WO2022099495A1

WO2022099495A1 - Ciphertext search method, system, and device in cloud computing environment

Info

Publication number: WO2022099495A1
Application number: PCT/CN2020/128029
Authority: WO
Inventors: 王树兰; 王凯文; 李采果
Original assignee: 深圳技术大学
Priority date: 2020-11-11
Filing date: 2020-11-11
Publication date: 2022-05-19

Abstract

The present invention provides a ciphertext search method, system, and device in a cloud computing environment, and a storage medium. The method comprises: encrypting a plaintext set by means of a client to obtain ciphertext, obtaining a ciphertext index table according to the ciphertext, randomly generating a user permission table, and uploading the ciphertext structure, user permission table, and ciphertext index table to a cloud server; receiving, by means of the client, a request for applying for a private key of the ciphertext, generating a search trapdoor, and sending same to the cloud server; matching a user attribute with a weight strategy tree, and if the matching succeeds, performing filtering by means of a search keyword and the ciphertext index table to obtain index ciphertext; the cloud server returning the intermediate value of the index ciphertext to the client, and decryption being performed to obtain a search result. The method allows for implementation of an efficient retrieval function for ciphertext data; by means of the characteristics of fully homomorphic encryption, control data of the cloud server is completely blurred, and the confidentiality and efficiency of cloud data processing are greatly improved; moreover, access strategies and users are in many-to-many relationships, and thus, keyword guessing attacks can be resisted.

Description

Ciphertext search method, system and device in cloud computing environment

technical field

The invention belongs to the technical field of data processing, and in particular relates to a ciphertext search method, system, device and storage medium in a cloud computing environment.

Background technique

As a distributed computing technology, cloud computing has almost unlimited computing power and storage space. However, since cloud data is beyond the control of users, data privacy and effective access control must be guaranteed during the use of these data. In order to solve the data privacy problem and realize effective operation, fully homomorphic encryption can be performed on the data, and the fully homomorphic encryption satisfies F(Enc(u))=Enc(F(u)). This feature is in line with the needs of cloud computing models such as cloud computing security and ciphertext retrieval. Therefore, the study of homomorphic encryption has important theoretical significance and application value. In addition, the implementation of the access control policy for cloud data requires the trusted entity approach in traditional access control, but the cloud service provider is no longer trusted, so the encryption party needs to encrypt the data before uploading it. In order to effectively implement access control to encrypted data, the CP-ABE encryption algorithm can be used. Although the classic CP-ABE scheme can achieve fine-grained access, it does not support ciphertext retrieval services. The fully homomorphic encryption scheme can realize ciphertext retrieval. However, it is vulnerable to chosen-plaintext attacks.

(1) Fully homomorphic encryption

Fully homomorphic encryption can perform arbitrary calculations on ciphertext without knowing the key. This special property makes fully homomorphic encryption have a wide range of application requirements, such as cloud computing data privacy security, multi-party computing, ciphertext retrieval Wait. The first fully homomorphic encryption scheme was proposed by Gentry in 2009. Since then, some fully homomorphic encryption schemes based on different difficult problems and some methods to improve the efficiency of fully homomorphic encryption have been proposed.

At present, the efficiency of fully homomorphic encryption is the main problem hindering its development, and the main reason for the low efficiency is that its ciphertext size is too large. Because each homomorphic calculation will cause the increase of ciphertext noise, especially the ciphertext multiplication calculation makes the ciphertext noise grow very fast. Homomorphic operations cannot be performed when the noise exceeds the bounds allowed by correct decryption. Therefore, in order to be able to perform more ciphertext homomorphic operations, large parameters must be set so that the ciphertext has enough space to accommodate noise, which directly leads to a sharp increase in the size of the ciphertext.

(2) Attribute-based encryption

Shamir first proposed the concept of identity-based encryption in 1979, and then many extended concepts of identity-based encryption were proposed, such as hierarchical identity-based encryption, identity-based broadcast encryption, and spatial encryption. Attribute-based encryption is also an extension of identity-based encryption. There are two types of attribute-based encryption: key-policy-based attribute-based encryption (CP-ABE) and ciphertext-policy-based attribute-based encryption (KP-ABE). In ciphertext policy-based attribute-based encryption, the plaintext message is encrypted under a predicate that can be expressed as a logical expression connected by AND, OR, and NOT gates. Each user obtains a key corresponding to a certain attribute set from the attribute authority. Decryption can succeed if and only if the set of attributes satisfies the above predicate. The opposite is true for key-policy-based attribute-based encryption: in this encryption system, the ciphertext corresponds to a set of attributes, and the user key corresponds to a predicate.

Attribute encryption has a wide range of application scenarios in practice, such as access control of distributed file systems, secure online social networks, and efficient broadcast encryption. In addition, most extensions of identity-based encryption can be regarded as a special case of attribute-based encryption. For example, broadcast encryption can be regarded as a special ciphertext policy-based attribute-based encryption. In this encryption system, the access structure is Predicates connected by OR gates. Attribute-based encryption is also an important tool to solve some theoretical problems in identity-based encryption systems. So far, attribute-based encryption has been used to solve the identity revocation problem in identity-based encryption and to construct accountable identity-based encryption schemes. . Due to the importance of attribute-based encryption in theory and practical applications, this encryption system has attracted extensive attention of researchers once it was proposed.

(3) Semantic space model

Computers have a hard time understanding what human language means. This severely limits our ability to communicate instructions to computers, limit their actions to interpret them to us, and limit their ability to analyze and process text. Semantic vector space models (VSMs) are the beginning of dealing with these limitations. The idea of VSM is to represent each document in the collection as a point in the space (a vector in the vector space). The closer the points in the space are, the more similar the semantic similarity is; the farther the points in the space are, the more distant they are semantically. A query of the user is represented as a point in the same space as a document (this query is called a pseudo-document). Documents are sorted by increasing distance from the query and presented to the user. However, there are still many deficiencies in the VSM semantic space, such as document subject classification, keywords, synonyms, etc., which will result in low search efficiency and high precision errors.

technical problem

The technical problem to be solved by the present invention is: aiming at the problems of the prior art, the present invention provides a ciphertext search method in a cloud computing environment.

technical solutions

In a first aspect, an embodiment of the present application provides a method for searching ciphertext in a cloud computing environment, the method comprising:

The client-based encryption party encrypts the plaintext set to obtain the ciphertext structure, obtains the ciphertext index table according to the ciphertext structure, randomly generates the user permission table, and uploads the ciphertext structure, the user permission table and the ciphertext index table to the Cloud server, the user permission table includes at least each user attribute class and a user weight policy tree corresponding to the attribute class, and the plaintext set includes at least one plaintext;

Based on the client receiving the request from the user to apply for the private key of the ciphertext structure, the user generates a corresponding search trapdoor after receiving the private key of the ciphertext structure and sends it to the cloud server, where the search trapdoor at least includes User attributes, search keywords, user private keys;

The cloud server matches the user attribute with the user weight policy tree, and if the user attribute is successfully matched with the user weight policy tree, the search keyword and the ciphertext index table are used for screening. , get the searched index ciphertext;

The cloud server returns the intermediate value of the index ciphertext to the client, and decrypts to obtain a search result.

In the second aspect, an embodiment of the present application provides a ciphertext search system in a cloud computing environment, the system comprising:

Encryption module: used to encrypt the plaintext set based on the client-side encryption party to obtain the ciphertext structure, obtain the ciphertext index table according to the ciphertext structure, randomly generate the user permission table, and combine the ciphertext structure, the user permission table and the ciphertext structure. The text index table is uploaded to the cloud server, the user permission table at least includes each user attribute class and the user weight policy tree corresponding to the attribute class, and the plaintext set includes at least one plaintext;

Generation module: used to receive a request from a user to apply for the private key of the ciphertext structure based on the client, after the user receives the private key of the ciphertext structure, generate a corresponding search trapdoor and send it to the cloud server, the The search trapdoor includes at least user attributes, search keywords, and user private keys;

Search module: used by the cloud server to match the user attribute with the user weight policy tree, and if the user attribute is successfully matched with the user weight policy tree, the search keyword is used to match the password with the password. The text index table is filtered to obtain the searched index ciphertext;

Decryption module: used by the cloud server to return the intermediate value of the index ciphertext to the client, and decrypt to obtain a search result.

In a third aspect, embodiments of the present application further provide a ciphertext search device in a cloud computing environment, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where When the processor executes the computer program, each step in the ciphertext search method in the cloud computing environment according to the first aspect is implemented.

In a fourth aspect, an embodiment of the present application further provides a storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements the method for searching ciphertext in a cloud computing environment as described in the first aspect of the various steps.

beneficial effect

The invention provides a ciphertext search method in a cloud computing environment. The method includes: encrypting a plaintext set based on a client-side encryption party to obtain a ciphertext structure, obtaining a ciphertext index table according to the ciphertext structure, and randomly generating user rights table, and upload the ciphertext structure, user permission table and ciphertext index table to the cloud server, the user permission table at least includes each user attribute class and the user weight policy tree corresponding to the attribute class, and the plaintext set at least Including a plaintext; based on the client receiving the request from the user to apply for the private key of the ciphertext structure, after the user receives the private key of the ciphertext structure, the user generates a corresponding search trapdoor and sends it to the cloud server, and the search The trapdoor at least includes user attributes, search keywords, and user private keys; the cloud server matches the user attributes with the user weight policy tree, and if the user attributes are successfully matched with the user weight policy tree, then The searched index ciphertext is obtained by filtering the search keyword and the ciphertext index table; the cloud server returns the intermediate value of the index ciphertext to the client, and decrypts to obtain a search result. The method realizes the efficient retrieval function of ciphertext data, adopts the weight strategy tree to optimize the access strategy and the latent semantics to optimize the space model, improves the retrieval accuracy through the access control strategy table and the document index table, and reduces the calculation amount of the ciphertext search; Using the characteristics of fully homomorphic encryption, the cloud server control data is completely fuzzed, efficient hiding strategies are realized, and the cloud server computing power is fully utilized to perform homomorphic addition/multiplication operations for access control and ciphertext retrieval, which can achieve efficient data dynamics The update greatly improves the confidentiality and efficiency of cloud data processing; and the access policy has a many-to-many relationship with users. Even if one user betrays, it will not affect other users, and it is based on the characteristics and attribute values of the weighted policy tree. The homomorphic encryption fuzzing is resistant to keyword guessing attacks.

Description of drawings

The specific structure of the present invention will be described in detail below in conjunction with the accompanying drawings

1 is a schematic flowchart of a method for searching ciphertext in a cloud computing environment according to the present invention;

2 is a schematic diagram of a sub-flow of a method for searching ciphertext in a cloud computing environment according to the present invention;

Fig. 3 is another sub-flow schematic diagram of a ciphertext search method in a cloud computing environment of the present invention;

Fig. 4 is another sub-flow schematic diagram of the ciphertext search method in a kind of cloud computing environment of the present invention;

Fig. 5 is another sub-flow schematic diagram of a ciphertext search method in a cloud computing environment of the present invention;

Fig. 6 is another sub-flow schematic diagram of a ciphertext search method in a cloud computing environment of the present invention;

Fig. 7 is another sub-flow schematic diagram of a ciphertext search method in a cloud computing environment of the present invention;

FIG. 8 is a schematic diagram of program modules of a method for searching ciphertext in a cloud computing environment according to the present invention.

Embodiments of the present invention

In order to make the purpose, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. The embodiments described above are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present invention.

1 is a schematic flowchart of a ciphertext search method in a cloud computing environment according to an embodiment of the present application. In this embodiment, the ciphertext search method in the above cloud computing environment includes:

Step 101: Encrypt the plaintext set based on the client to obtain a ciphertext structure, obtain a ciphertext index table according to the ciphertext structure, randomly generate a user authority table, and combine the ciphertext structure, user authority table and ciphertext index. The table is uploaded to the cloud server, the user permission table at least includes each user attribute class and a user weight policy tree corresponding to the attribute class, and the plaintext set includes at least one plaintext.

Among them, the encryption party first encrypts the plaintext set to obtain the ciphertext structure, wherein the plaintext set consists of multiple plaintext documents, and integrates the weight vector information of the keywords generated in the ciphertext structure to obtain a ciphertext index table. The selected attribute feature number randomly generates a user permission table, and the user permission table also includes the attribute class of each user and the user weight policy tree corresponding to the attribute class. Encryption, the number of attribute features selected by the encryption party is to select the user with these attributes to view the encrypted file, wherein the attribute class of the user includes at least the user ID, or other feature values. In this embodiment, only the user ID is proposed, Then upload the ciphertext structure, the ciphertext index table and the user permission table to the cloud server, wherein the information in the ciphertext index table and the user permission table has been subjected to homomorphic fuzzy encryption to form a table.

Step 102: Based on the client receiving the request from the user to apply for the private key of the ciphertext structure, the user receives the private key of the ciphertext structure and generates a corresponding search trapdoor and sends it to the cloud server, and the search trapdoor is sent to the cloud server. The gate includes at least user attributes, search keywords, and user private keys.

Among them, the search party is the user inputting their own attributes and the information of the keyword they want to search to apply for the private key of the ciphertext, the client sends the private key of the ciphertext structure to the user, and the user's local server will generate a search trap The search trapdoor integrates the information entered by the user and the obtained private key. Among them, input the attribute private key SK of the searcher, the searchable strategy tree R of the keyword, and the public key PK parameters, and then output the search trapdoor and upload it to the cloud server. The specific calculation is:

The searcher traverses the strategy tree R, selects the root node of the tree R, and sets it to be t∈G _zp , the local server is set to generate a random value d and store it, and calculate D _pai :

D _pai =D ^t *h ₁ ^d =g ^t(ac-r)/b+ad

By visiting the tree R, starting from the root node of the tree, visiting any node _i of the tree from top to bottom from the tree, select a polynomial qi, and satisfy qi ₌ d-1. Select the root node R ₀ of the tree R, let it be t∈G _zp , let

Assume

To access the set of all leaf nodes in the structure tree, we have

and do the following encryption calculations: Dk and Dk':

First take a random value t _j ∈ G _zp , for the attribute set

Do calculations, with arbitrary keywords

Both have the following calculations, generating Dj and Dj':

Here, a corresponding vector is generated according to the attributes of the searcher user. The vector is used for matching the user policy tree of the user permission table on the cloud server side. Each attribute value of the vector is fuzzy encrypted by the homomorphic algorithm:

V _{(ID, S)} = {*,*,,,*} (* is the value of the corresponding attribute)

Then, generate the user search keyword index vector, which is

_Itf ={*,*,,,,,*} (* is the TF value of the corresponding keyword)

The output format of the search trapdoor STK is:

Step 103: The cloud server matches the user attribute with the user weight policy tree, and if the user attribute is successfully matched with the user weight policy tree, the search keyword and the ciphertext index are used for matching. The table is filtered to obtain the searched index ciphertext.

Among them, 1. Upload the search trapdoor of the searcher, the cloud server CSP matches the user permission table and the ciphertext index table, determines the search range of the user ciphertext set, and calculates the similarity vector of the ciphertext document. The I _idf vector and the I _tf vector in the ciphertext CT and SKT, the corresponding _Itf-idf document vector is obtained, and then the corresponding document matching vector V is obtained through the corresponding LSA latent semantic space model parameter (X),

Document matching vector = V = I _tf-idf ^T *X

The similarity V between the ciphertext document and the search trapdoor is obtained, and the corresponding ciphertext is selected by screening the similarity measure of cosine calculation to generate a set of ciphertexts to be decrypted. The calculation is as follows:

2. Perform a matching operation on the filtered ciphertext set:

If y is a leaf node in the access policy A in the ciphertext CT, define j=att(y). For each attribute j∈A, calculate its intermediate value E _y .

If y is a non-leaf node of A, then define S _Z to be a set of nodes z of any size k _Z , and then calculate the intermediate value E _y as follows.

If y is the root node, then the intermediate value E is returned as follows.

3. If x is the leaf node of the searchable policy tree R, let w=key(x) be the key associated with the hash function, and for each attribute x∈R, calculate its intermediate value E _y' .

If x is a non-leaf node of R, then define S _Z' as a set of child nodes z' of any size of k _Z' , and then calculate the intermediate value E _y' as follows.

If x is the root node, then the intermediate value E is returned as follows.

4. According to the two values, E _y and E _y' , perform the corresponding formula operation and return the intermediate result

Select the next ciphertext that meets the similarity filter and start the calculation from step 2. The formula is as follows:

Step 104: The cloud server returns the intermediate value of the index ciphertext to the client, and decrypts to obtain a search result.

Among them, in step 103, the intermediate result is returned

After that, perform the decryption calculation:

An embodiment of the present application provides a ciphertext search method in a cloud computing environment. The method includes: encrypting a plaintext set based on a client-side encryption party to obtain a ciphertext structure, obtaining a ciphertext index table according to the ciphertext structure, and randomly generating a ciphertext index table. User permission table, upload the ciphertext structure, user permission table and ciphertext index table to the cloud server, the user permission table at least includes each user attribute class and the user weight policy tree corresponding to the attribute class, the description The corpus includes at least one plaintext; based on the client receiving the request from the user to apply for the private key of the ciphertext, the user receives the private key of the ciphertext structure and generates a corresponding search trapdoor and sends it to the cloud server. The search trapdoor at least includes user attributes, search keywords, and user private keys; the cloud server matches the user attributes with the user weight policy tree, and if the user attributes are successfully matched with the user weight policy tree , then filter the search keyword and the ciphertext index table to obtain the searched index ciphertext; the cloud server returns the intermediate value of the index ciphertext to the client, and decrypts to obtain the search result . The method realizes the efficient retrieval function of ciphertext data, adopts the weight strategy tree to optimize the access strategy and the latent semantics to optimize the space model, improves the retrieval accuracy through the access control strategy table and the document index table, and reduces the calculation amount of the ciphertext search; Using the characteristics of fully homomorphic encryption, the cloud server control data is completely fuzzed, efficient hiding strategies are realized, and the cloud server computing power is fully utilized to perform homomorphic addition/multiplication operations for access control and ciphertext retrieval, which can achieve efficient data dynamics The update greatly improves the confidentiality and efficiency of cloud data processing; and the access policy has a many-to-many relationship with users. Even if one user betrays, it will not affect other users, and it is based on the characteristics and attribute values of the weighted policy tree. The homomorphic encryption fuzzing is resistant to keyword guessing attacks.

Specifically, based on the foregoing embodiment, referring to FIG. 2 , FIG. 2 is a schematic diagram of a sub-flow of the ciphertext search method in the cloud computing environment of the present application. In this embodiment, the client-based encryption party encrypts the plaintext set to obtain the ciphertext structure. Specific steps include:

First perform attribute-based encryption on the plaintext;

Construct the TF vector of the keyword in the plaintext and the IDF vector of the keyword in the plaintext;

calculating the TF-IDF vector of the keywords in the plaintext set;

Perform latent semantic SVD dimension reduction calculation on the plaintext set to obtain a vector space model and I _idf ;

Homomorphic encryption The vector space model and I _idf generate the corresponding ciphertext structure.

Among them, construct the TF vector of the keywords in the plaintext and the keys in the plaintext

(where m _j is the number of words in document d _j , and n _ij is the number of words in which keyword t _i appears.)

(where |D| is the total number of documents,

The IDF for the number of documents in which the keyword t _i appears):

Calculate the TF-IDF vector _Atf-idf of the keywords in the plaintext set;

LSA latent semantic space model: A _tf-idf = USV ^T ,

LSA vector space model parameters: X=US ^T

First, encrypt the plaintext by the encryption party, select r ₀ ∈ G _zp and then calculate C and

The access tree T is constructed by visiting the structure A, starting from the root node of the tree, visiting any node _i of the tree from top to bottom, selecting a polynomial qi, and satisfying qi ₌ d-1. For the root node T ₀ of the tree, there is r ₀ ∈ G _zp , let

Let Y be the set of all leaf nodes in the access structure tree T, with y∈Y and perform the following encryption calculations: Cy and Cy'

Calculate the generated keyword set W in the semantic model, take a random value r _i ∈ G _zp for any keyword w ∈ W, have the following calculations to generate Cw and Cw'

The corresponding TF vector is constructed by the encryption party for its plaintext set, and the index matching vector of the corresponding document is taken as:

_Itf ={*,*,,,*} (* is the value of the corresponding key)

At the same time, the index vectors corresponding to all documents are integrated into the document index table in Table 2 on the cloud server side. Finally, the output format of the ciphertext CT is:

Here, the encryption party performs the latent semantic SVD dimension reduction calculation on the optimized weight access policy tree and determines the encrypted plaintext set (where the keyword weight is the TF-IDF weight) on the local server. The vector space model LSA and I _idf are homomorphic After encrypting to generate the corresponding ciphertext structure, upload it to the cloud server. LSA (latent semantic analysis), also known as LSI (latent semantic index), is a new indexing and retrieval method proposed by Scott Deerwester, Susan T. Dumais and others in 1990. This method, like the traditional vector space model (VSM), uses vectors to represent words (terms) and documents (documents), and judges the relationship between words and documents through the relationship between vectors (such as included angles); the difference is that LSA Mapping words and documents into the latent semantic space removes some "noise" in the original vector space and improves the accuracy of information retrieval. By generating a vocabulary base, a vocabulary-text matrix is formed (using TF-IDF to weight the word frequency). Each row in the initial matrix corresponds to a word, and each column corresponds to an article. M words and N articles can be expressed as the following M*N matrix, and then perform singular value decomposition in the figure, and reduce the matrix after SVD decomposition. dimension to construct the latent semantic space. The advantages of LSA are: low-dimensional spatial representation can describe synonyms, and synonyms will correspond to the same or similar topics; dimensionality reduction can remove some noise and make features more obvious; make full use of redundant data; unsupervised/complete automation; language-independent , strong practicability.

Specifically, based on the above embodiment, referring to FIG. 3 , FIG. 3 is a schematic diagram of another sub-flow of the ciphertext search method in the cloud computing environment of the application. In this embodiment, the specific steps of randomly generating a user permission table include:

Homomorphically encrypt the optimized user weight policy tree, where the user weight policy tree at least includes the number of attribute features selected by the encryption party;

Generate the optimized topic policy tree ciphertext, attribute class corresponding weight ciphertext, policy weight corresponding ciphertext set ciphertext.

Among them, after the optimized user weight policy tree is optimized, it not only reduces the storage cost of ciphertext, but also reduces the computational cost of encryption. After the optimized user weight policy tree is homomorphically encrypted, the three parts are Perform a homomorphic matching operation.

Specifically, based on the above embodiment, referring to FIG. 4 , FIG. 4 is a schematic diagram of another sub-flow of the ciphertext search method in the cloud computing environment of the present application. In this embodiment, the cloud server compares the user attributes with the The specific steps for matching the user weight policy tree include:

After performing the homomorphic algorithm fuzzy encryption on the user attribute, match it with the corresponding weight ciphertext of the attribute class;

If the matching is successful, the user attribute is then matched with the ciphertext of the topic policy tree to determine the authority of the user and the scope of the searchable ciphertext;

The ciphertexts of the ciphertext sets corresponding to the policy weights are then matched to lock the search range of the ciphertexts.

Among them, the searcher user uploads the attribute trapdoor vector of the search application. After the vector is also fuzzy encrypted by the homomorphic algorithm, the cloud server first performs the matching calculation of the attribute class corresponding weight ciphertext on the attribute ciphertext of the user. The approximate calculation principle is as follows:

Among them, in homomorphic encryption, the encryption party randomly generates large prime numbers r', p, q, and obtains the public parameter r'p, N=pq, hm.CT _attribute is the attribute trapdoor vector of the search party, hm.CT _DU is the encryption party's attribute trapdoor vector The attribute class of corresponds to the weight ciphertext,

Value=((hm.CT _attribute -hm.CT _DU )r'p)mod N

=((M _attribute -M _token )r'p+2 ^k r'pq(r _attribute -r _token ))mod N

=(M _attribute -M _token )r'p

It can be seen that since r'p is not 0, if Value=0, it means that the matching is successful, that is, the class corresponding to the attribute of the user exists in the table of the cloud server, and then through similar calculation, it can be concluded that the attribute is in the attribute The weight value within the class Value=ω _attribute .

The cloud server matches the ciphertext of the policy attribute tree, and performs tree sum/or calculation on the determined user attribute class after matching. And/or computations are implemented by homomorphic addition and subtraction operations. Multiply the ciphertext of the policy attribute tree after the operation with the multi-attribute weight value set Value _all = {ω ₁ , , ω _n }, and then match the ciphertext of the ciphertext set corresponding to the policy weight, and return Value=W _search , the Value value is the set ciphertext collection index, and the value is passed to the document index table for the next ciphertext retrieval.

In this process, in the process of transmission, storage, retrieval and processing of user data, except for the user's locality, no other entities have access to the user's plaintext data and its intermediate processing results. In addition, users do not need to obtain the private key p for homomorphic algorithm decryption, but only need to upload the product of random numbers r and p and N=pq used to encrypt the data, and use these two public parameters to perform homomorphic fuzzy encryption , since both r and p are large prime numbers, their product is also computationally intractable, thus ensuring that the private key p is not at risk of being leaked.

Specifically, based on the above embodiment, referring to FIG. 5 , FIG. 5 is a schematic diagram of another sub-flow of the ciphertext search method in the cloud computing environment of the present application. In this embodiment, if the user attribute and the user weight policy tree If the match is successful, the specific steps of obtaining the searched index ciphertext by filtering the keyword and the ciphertext index table include:

After the user attribute is successfully matched with the user weight policy tree;

The ciphertext correlation screening is performed between the search keywords and related parameters in the trapdoor and the ciphertext index table to obtain the searched index ciphertext, and the ciphertext index table at least includes the keyword vector in the ciphertext. .

Among them, after the user attributes are successfully matched with the user weight policy tree, the keywords searched by the users in the trapdoor will be filtered in the keywords in the ciphertext index table to obtain the most similar index ciphertext.

Specifically, based on the above-mentioned embodiment, referring to FIG. 6 , FIG. 6 is a schematic diagram of another sub-flow of the ciphertext search method in the cloud computing environment of the application. The specific steps of requesting the private key of the ciphertext structure include:

Generate the public key and the master private key used to generate the private key based on the client;

The user's private key is obtained based on the public key, the master private key, the user ID, and the user attribute.

where construct G ₀ is a bilinear group of order prime p, and

η is a random value. Let g be its generator, and the bilinear map e:G ₀ ×G ₀ →G _r defines two hash functions: H ₀ :{0,1} ^* →G ₀ and H ₁ :{0,1 } ^* →G _zp . Three random numbers a, b, c ∈ G _zp are selected in the group G _zp . The homomorphic encryption algorithm generates p, q, r is a random prime value, p is a private key, and generates a search public key parameter: {N=pq, RQ=rq}. Output the public key PK and the master key MK, where the public key PK contains the set of random value functions required by the homomorphic encryption algorithm.

MK={a, b, c}.

Enter the public key PK, the master private key MSK, the ID of the user applying for search, and the attribute set S of the searcher DU, and then output the attribute private key SK of the searcher. Take a random number r∈G _zp , and for each attribute in the attribute set S

A random number tj∈G _zp is selected, and the encrypted ID _t corresponding to the timestamp is assigned. Calculated to get:

SK _{(S, ID)} = {S, D = h ₄ ^(ac-r) , ID _t

Specifically, based on the foregoing embodiment, referring to FIG. 7 , FIG. 7 is a schematic diagram of another sub-flow of the ciphertext search method in the cloud computing environment of the present application. In this embodiment, the ciphertext search method in the cloud computing environment further includes:

When the encrypted data is deleted based on the client, by changing the access structure of the data;

After determining that the encrypted data is deleted, the cloud server returns the deleted file to the client.

Among them, when the encryptor wants to delete a certain type of encrypted data, the access structure of the file is changed by revoking the access permission attribute through the homomorphic operation. After the delete operation is executed, the cloud server will determine whether the current file has been deleted, and will return a Delete the file to the encryption party. Latent Semantic Space Model LSA does not support data update and is vulnerable to keyword guessing attacks, so the computing power of cloud servers and homomorphic encryption are used to solve this problem. The vector space model adopted in the scheme relies on tf-idf weights, where the inverse document frequency (idf) factor depends on the number of documents containing keywords. The idf factor of a keyword may change when files are added or removed. To avoid updating all searchable indexes when an update occurs, the document vectors should be independent of each other. Since the searchable index is built for each file, a possible solution would be to just store the tf value in the file vector and add another auxiliary vector to store the idf value for each key. This way the update is limited to the auxiliary vector, not all searchable indices. The cost is that during user search requests, tf-idf weights need to be computed to obtain relevance scores. Since the computation is on the server side, and the computing power on the server side is high, the overall efficiency is hardly affected by the update. Moreover, after using the homomorphic algorithm for encryption, guessing keyword attacks are avoided, and the whole process of the homomorphic operation is black-boxed, and there is no possibility of revealing the private key.

Among them, the realization of the homomorphic algorithm:

1. HOMO.Encrypt(PK,M',LSA)→CT': Homomorphic encryption algorithm, input the public parameter PK, the data owner DO generates the determined weight policy tree vector, semantic model parameters and document tf vector,

[ω is the number of documents]. Among them, q and r are random prime values, p is the private key, and the encrypted content is represented by binary bits, M'∈{0,1}. get encrypted ciphertext CT'

CT'={pq+ ^2k rq+M'}

The encryption algorithm here is an optimization of the initial algorithm. In view of the large improvement of the ciphertext generated by the original algorithm, a multi-bit binary is used to reduce the ciphertext size. The k power of 2 in the encryption formula represents the degree of bit reduction, so that The amount of calculation is greatly reduced, and in order to make the ciphertext noise too large after multiple homomorphic operations, the modulo exchange technology is used, that is, after each calculation of the ciphertext, it is multiplied by a decimal to reduce noise and control An increase in noise in the ciphertext.

Assuming that the modulo q is V ^j and the noise of both ciphertexts is V, then the noise is greater than V ² after the homomorphic multiplication operation, and the noise after the multiplication operation at the logj layer reaches the threshold, and the next step cannot be calculated. So to solve this situation, multiply each multiplication by 1/v. In the first operation the noise is X ² and then multiplied by 1/v, so the noise is reduced.

2. Calculate(CT', f _(update) )→CT ^* : The cloud server CSP inputs the specified CT', and the function f _(update) corresponding to the homomorphic operation sent by the data owner can realize the addition and synchronization of the ciphertext State and multiplication homomorphism calculation, perform attribute update operation on encrypted ciphertext CT' and dynamically modify parameters in LSA model in ciphertext.

The corresponding function operations here are: weight attribute strategy tree matching; weight calculation for tf-idf; latent semantic (lsa) calculation; dynamic modification of the value or weight of attribute strategy tree, user ID, and related document vectors.

The principle of homomorphic operation is as follows:

Homomorphism includes additive and multiplicative homomorphism, there are two ciphertexts c ₁ =m ₁ +2 ^k r ₁ q+pq and c ₂ =m ₂ +2 ^k r ₂ q+pq

Homomorphic addition correctness proof:

((c ₁ +c ₂ )mod p)mod 2 ^k =

[((m ₁ +m ₂ )+2 ^k q(r ₁ +r ₂ )+pq)mod p]mod 2 ^k =

((m ₁ +m ₂ )+2 ^k q(r ₁ +r ₂ ))mod 2 ^k =m ₁ +m ₂

Homomorphic multiplication correctness proof:

((c ₁ *c ₂ )mod p)mod 2 ^k =[((m ₁ +2 ^k r ₁ q)(m ₂ +2 ^k r ₂ q)+

p((m ₁ +2 ^k r ₁ q)+(m ₂ +2 ^k r ₂ q))q+(pq) ² )mod p]mod 2 ^k =

((m ₁ +2 ^k r ₁ q)(m ₂ +2 ^k r ₂ q))mod 2 ^k =

(m ₁ m ₂ +2 ^k (m ₁ r ₂ +m ₂ r ₁ +2 ^k r ₁ r ₂ )q)mod 2 ^k =m ₁ m ₂

However, in the above scheme, if pq is used as the public key, the private key p can be easily discovered, so the greatest common divisor problem is introduced into the above encryption algorithm, that is, some ciphertexts obtained by encrypting the plaintext with 0 are added.

{x _i :x _i =2 ⁿ r _i +pq _i }

Considering this set as a public key, and randomly selecting some subsets from this set and adding them to the encryption algorithm during encryption, the scheme is secure. Because the ciphertext of 0 is added, it has no effect on decryption.

Further, an embodiment of the present application also provides a ciphertext search device 200 in a cloud computing environment. Referring to FIG. 8 , FIG. 8 is a schematic diagram of a ciphertext search device module in a cloud computing environment in an embodiment of the present application. , the ciphertext search device 200 in the above cloud computing environment includes:

Encryption module 801: used to encrypt the plaintext set based on the client's encryption party to obtain a ciphertext structure, obtain a ciphertext index table according to the ciphertext structure, randomly generate a user authority table, and combine the ciphertext structure, user authority table and The ciphertext index table is uploaded to the cloud server, the user permission table at least includes each user attribute class and the user weight policy tree corresponding to the attribute class, and the plaintext set includes at least one plaintext;

Generation module 802: for receiving a request from a user to apply for the private key of the ciphertext based on the client, after the user receives the private key of the ciphertext structure, generates a corresponding search trapdoor and sends it to the cloud server. The search trapdoor at least includes user attributes, search keywords, and user private keys;

Search module 803: used by the cloud server to match the user attribute with the user weight policy tree, and if the user attribute is successfully matched with the user weight policy tree, then match the user attribute with the user weight policy tree through the search keyword. The ciphertext index table is filtered to obtain the searched index ciphertext;

Decryption module 804: used by the cloud server to return the intermediate value of the index ciphertext to the client, and decrypt to obtain a search result.

A ciphertext search device 200 in a cloud computing environment provided by an embodiment of the present application can implement: a client-based encryption party encrypts a plaintext set to obtain a ciphertext structure, obtains a ciphertext index table according to the ciphertext structure, and randomly generates User permission table, upload the ciphertext structure, user permission table and ciphertext index table to the cloud server, the user permission table at least includes each user attribute class and the user weight policy tree corresponding to the attribute class, the description The corpus includes at least one plaintext; based on the client receiving the request from the user to apply for the private key of the ciphertext, the user receives the private key of the ciphertext structure and generates a corresponding search trapdoor and sends it to the cloud server. The search trapdoor at least includes user attributes, search keywords, and user private keys; the cloud server matches the user attributes with the user weight policy tree, and if the user attributes are successfully matched with the user weight policy tree , then filter the search keyword and the ciphertext index table to obtain the searched index ciphertext; the cloud server returns the intermediate value of the index ciphertext to the client, and decrypts to obtain the search result . The method realizes the efficient retrieval function of ciphertext data, adopts the weight strategy tree to optimize the access strategy, and optimizes the space model of latent semantics, improves the retrieval accuracy through the access control strategy table and the document index table, and reduces the calculation amount of the ciphertext search; Using the characteristics of fully homomorphic encryption, the cloud server control data is completely fuzzed, efficient hiding strategies are realized, and the computing power of the cloud server is fully utilized to perform homomorphic addition/multiplication operations for access control and ciphertext retrieval, which can achieve efficient data dynamics The update greatly improves the confidentiality and efficiency of cloud data processing; and the access policy has a many-to-many relationship with users. Even if one user betrays, it will not affect other users, and it is based on the characteristics and attribute values of the weighted policy tree. The homomorphic encryption fuzzing is resistant to keyword guessing attacks.

Further, an embodiment of the present application also provides a ciphertext search device in a cloud computing environment, including a memory, a processor, and a computer program stored in the memory and running on the processor, the processing When the computer executes the computer program, each step in the above-mentioned ciphertext search method in a cloud computing environment is implemented.

Further, the present application also provides a storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements each step in the above-mentioned ciphertext search method in a cloud computing environment.

Each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically alone, or two or more modules may be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware, and can also be implemented in the form of software function modules. The integrated modules, if implemented in the form of software functional modules and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .

It should be noted that, for the convenience of description, the foregoing method embodiments are all expressed as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence. As in accordance with the present invention, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily all necessary to the present invention.

In the above-mentioned embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

The above is a description of the ciphertext search method, system, device and storage medium in a cloud computing environment provided by the present invention. There will be changes in the above, and in conclusion, the content of this specification should not be construed as a limitation to the present invention.

Claims

A ciphertext search method in a cloud computing environment, wherein the ciphertext search system includes a client and a cloud server, and the method includes:

The client-based encryption party encrypts the plaintext set to obtain the ciphertext structure, obtains the ciphertext index table according to the ciphertext structure, randomly generates the user permission table, and uploads the ciphertext structure, the user permission table and the ciphertext index table to the Cloud server, the user permission table includes at least each user attribute class and a user weight policy tree corresponding to the attribute class, and the plaintext set includes at least one plaintext;

Based on the client receiving the request from the user to apply for the private key of the ciphertext structure, the user generates a corresponding search trapdoor after receiving the private key of the ciphertext structure and sends it to the cloud server, where the search trapdoor at least includes User attributes, search keywords, user private keys;

The cloud server matches the user attribute with the user weight policy tree, and if the user attribute is successfully matched with the user weight policy tree, the search keyword and the ciphertext index table are used for screening. , get the searched index ciphertext;

The cloud server returns the intermediate value of the index ciphertext to the client, and decrypts to obtain a search result.
The method according to claim 1, wherein the encryption of the plaintext set by the client-based encryption party to obtain the ciphertext structure specifically includes:

First perform attribute-based encryption on the plaintext;

Construct the TF vector of the keyword in the plaintext and the IDF vector of the keyword in the plaintext;

calculating the TF-IDF vector of the keywords in the plaintext set;

Perform latent semantic SVD dimension reduction calculation on the plaintext set to obtain a vector space model and I idf ;

Homomorphic encryption The vector space model and I idf generate the corresponding ciphertext structure.
The method according to claim 2, wherein the randomly generated user permission table specifically comprises:

Homomorphically encrypt the optimized user weight policy tree, where the user weight policy tree at least includes the number of attribute features selected by the encryption party;

Generate the optimized topic policy tree ciphertext, attribute class corresponding weight ciphertext, policy weight corresponding ciphertext set ciphertext.
The method according to claim 3, wherein, the cloud server matching the user attribute with the user weight policy tree specifically includes:

After performing the homomorphic algorithm fuzzy encryption on the user attribute, match it with the corresponding weight ciphertext of the attribute class;

If the matching is successful, the user attribute is then matched with the ciphertext of the topic policy tree to determine the authority of the user and the scope of the searchable ciphertext;

The ciphertexts of the ciphertext sets corresponding to the policy weights are then matched to lock the search range of the ciphertexts.
The method according to claim 4, wherein, if the user attribute is successfully matched with the user weight policy tree, the keyword and the ciphertext index table are filtered to obtain the search index The ciphertext specifically includes:

After the user attribute is successfully matched with the user weight policy tree;

The ciphertext correlation screening is performed between the search keywords and related parameters in the trapdoor and the ciphertext index table to obtain the searched index ciphertext, and the ciphertext index table at least includes the keyword vector in the ciphertext. .
The method according to claim 5, wherein the receiving, by the client, a request from a searcher user to apply for the private key of the ciphertext structure specifically includes:

Generate the public key and the master private key used to generate the private key based on the client;

The user's private key is obtained based on the public key, the master private key, the user ID, and the user attribute.
The method of claim 1, further comprising:

When the encrypted data is deleted based on the client, by changing the access structure of the data;

After the cloud server determines that the encrypted data is deleted, it returns to the client to delete the file.
A ciphertext search system in a cloud computing environment, characterized in that the system includes:

Encryption module: used to encrypt the plaintext set based on the client-side encryption party to obtain the ciphertext structure, obtain the ciphertext index table according to the ciphertext structure, randomly generate the user permission table, and combine the ciphertext structure, the user permission table and the ciphertext structure. The text index table is uploaded to the cloud server, the user permission table at least includes each user attribute class and the user weight policy tree corresponding to the attribute class, and the plaintext set includes at least one plaintext;

Generation module: used to receive a request from a user to apply for the private key of the ciphertext structure based on the client, after the user receives the private key of the ciphertext structure, generate a corresponding search trapdoor and send it to the cloud server, the The search trapdoor includes at least user attributes, search keywords, and user private keys;

Search module: used by the cloud server to match the user attribute with the user weight policy tree, and if the user attribute is successfully matched with the user weight policy tree, the search keyword is used to match the password with the password. The text index table is filtered to obtain the searched index ciphertext;

Decryption module: used by the cloud server to return the intermediate value of the index ciphertext to the client, and decrypt to obtain a search result.
A ciphertext search device in a cloud computing environment, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that the processor executes the computer When the program is executed, each step in the ciphertext search method in the cloud computing environment according to any one of claims 1 to 7 is implemented.
A storage medium on which a computer program is stored, characterized in that, when the computer program is executed by a processor, the method in the ciphertext search method in the cloud computing environment according to any one of claims 1 to 7 is realized. each step.