ABAC policy mining method and system for heterogeneous access control systems

The ABAC policy mining method using user-permission tuples and an ant colony algorithm addresses the inefficiencies of traditional access control models by efficiently constructing and migrating to ABAC systems, enhancing policy mining efficiency and coverage.

CN116450729BActive Publication Date: 2024-12-03Chinese People's Liberation Army Cyberspace Force Information Engineering University
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310431235.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-04-20
Publication Date
2024-12-03
Estimated Expiration
2043-04-20

AI Technical Summary

Technical Problem

Existing access control models like DAC, MAC, and RBAC struggle to meet the fine-grained and expressive security requirements of complex dynamic systems, and manually constructing ABAC policies is difficult and costly due to the flexibility and large number of subjects and objects.

Method used

An ABAC policy mining method using user-permission tuples and an ant colony algorithm to simulate constraint paths, enabling efficient construction and migration of access control systems by generating optimal policy constraints and simplifying policies.

Benefits of technology

Improves policy mining efficiency by reducing WSC by 34.6%, increasing covered user-permission tuples by 1.03 times, and facilitating the construction and migration of heterogeneous access control systems to ABAC systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116450729B_ABST
    Figure CN116450729B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of access control technology, and in particular to an ABAC policy mining method and system for heterogeneous access control systems, which extracts user-authority tuples and corresponding attribute relationships from heterogeneous access control systems based on access control lists; traverses all attributes of users and resources in user-authority tuples, and obtains constraint sets satisfied by different user attributes and resource attributes by comparing attribute value relationships; uses the constraint set as an ant colony algorithm input, obtains policy constraints corresponding to the best path by simulating different constraints as cities with path distances in the ant colony algorithm, and adds the policy constraints to the policy set; selects the final policy results required for mining from the policy set based on preset policy quality evaluation criteria and the size of user-authority tuples that the policy can cover. The present invention can improve the efficiency of policy mining and the quality of the mining policy set, is helpful for the construction and migration of access control systems, and can be used for the migration of heterogeneous access control systems to ABAC access control systems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of access control technology, and in particular to an ABAC strategy mining method and system for heterogeneous access control systems. Background Art

[0002] With the rapid development of the Internet, various data resources have shown explosive growth, and are also facing many security challenges. Access control systems are a key component of information systems, and their role is to protect information resources from unauthorized access. Access control policies are a direct reflection of the security intention and usage intention of information systems, and are also an important prerequisite for the implementation of access control. In distributed big data application scenarios, it is necessary to manage a large number of user access rights in applications and business systems distributed in different computing environments. With the rapid development of information technology, traditional access control models such as DAC (Discretionary access control), MAC (Mandatory access control), and RBAC (Role-Based access control) can no longer meet the fine-grained and expressive security requirements of emerging applications. Today, the attribute-based access control (ABAC) model provides a high degree of flexibility, promotes security and information sharing, overcomes the role explosion and other problems associated with RBAC, and provides a more flexible method to solve the authorization requirements of complex dynamic systems. In practical applications, different business systems may adopt different access control models. We need a universal method to migrate the non-ABAC access control system to the ABAC system, mine the ABAC policy based on the security intent of the original access control policy, and build the ABAC policy set.

[0003] The existing methods for constructing ABAC policy sets are mainly divided into two types: top-down and bottom-up. The basic idea of ​​the top-down method is to combine security requirements, decompose the business into independent functional units, refine the functional units to specific accesses, and construct access control policies based on these accesses. The usual method is to convene experts for manual analysis, or use natural language processing technology to analyze the ABAC policy set from the security requirements document; the bottom-up method is also called policy mining, which usually uses a policy mining algorithm. The basic idea is to combine the subject and object attribute data to mine ABAC policies from the subject-object permission relationship. Due to the flexibility of ABAC, when the number of subjects and objects and the number of subject-object attributes are large, manually developing ABAC policies is not only difficult and costly, but also difficult to apply ABAC in practice. Summary of the invention

[0004] To this end, the present invention provides an ABAC policy mining method and system for heterogeneous access control systems, which solves the problems of construction, migration and integration of access control policies in heterogeneous access control systems, and facilitates the implementation of migration of heterogeneous access control systems to ABAC access control systems.

[0005] According to the design scheme provided by the present invention, an ABAC policy mining method for heterogeneous access control systems is provided, comprising:

[0006] Extract user-permission tuples and corresponding attribute relationships from heterogeneous access control systems based on access control lists, where user-permission tuples are used to describe the user's operation permissions on resources;

[0007] Traverse all attributes of users and resources in the user-permission tuple, and obtain the constraint set satisfied by different user attributes and resource attributes by comparing the attribute value relationship;

[0008] The constraint set is used as the input of the ant colony algorithm. The policy constraint corresponding to the optimal path is obtained by simulating different constraints as cities with path distances in the ant colony algorithm, and the policy constraint is added to the policy set.

[0009] The final policy results to be mined are selected from the policy set based on the preset policy quality evaluation criteria and the size of the user-authority tuple that the policy can cover.

[0010] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, the user-permission tuple is represented as a triple containing user u, resource r and operation o.

[0011] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, by simulating different constraints as cities with path distances in the ant colony algorithm to obtain the policy constraints corresponding to the optimal path, it includes:

[0012] First, obtain the user group with the same constraint set result, build the user attribute expression and resource attribute expression according to the conditions that the users and resources in the user group need to meet, and generate the initial policy;

[0013] Then, different cities in the ant colony algorithm are used to match the constraints in the initial strategy, the distance values ​​between cities are randomly set, the pheromone concentration is used to calculate the possibility of path selection and generate different generalization methods, by finding the best path to travel through the city, and using the generalization method to process the strategy constraints in turn, and the optimal generalization method and its corresponding optimal quality strategy are obtained through evaluation.

[0014] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, the optimal generalization method and its corresponding optimal quality policy are obtained by evaluation, and the generalization method is evaluated by a preset quality function, wherein the quality function Q policy Expressed as: Q policy (ρ,UP)=|[ρ]∩UP| / WSC(ρ), ρ is the initial policy, UP is the set of user-authority tuples, and WSC() is the preset policy quality evaluation standard.

[0015] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, in evaluating the generalization method through a preset quality function, the quotient of the quality function value before generalization and the quality function value after generalization is used as the result value for evaluating this generalization, and the optimal generalization method and its corresponding optimal quality strategy are found according to the size of the result value.

[0016] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, in the policy quality evaluation, a preset policy quality evaluation standard is set according to the number of constraints and the custom weights and using weighted structural complexity.

[0017] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, according to the preset policy quality evaluation standard and the size of the user-authority tuple that the policy can cover, the final policy result required to be mined is selected from the policy set, including:

[0018] First, quality processing is performed on the policy set, and the quality processing includes policy merging and policy simplification processing;

[0019] Then, policies that meet the preset quality evaluation criteria are continuously selected from the processed policy set, and the final policy results required for mining are obtained based on the size of the covered user-authority tuple.

[0020] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, the policy merging process includes: if the user-permission tuple covered by the current policy is included in the user-permission principle covered by another policy, the current policy is deleted; if the constraints returned by the two policies are the same, under the premise that the policies are valid, a synthetic policy is generated by merging the attribute relationship of the two policies.

[0021] As the ABAC policy mining method for heterogeneous access control systems of the present invention, further, the policy simplification processing includes: if the multi-valued attribute of the current policy has an attribute value a, (a, b), the attribute value of the current policy is adjusted to a; if the attribute value of the current policy includes a and the attribute value b must exist, the attribute value {a, b} of the current policy is removed from the value b; for overlapping attribute values ​​and overlapping operation values ​​in two policies, keep the attribute value and operation value of one of the policies unchanged, and delete the overlapping attribute value and overlapping operation value in the other policy.

[0022] Furthermore, the present invention also provides an ABAC policy mining system for heterogeneous access control systems, comprising: a data extraction module, a constraint construction module, a policy acquisition module and a policy output module, wherein:

[0023] The data extraction module is used to extract the user-permission tuple and the corresponding attribute relationship from the heterogeneous access control system according to the access control list, wherein the user-permission tuple is used to describe the user's operation authority over the resource;

[0024] The constraint building module is used to traverse all attributes of users and resources in the user-permission tuple and obtain the constraint set satisfied by different user attributes and resource attributes by comparing the attribute value relationship;

[0025] A strategy acquisition module is used to take the constraint set as the input of the ant colony algorithm, obtain the strategy constraint corresponding to the optimal path by simulating different constraints as cities with path distances in the ant colony algorithm, and add the strategy constraint to the strategy set;

[0026] The policy output module is used to select the final policy results required for mining from the policy set based on the preset policy quality evaluation criteria and the size of the user-authority tuple that the policy can cover.

[0027] Beneficial effects of the present invention:

[0028] The present invention uses user-permission tuples as policy mining seeds to generate ABAC policies. The algorithm iterates user-permission tuples as policy mining seeds, combines attribute relationships, and converts restrictions in attribute expressions into constraint relationships. The policy mining process is analogized to the ant colony algorithm to solve the TSP problem for policy mining, thereby improving the efficiency of policy mining, facilitating the construction and migration of access control systems, and solving the problem of heterogeneous policy fusion. The quality of the policy set is further improved by improving the strategy through methods such as merging and simplification. Experimental data is used to further show that in the case of policy mining with a high degree of complexity, the scheme of this case can also reduce the WSC value of the mining result by 34.6%, increase the number of covered user-permission tuples by 1.03 times, and improve the efficiency by 58.3%. It can solve the problems of construction, migration and fusion of access control policies in heterogeneous access control systems, and can be used to migrate heterogeneous access control systems to ABAC access control systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the ABAC policy mining process for heterogeneous access control systems in the embodiment;

[0030] Figure 2 This is a schematic diagram of the implementation principle of the ABAC strategy mining algorithm in the embodiment;

[0031] Figure 3 This is a comparison of the running time of different ant numbers in the embodiment;

[0032] Figure 4 It is a comparison diagram of different ant numbers WSC in the embodiment;

[0033] Figure 5 The experimental results of the pheromone importance control variables in the embodiment are shown in FIG.

[0034] Figure 6 The experimental results of the pheromone volatility factor control variables in the embodiment are shown in FIG.

[0035] Figure 7 The experimental results of the pheromone constant control variables in the embodiment are shown in FIG.

[0036] Figure 8 It is a schematic diagram of the experimental results of the iteration number control variable in the embodiment;

[0037] Fig. 9 This is a schematic diagram of the algorithm running time comparison in the embodiment;

[0038] Fig.10 It is a comparison diagram of the policy set WSC in the embodiment;

[0039] Fig.11 Schematic diagram of the average coverage of user-authority tuples by policies in the embodiment. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the present invention clearer and more understandable, the present invention is further described in detail below in conjunction with the accompanying drawings and technical solutions.

[0041] The ABAC model uses the attributes of the subject and the object as decision factors for access control. It can flexibly determine whether to grant corresponding access rights based on the security intent of the ABAC policy and the attribute set of the entity. The ABAC model provides a more flexible way to solve the authorization requirements of complex dynamic systems. In the process of migrating heterogeneous access control systems to ABAC access control systems, in order to retain the security intent of the original system, it is necessary to mine ABAC policies from the access control policies of the original system. Regardless of the access control model used by the original system, its policies can be converted into user-permission tuples.

[0042] The relevant theories about attributes and attribute restrictions in common ABAC policy language structures are introduced as follows:

[0043] ABAC policy includes user attributes and resource attributes, defining user set U and user attribute set A u , function d u (u,a) represents the specific value of attribute a in user u. Similarly, define resource set R and resource attribute set A r , function d r (r,a) represents the specific value of attribute a in resource r. u Can be divided into single-valued attribute set A u,1 and multi-valued attribute set A u,m . A r Can be divided into single-valued attribute set A r,1 and multi-valued attribute set A r,m . Define Val s is the set of all attribute values, and ⊥ is used to represent unknown attribute values. The multi-valued attribute set can be represented as Val m =set(Val s \{⊥})∪⊥, where set(S) is the power set of set S. For single-valued attributes, use d u (u,a)=v means that the user satisfies a certain attribute value v.

[0044] Attribute expressions are used to express the conditions that users and resources must satisfy in a policy. User attribute expressions (UAE) use function e to restrict user attributes. e(a) is called the attribute restriction on attribute a. The value of e(a) is the attribute base Val. s \{⊥} or Val m\{⊥}, depending on whether attribute a is a single-valued attribute or a multi-valued attribute. When e(a) is T, no restriction is imposed on the attribute value. Define attr(e) to represent the attribute set used by e, attr1(e) and attr m (e) represents the single-valued attribute set and multi-valued attribute set used by e. When the user satisfies a certain attribute restriction e, it is represented by u|=e. Definition [e] u is the set of users in the user set U that satisfy the attribute restriction e, that is, [e] u = {u∈U|u|=e}. For multi-valued attributes, use Indicates that the user has a certain attribute value v and may have other attribute values.

[0045] Example 1: Hypothesis

[0046] A u,1 ={subject,position},A u,m ={courses},

[0047] e1(subject)={math},e1(position)={graduate,undergraduate},e1(courses)={{C1,C2}}

[0048] The user attribute expression indicates that the mathematics department user is a graduate student or undergraduate, and the courses include C1 and C2.

[0049] For multi-valued attributes in resource attribute expressions (RAEs), use d r (r,a)=v means that the resource has a certain attribute value v, and the rest is the same as the user attribute expression.

[0050] To increase readability, for the attribute restriction e(a), when e(a) is not T, let v = e(a). When a is a single-valued attribute, the attribute restriction e is written as a∈v. If there is only element v in the set s, it can be written as a=s. When a is a multi-valued attribute, the attribute restriction e is written as If the multi-valued attribute set s has only element v, the attribute restriction e can be written as

[0051] In ABAC policies, constraints are used to express the relationship between user attributes and resource attributes. u,1 ∈A u,1 , a u,m ∈A u,m ,a r,1 ∈A r,1 , a r,m ∈A r,m When Defined as a basic constraint. A constraint set contains multiple basic constraints.

[0052] Example 2:

[0053]

[0054] It means that the user's hobbies include all values ​​of the subject attribute of the resource, and the user's subject includes a certain attribute value of the available subject attribute of the resource.

[0055] A user-permission tuple is a triple consisting of user u, resource r, and operation o, indicating that user u has the permission to perform operation o on resource r.

[0056] The ABAC strategy is a u ,e r ,O,c>tuple, where e u is the expression of all user attributes, e r are all resource attribute expressions, O is the operation set, and c is the constraint set.

[0057] Example 3: For a strategy using constraints,

[0058]

[0059] Indicates that students can read the grades of the courses they are taking. For a strategy that does not use constraints, for a strategy that does not use constraints:

[0060] Example 4: For a strategy that does not use constraints,

[0061]

[0062] Indicates that the Manager can read all results.

[0063] The attribute restrictions in the attribute expression that are the same as the constraint attributes are deleted, and the constraints are added to the strategy, which is called a generalization operation.

[0064] The ant colony algorithm was proposed by Marco Dorigo and others in the 1990s, and was inspired by the foraging process of ants. The algorithm simulates the collective path-finding behavior of ants in nature. It is a heuristic bionic evolutionary algorithm that has been successfully applied to solve the traveling salesman problem (TSP). Ants transmit information through a pheromone. During the movement, they can leave this substance on the path they pass through, sense its concentration, and guide their path-finding process. The algorithm continuously seeks the optimal solution through the release, volatilization, and accumulation of pheromones. It adopts a distributed parallel computing mechanism and has strong robustness, but it is prone to falling into local optimal solutions. Taking the solution of the TSP problem as an example, the main process of the algorithm is as follows:

[0065] 1) Set the number of ants according to the specific problem and search in parallel.

[0066] 2) After each ant completes a search, it releases pheromones along the path it passes (the initial pheromone concentration is set to an equal value), and the pheromone concentration is proportional to the quality of the solution.

[0067] 3) Ants use random search to select a path based on the pheromone concentration and the distance between the two cities. Generally, the shorter the distance between the two cities, the greater the amount of pheromone between the paths, and the greater the probability that the ants will choose that edge.

[0068] 4) When all ants have finished searching once, one round of iteration is completed. Each iteration updates the pheromone concentration on all paths, including the evaporation of the original pheromone and the increase of the pheromone concentration.

[0069] 5) After reaching the predetermined number of iterations or the solution no longer changes, the algorithm ends and the current optimal solution is taken as the optimal solution to the problem.

[0070] In the embodiment of the present invention, for the strategy mining under the complexity of the attribute relationship between users and resources, see Figure 1 As shown, an ABAC policy mining method for heterogeneous access control systems is provided, including:

[0071] S101. Extracting user-authority tuples and corresponding attribute relationships from a heterogeneous access control system according to an access control list, wherein the user-authority tuple is used to describe the user's operation authority over a resource.

[0072] S102: traverse all attributes of the user and the resource in the user-authority tuple, and obtain the constraint set satisfied by different user attributes and resource attributes by comparing the attribute value relationship.

[0073] S103: Using the constraint set as an input of the ant colony algorithm, simulating different constraints as cities with path distances in the ant colony algorithm to obtain the policy constraints corresponding to the optimal path, and adding the policy constraints to the policy set.

[0074] S104: Select the final policy result to be mined from the policy set based on the preset policy quality evaluation standard and the size of the user-authority tuple that the policy can cover.

[0075] The policy form of the access control list ACL can be expressed as a <U,R,O p ,UP>tuple, where U is the set of users, R is the set of resources, O p is a collection of operations, represents a set of user-permission tuples, which are obtained by combining access control lists. If the ABAC policy π and the ACL policy tuple <U,R,O p ,UP>have the same user set, resource set and operation set, and [π]=UP, then the security meaning expressed by the two Figure 1To.

[0076] The ABAC policy mining problem can be transformed into: using the given ACL policy tuple π0, combined with user attributes, resource attributes, user attribute values, and resource attribute values, construct an ABAC policy π=<U,R,O p ,A u ,A r ,d u ,d r ,Rules>makes the security intention it expresses consistent with π0 and has a higher policy quality.

[0077] In the embodiment of this case, the user-authority tuple is used as a policy mining seed to generate an ABAC policy. The policy mining process is analogized to the ant colony algorithm for solving the TSP problem. The non-ABAC access control system can be migrated to the ABAC system. The ABAC policy can be mined and the ABAC policy set can be constructed based on the security intent of the original access control policy to meet the authorization requirements of complex dynamic systems.

[0078] Furthermore, the embodiment of this case obtains the policy constraints corresponding to the optimal path by simulating different constraints as cities with path distances in the ant colony algorithm, which can be designed to include the following contents:

[0079] First, obtain the user group with the same constraint set result, build the user attribute expression and resource attribute expression according to the conditions that the users and resources in the user group need to meet, and generate the initial policy;

[0080] Then, different cities in the ant colony algorithm are used to match the constraints in the initial strategy, the distance values ​​between cities are randomly set, the pheromone concentration is used to calculate the possibility of path selection and generate different generalization methods, by finding the best path to travel through the city, and using the generalization method to process the strategy constraints in turn, and the optimal generalization method and its corresponding optimal quality strategy are obtained through evaluation.

[0081] For the policy mining results, an evaluation standard can be predefined to evaluate the quality of the mined ABAC policies. In this embodiment, the preset policy quality evaluation standard can be set based on the number of constraints and custom weights and using weighted structural complexity. Among them, the calculation method using weighted structural complexity (WSC) as the policy quality evaluation standard can be described as follows:

[0082]

[0083] WSC(<e u ,e r ,O,c>)=w1WSC(e u )+w2WSC(e r )+w3|O|+w4|c|

[0084]

[0085] Among them, |s| is the number of attribute restrictions, w i The weights are defined for different situations. The usability study of access control policies shows that simpler policies are easier to manage. The lower the WSC value, the higher the quality of the mined ABAC policies.

[0086] Combining ACL (Access control list) with subject and object attribute data to mine ABAC policies, the embodiment of this case also provides an example of a policy mining implementation algorithm. The core of the algorithm is to loop the user-permission tuples in the ACL, use the selected tuple as the seed for building the ABAC policy, calculate the available constraints of the tuple, and use the constraint replacement method to replace the corresponding attribute restrictions in the attribute expression. Under the premise of ensuring the effectiveness of the policy, define the policy quality function, use the ant colony algorithm to find the best processing method, and add the processed high-quality policy to the candidate policy set until all user-permission tuples are covered. Finally, the algorithm is optimized through methods such as merging and simplification to improve the quality of the policy and obtain the optimal policy set.

[0087] Loop through the user-permission tuples in the ACL, combine the attribute calculation attribute expression, and generate the initial ABAC policy. Calculate the constraint set satisfied by different user attributes and resource attributes, traverse each constraint in the constraint set, and generate the final ABAC policy through continuous generalization operations. The constraint set is used as the input of the ant colony algorithm. On the premise of ensuring that all user-permission tuples can be covered, find a generalization method that makes the policy have the best quality, and add the corresponding optimal policy to the policy set as the mining result. The quality of the policy set is continuously improved through subsequent optimization methods such as merging and simplification. In the specific implementation of the algorithm, see Figure 2 As shown in the figure, it can be divided into the following contents: select a user-permission tuple and calculate the constraint relationship satisfied between the user and the resource; calculate the attribute expressions satisfied by the user and the resource respectively and generate the initial strategy; use different generalization methods, that is, replace the corresponding attribute restrictions with constraint relationships; define the quality function, use the ant colony algorithm to find the generalization method with the best quality, and add the corresponding optimal strategy to the strategy set; merge, simplify, reduce constraints and other operations on the strategy set generated after processing all user-permission tuples; continue to select the strategy with the lowest WSC from the strategy set until all user-permission tuples are covered.

[0088] Figure 2The main process of the policy mining algorithm shown in is shown in Algorithm 1. The algorithm uses the computeConstraint function to calculate the constraint relationship between users and resources, and then performs specific generalization operations and selects the optimal policy through the addRule function. Finally, the merge, simplify, and eliconstraint functions are used to improve the quality of the policy set, and low WSC policies are continuously selected to cover all user-permission tuples as the final mining results.

[0089]

[0090]

[0091] The algorithm input of the policy mining algorithm is the user-permission tuple and the corresponding attribute relationship. The user-permission tuple can be obtained directly or extracted from a heterogeneous access control system. The computeConstraint function finds the constraint relationship by traversing all attributes of user u and resource r in the user-permission tuple and comparing their attribute value relationships to generate a constraint set.

[0092] Function addRule is the main function that uses the ant colony algorithm to find the optimal strategy. Among them, the antSearch function performs specific generalization operations, and the evaluation function evaluates different generalization methods by defining a quality function to find the optimal strategy corresponding to the optimal generalization method.

[0093] The addRule function calls the computeUAE and computeRAE functions, traverses the ACL policy tuple and attribute relationship, calculates the user attribute expression and resource attribute expression, generates the initial policy ρ, and then generalizes ρ through the antSearch function, and takes the returned optimal policy as the function result. The specific steps are shown in Algorithm 2.

[0094]

[0095] The antSearch function solves the TSP problem by imitating the ant colony algorithm, using different cities as analogies for different constraints, and finding the best path to travel through the cities to solve the optimal generalization method and optimal strategy. Its main process is shown in Algorithm 3.

[0096]

[0097]

[0098] The initPheromones function initializes the pheromone concentration. In this algorithm, the initial pheromone concentration is set to 1. The initHeuristics function initializes the distance matrix between cities and randomly sets the distance values ​​between cities. The generateSolutions function calculates the possibility of path selection through pheromone concentration and generates different generalization methods. The updateBestsolution function compares the fitness values ​​returned by the evaluation function each time, selects the generalization method with the highest evaluation and stores the corresponding strategy processing results. The updatePheromones function updates the pheromone concentration. The above functions are the basic process of the ant colony algorithm to solve the TSP problem, and the specific process will not be explained.

[0099] The evaluation function is the core function for solving problems using the ant colony algorithm. It performs specific generalization operations and defines a quality function to evaluate the quality of the generalized strategy as a basis for judging the quality of the generalization method. policy The definition is as follows:

[0100] Q policy (ρ,UP)=|[ρ]∩UP| / WSC(ρ)

[0101] In order to comprehensively consider the effect of generalization, the policy quality is defined by the size of user-permission tuples and WSC size that the generalized policy can cover. The more tuples are covered, the lower the WSC is, and the higher the policy quality is.

[0102] The generalization process processes the strategies in turn according to the constraint order of this generalization method. Under the premise of ensuring the effectiveness of the strategy, that is, the coverage tuple does not exceed the original tuple and the quality of the strategy after generalization is improved, the quotient of the quality function value of the strategy before generalization and the quality function value of the strategy after generalization is used as the result value of evaluating this generalization. The higher the quality function value after generalization, the smaller the result value, the higher the evaluation of this generalization method, and the higher the concentration of released pheromones. Through the continuous pathfinding of the ant colony, it tends to find the optimal generalization method and return the corresponding strategy processing result. The specific process is shown in Algorithm 4.

[0103]

[0104]

[0105]

[0106] In improving the quality of policy sets and selecting policies, we can reduce unnecessary constraints by merging and simplifying policies and improve the quality of policy sets. We can also cover user-authority tuples by continuously selecting low WSC policies as the final output mining results.

[0107] The merge function merges the mined strategies to improve the quality of the strategy set. It can mainly handle the following two situations:

[0108] 1) If a user-permission tuple that a policy can cover is included in another policy, then this policy is redundant and should be deleted.

[0109] 2) If the constraint calculation results returned by the two strategies are the same, then under the premise that the strategies are valid, the attribute relationships of the two strategies are merged to generate a composite strategy.

[0110] The simplify function simplifies the mined strategies themselves to improve the quality of the strategy set. The specific method can be designed as follows:

[0111] 1) When a multi-valued attribute has attribute values ​​a, (a, b), it can be simplified to attribute value a.

[0112] 2) When the attribute value contains a and the attribute value b must exist, the attribute value {a,b} can be removed from the value b.

[0113] 3) Eliminate overlapping attribute values ​​in the policy. For example, for the following two policies, the attribute value a1 can be deleted from the first policy.

[0114] <subject∈{a1,a2}∧identity=b1,class=c1,read,subject=subject>

[0115] <subject∈{a1}∧identity=b1,class={c1,c2},read,subject=subject>

[0116] 4) Delete overlapping operation values ​​in the strategy. For example, for the following two strategies, read can be deleted from the second strategy.

[0117] <subject=a1, class=c1, read, subject=subject>

[0118] <subject=a1∧identity=b1, class=c1,{read,write},subject=subject>

[0119] The eliconstraint function removes atomic constraints, reduces the WSC value, and improves the quality of the strategy while ensuring the effectiveness of the strategy. This function only takes the effectiveness of the strategy as the goal, retains only necessary constraints, finds unnecessary constraints and removes them to reduce the WSC value of the strategy.

[0120] Finally, in order to reduce the overall WSC of the policy set, without affecting the coverage of all user-permission tuples, strategies are continuously selected from the policy set with low WSC as the goal as the final mining result.

[0121] Furthermore, based on the above method, an embodiment of the present invention also provides an ABAC policy mining system for heterogeneous access control systems, comprising: a data extraction module, a constraint construction module, a policy acquisition module and a policy output module, wherein:

[0122] The data extraction module is used to extract the user-permission tuple and the corresponding attribute relationship from the heterogeneous access control system according to the access control list, wherein the user-permission tuple is used to describe the user's operation authority over the resource;

[0123] The constraint building module is used to traverse all attributes of users and resources in the user-permission tuple and obtain the constraint set satisfied by different user attributes and resource attributes by comparing the attribute value relationship;

[0124] A strategy acquisition module is used to take the constraint set as the input of the ant colony algorithm, obtain the strategy constraint corresponding to the optimal path by simulating different constraints as cities with path distances in the ant colony algorithm, and add the strategy constraint to the strategy set;

[0125] The policy output module is used to select the final policy results required for mining from the policy set based on the preset policy quality evaluation criteria and the size of the user-authority tuple that the policy can cover.

[0126] In order to verify the effectiveness of this solution, the following is a further explanation based on experimental data:

[0127] Use data generation algorithms to generate experimental data sets of different complexity, and use synthetic strategies to guide the generation of attribute data to generate data sets of various sizes and different characteristics. The different sizes of the average value of the number of constraints satisfied by the user-permission tuple in the data set are studied to represent policy mining of different complexity levels. Generally, the larger the value, the more complex the relationship between the user and the resource, and the higher the complexity of policy mining. The goal is to improve the discrimination of the number of constraints and generate data sets with different numbers of constraints. Due to the randomness of the data generation algorithm, other parameters cannot be guaranteed to be exactly the same. In the experiment, the weight w in the WSC definition i Equal to 1, the data set parameters are shown in the following table.

[0128] Table 1 Dataset parameters

[0129] Dataset number Constrained quantity mean WSC ACL Number of strategies 1 4.3 900 1176 10 2 8.4 909 1086 10 3 12.2 921 978 10 4 15.4 975 1032 10 5 19.5 1078 1361 10

[0130] The strategy mining algorithm in the embodiment of this case is used to conduct strategy mining on different data sets, and the WSC of the synthetic strategy and the WSC of the mining result are calculated. Since the synthetic strategy is always unnecessarily complex, it can be considered that the mining is effective if the WSC of the mining result is lower than the WSC of the synthetic strategy, and the lower the WSC value, the better the mining effect. The parameter selection of the ant colony algorithm in the algorithm is studied, and the Rhapsody algorithm is used as a control algorithm for performance comparison. Among them, the Rhapsody algorithm mines ABAC strategies from sparse logs, and defines a new metric - reliability, which is used to measure the looseness of the extracted strategy. However, it cannot mine strategies from logs with a large number of attributes, because the number of extracted strategies grows exponentially with the number of system attributes.

[0131] In the parameter setting of the ant colony algorithm, the number of ants m is generally set to 1.5 times the target number. When the number of ants is large, it may affect the efficiency of the algorithm; when the number is small, the solution may not be satisfactory. The setting interval of the pheromone importance α is [0,5]. The larger the value, the easier it is for the ant colony to converge, but the randomness of the search is weakened. The setting interval of the pheromone volatility factor ω is [0.2,0.5]. If the value is too large, it is not easy to find the optimal solution, and if the value is too small, it is easy to affect the efficiency of the algorithm. The setting interval of the pheromone constant δ is [10,1000]. The larger the problem scale, the higher the value should be. The setting interval of the number of iterations is [100,500]. Too high a number of iterations has no substantial meaning for the algorithm.

[0132] For different numbers of constraints, different ratios of the number of constraints to the number of ants m are used as variables to test and analyze the impact of the number of ants in the ant colony algorithm on the algorithm running time and mining results WSC under different numbers of constraints. The other parameter settings are all within the effective range. The results are shown in Table 2.

[0133] Table 2 The impact of the number of ants on the algorithm performance under different constraints

[0134]

[0135] Convert the tabular data into a chart to more intuitively show the impact of different numbers of ants on the algorithm performance under different numbers of constraints, such as Figure 3 and 4As shown in the figure, as the number of ants increases, when the number of constraints is large, the running time increases and the efficiency of the algorithm decreases. When the number of constraints is small, the WSC values ​​calculated with different numbers of ants tend to be stable. When the number of constraints is large, it is more difficult for the ant colony algorithm to find the optimal solution. The WSC value fluctuates up and down but the fluctuation range is small, which can be considered to be within an acceptable range. Considering the quality of the solution and the efficiency of the algorithm, the number of ants twice the number of constraints can be selected as the input of the algorithm. The selection of the ideal number of ants under different numbers of constraints is similar to the selection of the number of ants under different numbers of cities in the TSP problem, which proves the applicability of the ant colony algorithm to solve this problem.

[0136] Using the control variable method, each parameter of the ant colony algorithm is used as a variable. When constraint.size=8.4, m=16 is set to analyze the impact of different parameters on the algorithm running time and mining result WSC. In each control variable experiment, the remaining parameter settings are valid values ​​within the interval and are adjusted according to the ideal parameters obtained in the experiment. Due to the small scale of the problem, in the control variable experiment of the pheromone constant, only the control variable experiment is attempted in [10,100]. The experimental results are shown in Figures 5 to 8 As shown in the figure. Analysis of the experimental results shows that the values ​​of α, δ, and ω parameters have little effect on the performance of the algorithm. Based on the experimental results, we select the optimal or better results of each control variable experiment as our parameter settings. When the number of iterations n increases, there is no effect on WSC, but the algorithm running time increases significantly. In addition, since the relationship between different constraints is different from the distance between different cities in the TSP problem, the importance of the algorithm heuristic factor σ can be set to 0.0.

[0137] In summary, all parameters of the ant colony algorithm can be set as shown in Table 3.

[0138] Table 3 Ant colony algorithm parameters

[0139] Number of ants m constraint.size×2 Pheromone Importance α 3.0 Heuristic factor importance σ 0.0 Pheromone constant δ 50.0 Pheromone Volatilization Factor ω 0.4 Iteration number n 100

[0140] The strategy mining algorithm in this embodiment is compared with the comparison algorithm proposed by Xu et al. in terms of mining efficiency and strategy quality. The algorithm running time under different numbers of constraints is shown in the following figure. Fig. 9 As shown in the figure, the contrast algorithm proposed by Xu et al. is the first ABAC policy mining algorithm. The algorithm traverses the user-permission tuple and the subject-object attribute data to generate ABAC policies, and optimizes them by merging and simplifying the policies. It works well for simple policy mining situations, but its performance drops significantly as the complexity of the user-permission tuple increases.

[0141] Depend on Fig. 9It can be seen that when the number of constraints is small, the algorithm initialization takes a long time due to the ant colony algorithm used in the algorithm in this embodiment. As the number of constraints increases, the efficiency of the control algorithm decreases significantly due to the traversal method used to find the optimal generalization method, and the running time increases exponentially with the increase in the number of constraints. By taking advantage of the parallel search of the ant colony algorithm, the efficiency of the algorithm is greatly improved, and it can effectively handle strategy mining in complex situations.

[0142] The WSC value of the policy set under different numbers of constraints is compared with the average coverage of the policy user-permission tuple value. The results are as follows: Fig.10 , Fig.11 As shown, the algorithm in the embodiment of this case has a lower WSC value. This is mainly because in the antSearch function, due to the limitations of the ant colony algorithm itself, the generalization method sought may be relatively optimal. Therefore, for policy mining in complex situations, the ant colony algorithm in this case quickly and preliminarily screens effective generalization constraints, and then deletes all unnecessary constraints through the eliconstraint function, retains necessary constraints, and tends to mine access control policies with fewer constraints and simpler simplicity without violating the access control intention, so as to perform more macro access control. Analysis of the experimental results shows that compared with the control algorithm, the mining results of this case have an average WSC decrease of 34.6%, and an average increase of 1.03 times in the coverage of user-permission tuples. When constraint.size = 19.5, that is, in the case of policy mining with a higher degree of complexity, the algorithm efficiency is improved by 58.3%.

[0143] The above experimental data further show that the use of the characteristics of the ant colony algorithm in this solution can improve the efficiency of policy mining, help build and migrate access control systems, solve the problem of heterogeneous policy fusion, and facilitate the application of authorization requirements in complex dynamic systems.

[0144] Unless otherwise specifically stated, the relative steps, numerical expressions and values ​​of the components and steps set forth in these embodiments do not limit the scope of the present invention.

[0145] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0146] The units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person of ordinary skill in the art may use different methods to implement the described functions for each specific application, but such implementation is not considered to be beyond the scope of the present invention.

[0147] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits, and accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present invention is not limited to any specific form of combination of hardware and software.

[0148] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An ABAC policy mining method for heterogeneous access control systems, characterized by: Include: Extract user-permission tuples and corresponding attribute relationships from heterogeneous access control systems based on access control lists, where user-permission tuples are used to describe the user's operation permissions on resources; Traverse all attributes of users and resources in the user-permission tuple and obtain the constraint set satisfied by different user attributes and resource attributes by comparing the attribute value relationships; The constraint set is used as the input of the ant colony algorithm. The policy constraints corresponding to the optimal path are obtained by simulating different constraints as cities with path distances in the ant colony algorithm. The policy constraints are then added to the policy set. The policy constraints corresponding to the optimal path are obtained by simulating different constraints as cities with path distances in the ant colony algorithm. The method includes: first, obtaining a user group with the same constraint set result, constructing user attribute expressions and resource attribute expressions based on the conditions that users and resources in the user group need to meet, and generating an initial policy; then, matching the constraints in the initial policy with different cities in the ant colony algorithm, randomly setting the distance values ​​between cities, calculating the possibility of path selection using pheromone concentration and generating different generalization methods, finding the best path to travel through the cities, and using the generalization method to process the policy constraints in turn, and obtaining the optimal generalization method and its corresponding optimal quality policy through evaluation; The final policy results required for mining are selected from the policy set based on the preset policy quality evaluation criteria and the size of the user-authority tuple that the policy can cover. The process of selecting the final policy results required for mining includes: first, performing quality processing on the policy set, and the quality processing includes policy merging and policy simplification processing; then, continuously selecting policies that meet the preset quality evaluation criteria from the processed policy set, and obtaining the final policy results required for mining based on the size of the covered user-authority tuple.

2. The ABAC policy mining method for heterogeneous access control systems according to claim 1 is characterized in that: A user-permission tuple is represented as a triple consisting of user u, resource r, and operation o.

3. The ABAC policy mining method for heterogeneous access control systems according to claim 1 is characterized in that: The optimal generalization method and its corresponding optimal quality strategy are obtained through evaluation, and the generalization method is evaluated by the preset quality function, where the quality function Q policy Expressed as: Q policy (ρ,UP)=|[ρ]∩UP| / WSC(ρ), ρ is the initial policy, UP is the set of user-authority tuples, and WSC() is the preset policy quality evaluation standard.

4. The ABAC policy mining method for heterogeneous access control systems according to claim 3 is characterized in that: When evaluating the generalization method through a preset quality function, the quotient of the quality function value before generalization and the quality function value after generalization is used as the result value of evaluating this generalization. The optimal generalization method and its corresponding optimal quality strategy are found based on the size of the result value.

5. The ABAC policy mining method for heterogeneous access control systems according to claim 3 or 4, characterized in that: In the strategy quality evaluation, the preset strategy quality evaluation criteria are set based on the number of constraints and custom weights and using weighted structural complexity.

6. The ABAC policy mining method for heterogeneous access control systems according to claim 1 is characterized in that: Policy merging processing includes: if the user-permission tuple covered by the current policy is included in the user-permission principle covered by another policy, the current policy is deleted; if the constraints returned by the two policies are the same, a composite policy is generated by merging the attribute relationships of the two policies, provided that the policies are valid.

7. The ABAC policy mining method for heterogeneous access control systems according to claim 1 is characterized in that: Policy simplification processing includes: if the multi-valued attribute of the current policy has attribute value a, (a, b), then the attribute value of the current policy is adjusted to a; if the attribute value of the current policy contains a and the attribute value b must exist, then the attribute value {a, b} of the current policy is removed from the value b; for overlapping attribute values ​​and overlapping operation values ​​in two policies, keep the attribute value and operation value of one policy unchanged, and delete the overlapping attribute value and overlapping operation value in the other policy.

8. An ABAC policy mining system for heterogeneous access control systems, characterized in that: It includes: data extraction module, constraint construction module, strategy acquisition module and strategy output module, among which, The data extraction module is used to extract user-permission tuples and corresponding attribute relationships from the heterogeneous access control system based on the access control list, wherein the user-permission tuple is used to describe the user's operation permissions on the resource; The constraint construction module is used to traverse all attributes of users and resources in the user-permission tuple and obtain the constraint set satisfied by different user attributes and resource attributes by comparing the attribute value relationship; The strategy acquisition module is used to take the constraint set as the input of the ant colony algorithm, obtain the strategy constraints corresponding to the optimal path by simulating different constraints as cities with path distances in the ant colony algorithm, and add the strategy constraints to the strategy set. The strategy constraints corresponding to the optimal path are obtained by simulating different constraints as cities with path distances in the ant colony algorithm. The module includes: first, obtaining a user group with the same constraint set result, constructing user attribute expressions and resource attribute expressions based on the conditions that users and resources in the user group need to meet, and generating an initial strategy; then, matching the constraints in the initial strategy with different cities in the ant colony algorithm, randomly setting the distance values ​​between cities, calculating the possibility of path selection using pheromone concentration and generating different generalization methods, finding the best path to travel through the cities, and using the generalization method to process the strategy constraints in turn, and obtaining the optimal generalization method and its corresponding optimal quality strategy through evaluation; The policy output module is used to select the final policy results required for mining from the policy set based on the preset policy quality evaluation criteria and the size of the user-permission tuple that the policy can cover. The process of selecting the final policy results required for mining includes: first, performing quality processing on the policy set, and the quality processing includes policy merging and policy simplification processing; then, continuously selecting policies that meet the preset quality evaluation criteria from the processed policy set, and obtaining the final policy results required for mining based on the size of the covered user-permission tuple.