Integrated anonymous authentication and hierarchical access method oriented to large model as service

Abstract

The invention relates to the technical field of privacy protection, and discloses a large model as a service-oriented integrated anonymous authentication and hierarchical access method, which comprises the following steps of: generating a server key pair through set system parameters; constructing a random number list of the user; the user registers to the server through the constructed zero-knowledge proof, the commitment value and the applied authority to apply for the member authority, the server issues an anonymous certificate to the user, and the user carries out blind removal on an anonymous signature and then carries out calculation to obtain the certificate; the user blinds the voucher and then sends the voucher to the server for accessing the server, the server judges whether the user is still in a member period or not, if yes, the updated anonymous signature is issued to the user, and after the user conducts blind removal operation on the anonymous signature, the updated voucher is calculated so as to judge whether the updated anonymous signature issued by the server is safe or not. According to the invention, through the non-linkability and anonymity, the authentication information of the same user in different sessions cannot be associated, and the real identity information of the user can be effectively hidden.

Description

Technical Field

[0001] The present invention relates to the field of privacy protection technology, and in particular to an integrated anonymous authentication and hierarchical access method for large model-as-a-service. Background Art

[0002] Large Language Models (LLMs) are widely used in various fields in text understanding and generation tasks, and have become an important tool for improving production efficiency. With the popularity of large language models, the issue of user privacy protection has become increasingly prominent, especially when users interact with models, the risk of privacy leakage has increased significantly. Users usually rely on online AI as a Service (AIaaS) provided by large language model service providers to interact. In this model, the user's identity information, interaction data, and behavior models may be collected and stored by the service provider. Since service providers can build detailed user portraits based on user interaction behaviors, search records and other data, users' privacy security faces severe challenges. Summary of the invention

[0003] The purpose of the present invention is to achieve dual protection characteristics of user identity in the registration and authentication stages, namely, unlinkability and anonymity, through a blind signature mechanism, to ensure that the authentication information of the same user in different sessions cannot be associated, and to effectively hide the user's real identity information, thereby providing an integrated anonymous authentication and hierarchical access method for large models as a service.

[0004] In order to achieve the above-mentioned object of the invention, the embodiment of the present invention provides the following technical solutions: The integrated anonymous authentication and hierarchical access method for large model-as-a-service includes the following steps: Step 1: Generate the server's key pair by setting the system parameters; build the user's random number list; Step 2: The user registers with the server through the constructed zero-knowledge proof to apply for membership rights. The server issues an anonymous certificate to the user, and the user deblinds the anonymous signature and calculates the certificate; Step 3: The user blinds the credential and sends it to the server to access the server. The server determines whether the user is still within the membership period. If so, it issues an updated anonymous signature to the user. After the user unblinds the anonymous signature, it calculates the updated credential to determine whether the updated anonymous signature issued by the server is secure.

[0005] Compared with the prior art, the present invention has the following beneficial effects: anonymity, unlinkability and authentication.

[0006] Anonymity: When a user registers with a server, the server will issue an anonymous signature to the user. Before the user uses the anonymous signature for authentication, the anonymous signature must be deblinded to obtain the credential cred. The blinded cred is used for authentication. The user can complete the registration and authentication operations without revealing his or her true identity.

[0007] Unlinkability: During the registration phase, the system uses zero-knowledge proof technology to protect the user's identity information, making it impossible for external attackers to establish an association with the user. During the authentication process, the system uses an attribute-based anonymous authentication protocol to ensure that attackers cannot track the user's identity through the authentication process; at the same time, the server side uses an anonymous signature protection mechanism to ensure that the same user cannot be identified by analyzing different credentials.

[0008] Authentication: refers to verifying the legitimacy and authenticity of users. In this solution, the server authenticates users to ensure that only legitimate users can access its services, thereby fundamentally protecting the security of the system. Authentication is the core foundation of access control. Only authenticated users can be granted corresponding permissions to access system resources or perform specific operations. This mechanism effectively prevents impersonation and unauthorized access by illegal users, providing an important guarantee for the security and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0011] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0012] It should be noted that similar numbers and letters represent similar items in the following figures, so once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance, or implying any such actual relationship or order between these entities or operations. In addition, the terms "connected", "connected", etc. can be directly connected between elements, or indirectly connected via other elements.

[0013] Embodiment 1: The present invention is achieved through the following technical solutions: Figure 1 As shown, the integrated anonymous authentication and hierarchical access method for large model as a service includes the following steps: Step 1: Generate the server's key pair by setting system parameters; build the user's random number list.

[0014] Step 1-1, set parameters: Given security parameter 1 λ , λ is used to determine the security level of the system and the size of related parameters. For example, λ determines the bit length of the order p of the cyclic group, and can also ensure that the collision resistance of the collision-resistant hash function is strong enough.

[0015] The server (i.e., the large language model server) generates a bilinear mapping e: G1×G2→G T , where G1, G2 and G T are all cyclic groups of order p; the generators of G1 are g and w, and the generators of G2 are .

[0016] Choose a collision-resistant hash function H:{0,1}*→Z p *, where Z p * represents the multiplicative group of integers modulo p, that is, the set {1,2,...,p-1}, where p is a prime number.

[0017] The public parameters are expressed as pp=(e,G1,G2,G T ,p,g,w, ,H).

[0018] Step 1-2, generate a key: Server selects (x,y1,y2,y3,z)←Z p *,calculate: The server generates a private key SK M=(X,y1,y2,y3,z); Server generates public key .

[0019] Step 1-3, generate a random number list: according to the remaining membership days Ld of user Ui, divide the random number list into 12 rows, and the remaining membership days Ld corresponding to each row are [0,30), [30,60), [60,90), ..., [330,360), which are represented by types TP1, TP2, ..., TP12 respectively, and the update time of each row is 1 month, 2 months, ..., 12 months respectively. In the initial state, the pseudo-random number d in the random number list of user Ui sid All are blank.

[0020] Table 1 Random number list Step 2: The user registers with the server through the constructed zero-knowledge proof to apply for membership rights. The server issues an anonymous certificate to the user, and the user de-blinds the anonymous signature and calculates the certificate.

[0021] Step 2-1, user Ui selects a random number t←Z p *, and its own IDi (a number), calculate the commitment value: Constructing a zero-knowledge proof: Among them, Pok is a symbol used to explain the set.

[0022] The permissions applied for by the user are ACLs, which represent the membership permissions applied for by the user, such as one-day membership, monthly membership, quarterly membership, and annual membership; the user pays the amount for the corresponding permissions.

[0023] User Ui will , , ACL is sent to the server.

[0024] Step 2-2, the server receives , , ACL, verify zero-knowledge proof , if the verification is successful, it means that the user is a legitimate user with the corresponding ID. After the zero-knowledge proof verification is passed, it is determined whether the permission applied by user Ui matches the amount paid. If not, it is changed to the permission ACL` corresponding to the actual payment amount of the user, and the access days val corresponding to ACL` is generated; if it matches, the days val corresponding to ACL is generated. Select two random numbers (r,u)←Z p *, calculate the pseudo-random number d sid=H(u,Lt), where Lt is the current registration time, H is a collision-resistant hash function, and RT=(val,Lt). The server then calculates the anonymous signature of user Ui : Right now: The server will ,d sid , RT is sent to user Ui.

[0025] Step 2-3: User Ui receives the anonymous signature After that, it needs to be deblinded, and the user Ui calculates the credential cred: The validity of the signature is verified by checking whether the following equation is true: Among them, e represents the bilinear mapping operation; after verification, user Ui stores cred, d sid , RT.

[0026] Step 3: The user blinds the credential and sends it to the server to access the server. The server determines whether the user is still within the membership period. If so, it issues an updated anonymous signature to the user. After the user unblinds the anonymous signature, it calculates the updated credential to determine whether the updated anonymous signature issued by the server is secure.

[0027] When user Ui accesses the large language model server, he sends the question query and credential cred to the large language model server. After the server verifies the credential cred of user Ui, it sends the answer to the question back to user Ui. In detail: Step 3-1, user Ui selects three random numbers (α, θ, β)←Z p *, and calculate the blinded credential cred`: Calculate the commitment value: Generate a zero-knowledge proof: User Ui will cred`, , ,d sid , RT, query and current authentication time Lt` are sent to the server.

[0028] Step 3-2: After receiving cred, , ,d sid , RT, query, Lt`, calculate Lt`-Lt=d, where d represents the number of days between the authentication time and the registration time, and determine whether d exceeds val; If d exceeds val, the user is reminded that the Ui membership has expired and needs to pay again; If d does not exceed val, then calculate val-d=Ld, where Ld represents the remaining days that user Ui can access the server; if 0≤Ld<30, then check the pseudo-random number d in the type TP1 row in the random number list sid If it exists, it means that the blinded credential cred` has been used and the service is refused; if it does not exist, d sid Join TP1 and provide services; similarly, if 30≤Ld<60, perform the same operation in TP2; and so on; if 330≤Ld<360, perform the same operation in TP12.

[0029] Step 3-3, server verifies zero-knowledge proof ; After verification, the server selects two random numbers (ru`, k)←Z p * Updated sid , d sid * = H(ru`, Lt), d sid * is the updated pseudo-random number; the server calculates the updated anonymous signature : And the answer to the query and the updated anonymous signature ,d sid *, Ld is returned to user Ui.

[0030] Step 3-4: User Ui receives the anonymous signature After that, you need to perform a deblinding operation and update your own credentials ,in: User Ui checks whether the following is true: If true, user Ui stores cred*, d sid *, RT; if it is not true, it means that there is a problem with the anonymous signature issued by the server, such as being tampered with or attacked.

[0031] In summary, when a user registers with a large language model server, the server will anonymously issue an anonymous signature to the user. Before the user uses the anonymous signature for authentication, the anonymous signature will be deblinded to obtain a credential. When using the deblinded credential for authentication, the user will add a zero-knowledge proof to prove the legitimacy of the user's identity without revealing the user's real identity, thereby achieving anonymous authentication and ensuring that no link can be established between the signature used in the authentication process and the user's real identity, thereby effectively preventing the user's identity from being tracked or revealed. In this way, anonymous authentication technology not only ensures the anonymity of the user's identity, but also achieves the unlinkability of the user in different interaction scenarios and improves the authentication efficiency. This design significantly improves the level of user privacy protection, especially in high-privacy risk scenarios such as large model interactions, which can effectively prevent user identity information from being abused or leaked.

[0032] Based on the anonymous authentication mechanism of randomized credentials (such as zero-knowledge proof and anonymous signature), a unique pseudo-identity is generated for each user's visit. The same user uses different identities in different queries. Even if the user's query information is obtained, it cannot be linked to the user's real identity. The user's privacy is protected and the tracking mode and behavior data of the user's identity are prevented from being abused. The authentication mechanism of this method is implemented based on blind signature technology to ensure that the user's identity information and behavior data cannot be linked or tracked when interacting with the large model, which effectively solves the problem that the service provider in the existing large cloud-based language model will reveal the user's real identity. This method dynamically generates unlinkable anonymous credentials, so that each time the user accesses the large model, an anonymous and unassociated authentication identifier is used. The user's privacy data is protected at a higher level, while ensuring the security and credibility of the large model service.

[0033] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An integrated anonymous authentication and hierarchical access method for large models as a service, characterized by: The following steps are involved: Step 1: Generate a server key pair by setting system parameters; Build a random number list for the user; Step 2: The user registers with the server through the constructed zero-knowledge proof to apply for membership rights. The server issues an anonymous certificate to the user, and the user deblinds the anonymous signature and calculates the certificate; Step 3: The user blinds the credential and sends it to the server to access the server. The server determines whether the user is still within the membership period. If so, it issues an updated anonymous signature to the user. After the user unblinds the anonymous signature, it calculates the updated credential to determine whether the updated anonymous signature issued by the server is secure.

2. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 1 is characterized in that: In step 1, the system parameters set include: Given security parameter 1 λ ,λ is used to determine the security level of the system and the size of related parameters; The server generates a bilinear map e: G1×G2→G T , where G1, G2 and G T are all cyclic groups of order p; the generators of G1 are g and w, and the generators of G2 are ; Choose a collision-resistant hash function H:{0,1}*→Z p *, where Z p * represents the multiplicative group of integers modulo p; The public parameters are expressed as pp=(e,G1,G2,G T ,p,g,w, ,H).

3. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 2 is characterized in that: In step 1, the step of generating a key pair of the server by setting the system parameters includes: Server selects (x,y1,y2,y3,z)←Z p *,calculate: The server generates a private key SK M =(X,y1,y2,y3,z); Server generates public key .

4. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 3 is characterized in that: In step 1, the step of constructing a user's random number list includes: According to the remaining membership days Ld of user Ui, the random number list is divided into 12 rows, and the remaining membership days Ld corresponding to each row are [0,30), [30,60), [60,90), ..., [330,360), and TP1, TP2, ..., TP12 are used to represent the type of the interval in which the remaining membership days are located; in the initial state, the pseudo-random number d of each row in the random number list of user Ui sid All are blank.

5. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 4 is characterized in that: In step 2, the user registers with the server through the constructed zero-knowledge proof to apply for membership rights, including: User Ui chooses a random number t←Z p *, and its own IDi; Calculate the commitment value: Constructing a zero-knowledge proof: Among them, Pok is a symbol used to explain the set; The permissions applied for by the user are ACLs, which represent the membership permissions applied for by the user, and the amount paid by the user for the corresponding permissions; User Ui will , , ACL is sent to the server.

6. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 5 is characterized in that: In step 2, the step of the server issuing an anonymous credential to the user includes: The server receives , , ACL, verify zero-knowledge proof , if the verification is successful, it means that the user is a legitimate user with the corresponding ID; After the zero-knowledge proof verification is passed, it is determined whether the permissions applied for by user Ui match the amount paid. If not, it is changed to the permission ACL` corresponding to the actual payment amount of the user, and the access days val corresponding to the ACL` is generated; if it matches, the days val corresponding to the ACL is generated; Choose two random numbers (r,u)←Z p *, calculate the pseudo-random number d sid =H(u,Lt), where Lt is the current registration time, H is the anti-collision hash function, let RT=(val,Lt); The server calculates the anonymous signature of user Ui : The server will d sid , RT is sent to user Ui.

7. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 6 is characterized in that: In step 2, the user performs deblinding on the anonymous signature and then calculates the credential, including: User Ui receives an anonymous signature After that, the deblinding operation is performed and the user Ui calculates the credential cred: The validity of the signature is verified by checking whether the following equation is true: Where, e represents a bilinear mapping operation; After verification, user Ui stores cred, d sid , RT.

8. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 7 is characterized in that: In step 3, the user blinds the credentials and sends them to the server to access the server, including: User Ui selects three random numbers (α, θ, β)←Z p *; Calculate the blinded credential cred`: Calculate the commitment value: Generate a zero-knowledge proof: User Ui will cred`, , d sid , RT, question query and current authentication time Lt` are sent to the server.

9. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 8, characterized in that: In step 3, the server determines whether the user is still within the membership period, and if so, issues an updated anonymous signature to the user, including: The server receives cred`, , d sid , RT, query, Lt`, calculate Lt`-Lt=d, where d represents the number of days between the authentication time and the registration time, and determine whether d exceeds val; If d exceeds val, the user is reminded that the Ui membership has expired and needs to pay again; If d does not exceed val, then calculate val-d=Ld, where Ld represents the remaining days that user Ui can access the server; check the pseudo-random number d in the type row of the remaining days Ld in the random number list sid If it exists, it means that the blinded credential cred` has been used and the service is refused; if it does not exist, d sid Join the category and provide services; Server verifies zero-knowledge proof ; After verification, the server selects two random numbers (ru`, k)←Z p * Updated sid , d sid * = H(ru`, Lt), d sid * is the updated pseudo-random number; The server calculates the updated anonymous signature : And the answer to the query and the updated anonymous signature d sid *, Ld is returned to user Ui.

10. The integrated anonymous authentication and hierarchical access method for large model as a service according to claim 9, characterized in that: In step 3, after the user performs a deblinding operation on the anonymous signature, the user calculates the updated credential to determine whether the updated anonymous signature issued by the server is safe, including: User Ui receives an anonymous signature After that, perform the deblinding operation and update your own credentials ,in: User Ui checks whether the following is true: If true, user Ui stores cred*, d sid *, RT; if this is not true, it means that the anonymous signature issued by the server is not secure.

Images