, perform cryptographic signing and authentication on the identity chain, and obtain an initial authentication result set; wherein the authentication is used to reject any identity chain that fails to pass the match;

[0165] An identity chain enhancement unit, configured to perform an adversarial sample attack on the initial authentication result set, thereby increasing the detection capability of the authentication result set and preventing forged identity attacks, and generating a final authentication result set;

[0166] The adversarial sample attack achieves adversarial sample detection by calculating identity information perturbation metrics, weighted perturbation evaluation, and dynamic regularization terms.

[0167] A user access permission setting unit is configured to obtain user identity credibility based on the authentication result set, obtain resource information requested by the user, dynamically adjust the access permission requested by the user based on the user identity credibility, and output the user access permission;

[0168] The access rights optimization unit is used to monitor the identity chain in real time. Based on the dynamically calculated user identity credibility and real-time access records, it updates access rights through the identity credibility update model and triggers further security measures or permission adjustments through anomaly detection.

[0169] The identity credibility update model dynamically adjusts the user's identity credibility score according to the user's behavior history, current access record and system status, and controls the access rights of each user based on a preset identity credibility threshold.

[0170] The foregoing description of specific embodiments of the present disclosure is intended to illustrate a method for performing a multi-tasking process. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0171] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0172] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0173] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the embodiments of this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0174] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0175] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0176] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0177] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0178] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0179] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0180] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0181] This specification may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0182] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0183] Finally, it should be noted that the lithium battery pack chip balancing control platform disclosed in the embodiment of the present invention is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, ordinary technicians in this field should understand that it is still possible to modify the technical solutions recorded in the aforementioned embodiments, or to replace some of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A dynamic access control management method based on zero trust architecture, characterized in that: The method comprises the following steps: Acquiring original feature data, constructing feature data, and assigning labels to the feature data; wherein the original feature data is obtained through preprocessing; Generate an identity chain based on the feature data and the tag, encrypt and sign the identity chain, and authenticate it to obtain an initial authentication result set; wherein the authentication is used to reject any identity chain that cannot pass the match; Performing an adversarial sample attack on the initial authentication result set to increase the detection capability of the authentication result set and prevent forged identity attacks, and generating a final authentication result set; The adversarial sample attack achieves adversarial sample detection by calculating identity information perturbation metrics, weighted perturbation evaluation, and dynamic regularization terms. Obtaining user identity credibility based on the authentication result set, obtaining resource information requested by the user, dynamically adjusting the access rights requested by the user based on the user identity credibility, and outputting the user access rights; Real-time identity chain monitoring, based on dynamically calculated user identity credibility and real-time access records, updates access rights through the identity credibility update model, and triggers further security measures or permission adjustments through anomaly detection; The identity credibility update model dynamically adjusts the user's identity credibility score according to the user's behavior history, current access record and system status, and controls the access rights of each user based on a preset identity credibility threshold.

2. The dynamic access control management method based on zero trust architecture according to claim 1 is characterized in that: The preprocessing includes data cleaning, normalization, denoising and label generation.

3. The dynamic access control management method based on zero trust architecture according to claim 1 is characterized in that: Generating an identity chain based on the feature data and the tag, encrypting and signing the identity chain, and authenticating the identity chain to obtain an initial authentication result set specifically includes: Build a corresponding identity chain based on each feature data through a mapping function as a unique identity identifier; Encrypted identity chain to obtain encrypted identity information; Sign the encrypted identity information through the signature algorithm to obtain the corresponding signature result; A weighted regularization term is designed to adjust the matching degree between the encrypted identity information and the signature, reducing overmatching of low-trust identity information. The corresponding identity chain, the corresponding signature result and the weighted regularization item are combined into an authentication result to construct a final authentication result set.

4. The dynamic access control management method based on zero trust architecture according to claim 3 is characterized in that: The design weighted regularization term includes: Get the matching degree between the identity chain and its corresponding signature result; Based on the matching degree and the validity indicator function of the identity information, a weighted regularization term is obtained by weighted summation; The corresponding identity chain, the corresponding signature result and the weighted regularization item are passed through a verification function to generate an authentication result.

5. The dynamic access control management method based on zero trust architecture according to claim 1 is characterized in that: The adversarial sample attack is performed on the initial authentication result set to increase the detection capability of the authentication result set and prevent forged identity attacks, and generate a final authentication result set, specifically including: By comparing the difference between the predicted result of each identity information obtained in the preset authentication model and the original authentication result, it is determined whether the information is affected by the adversarial sample attack and the perturbation measure of the corresponding identity information is generated; Performing weighted perturbation evaluation analysis on the perturbation metric corresponding to the identity information to generate a weighted perturbation metric; Regularization terms are introduced to penalize abnormal disturbances during the perturbation identification process, thereby improving the ability to filter adversarial samples. Combining the weighted perturbation measure and the regularization term, we generate an adversarial sample detection decision function to determine whether to accept a piece of identity information: For identity information whose sum of the weighted perturbation metric and the regularization term exceeds a preset threshold, the identity information will be rejected; otherwise, the identity authentication process will continue.

6. The dynamic access control management method based on zero trust architecture according to claim 5 is characterized in that: The determination of whether the information is affected by the adversarial sample attack is specifically as follows: If the difference between the predicted result and the original authentication result exceeds the set threshold, it is considered that the identity information may be vulnerable to adversarial attacks; The weighted perturbation estimation analysis weights the perturbation metric according to the sensitivity of the features.

7. The dynamic access control management method based on zero trust architecture according to claim 1 is characterized in that: The steps of obtaining the user identity credibility based on the authentication result set, obtaining the resource information requested by the user, dynamically adjusting the access rights requested by the user based on the user identity credibility, and outputting the user access rights specifically include: Use dynamic credibility update rules to update user identity credibility and generate dynamic credibility; Based on the sensitivity of the resource request and the user's behavioral activity, the dynamic credibility is comprehensively scored to further adjust the identity credibility; Adopting a preset dynamic access rights calculation model, it determines whether to allow a user to access a resource based on different dynamic credibility and resource request sensitivity, and generates dynamically calculated user access rights. The user access rights include: a specific access rights level and an access rights vector.

8. The dynamic access control management method based on zero trust architecture according to claim 7 is characterized in that: The dynamic credibility update rule calculates the updated credibility by combining the historical credibility and the currently evaluated credibility in a weighted average manner to obtain the dynamic credibility; the dynamic credibility changes according to the time of the authentication, and the authentication change is based on the updated version of the authentication result obtained by the adversarial sample detection; The access rights calculation rules of the preset dynamic access rights calculation model are as follows: Among them, A(t) is the user access permission, θ high and θ low are high confidence threshold and low confidence threshold, which are used to determine the upper and lower limits of access; scale(C(t),θ low ,θ high ) is mapped to the range [0,1] according to the dynamic identity credibility C(t), indicating partial access rights; C(t) is the dynamic identity credibility.

9. The dynamic access control management method based on zero trust architecture according to claim 1, characterized in that: The real-time identity chain monitoring is performed, and based on the dynamically calculated user identity credibility and real-time access records, access rights are updated through the identity credibility update model, and further security measures or permission adjustments are triggered through anomaly detection, specifically including: Adjust the user's identity credibility score based on the user's historical behavior, real-time request records, and current access rights; if the user's behavior is normal, the identity credibility score will increase; if abnormal behavior occurs, the score will decrease; Based on the real-time monitored identity credibility score, the access permission adjustment threshold is dynamically calculated. If the identity credibility score exceeds the adjusted access permission adjustment threshold, the user is allowed to access the corresponding resources. At the same time, during the real-time security adjustment process, anomaly detection is performed based on the current identity credibility score and the average of the identity credibility scores. If the difference between the user's identity credibility score and the average of the identity credibility scores exceeds the preset threshold, the user behavior is considered abnormal, triggering further security monitoring or permission adjustment.

10. A dynamic access control management system based on zero trust architecture, characterized in that: The system comprises: An implicit identity feature acquisition unit, configured to acquire original feature data, construct feature data, and assign labels to the feature data; wherein the original feature data is obtained through preprocessing; An identity chain generation unit is configured to generate an identity chain based on the feature data and the tag, perform cryptographic signing and authentication on the identity chain, and obtain an initial authentication result set; wherein the authentication is used to reject any identity chain that fails to pass the match; An identity chain enhancement unit, configured to perform an adversarial sample attack on the initial authentication result set, thereby increasing the detection capability of the authentication result set and preventing forged identity attacks, and generating a final authentication result set; The adversarial sample attack achieves adversarial sample detection by calculating identity information perturbation metrics, weighted perturbation evaluation, and dynamic regularization terms. A user access permission setting unit is configured to obtain user identity credibility based on the authentication result set, obtain resource information requested by the user, dynamically adjust the access permission requested by the user based on the user identity credibility, and output the user access permission; The access rights optimization unit is used to monitor the identity chain in real time. Based on the dynamically calculated user identity credibility and real-time access records, it updates access rights through the identity credibility update model and triggers further security measures or permission adjustments through anomaly detection. The identity credibility update model dynamically adjusts the user's identity credibility score according to the user's behavior history, current access record and system status, and controls the access rights of each user based on a preset identity credibility threshold.

Images