A verifiable dense-state data fusion method

By constructing a encrypted data fusion model and utilizing key initialization and homomorphic encryption algorithms on a polynomial quotient ring, the problems of low security and efficiency in the data fusion process of traditional encryption technologies are solved, achieving efficient and secure encrypted data fusion, which is suitable for peer-to-peer and cloud-based scenarios.

CN117909997BActive Publication Date: 2026-11-17NO 30 INST OF CHINA ELECTRONIC TECH GRP CORP
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202311698165.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-12
Publication Date
2026-11-17
Estimated Expiration
2043-12-12

AI Technical Summary

Technical Problem

Traditional encryption technologies suffer from low security and low efficiency during data fusion, especially during post-decryption fusion analysis, where they are vulnerable to man-in-the-middle attacks and reduce system availability.

Method used

A verifiable encrypted data fusion method is adopted. By constructing an encrypted data fusion model, key initialization on a polynomial quotient ring and homomorphic encryption algorithm are used to realize point-to-point decentralized and cloud-based fusion computing. It supports distributed collaborative computing of encrypted data and ensures data security and correctness through authentication code verification.

Benefits of technology

It enables efficient fusion computation of encrypted data without decryption, ensuring data security and availability, preventing man-in-the-middle attacks, reducing cross-domain collaboration costs, and supporting distributed collaborative computing and privacy protection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117909997B_ABST
    Figure CN117909997B_ABST
Patent Text Reader

Abstract

This invention provides a verifiable encrypted data fusion method, comprising: S100, constructing an encrypted data fusion model; S200, performing encrypted data fusion based on the encrypted data fusion model; S201, key initialization stage; S202, data encryption and authentication code generation stage; S203, encrypted data fusion calculation stage; and S204, encrypted data decryption stage. This invention allows institutions or individuals to control their own data privacy, supports distributed collaborative computation directly on encrypted text, prevents malicious behavior by participating parties, achieves verifiable encrypted text fusion computation among a group of participants, ensures data security and the authenticity and trustworthiness of information, and reduces the cost of cross-domain and cross-departmental collaboration. Its computational characteristics mainly include distributed collaborative computation, input privacy, and robustness.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of data encryption technology, and more specifically, to a verifiable encrypted data fusion method. Background Technology

[0002] Currently, a number of "city brains" have emerged in fields such as e-government, digital healthcare, and ride-hailing. While highlighting applications like "at most one visit," "data travels more," and "precision services" have brought benefits to the public, they have also generated a large amount of public and personal data. Driven by profit, external attacks are frequent, with adversaries often using vulnerability exploitation and protection bypass to infiltrate the internal networks of enterprises or organizations to steal or destroy data. While external data security risks continue to escalate, internal data security threats are becoming increasingly severe due to intentional or unintentional actions by administrators, illegal use of privacy data, and improper data usage. For example, over 200 million user information records from a telecommunications company were sold; Beniao Social leaked 400GB of data containing over 300 million user records; Amazon was heavily fined by the EU for violating data privacy regulations; and Didi's app illegally collected personal information. These incidents, along with Didi's IPO in the US, have raised concerns about the leakage of citizens' behavioral data.

[0003] Data encryption is an effective means of data protection, but encrypted data is difficult to analyze and statistically process directly like plaintext, reducing data usability. Traditional encryption technologies face the following problems: downloading encrypted data to the user's end and then decrypting and analyzing it leads to low efficiency. Decrypting encrypted data on the server side and then fusing it for analysis reduces security. Furthermore, directly fusing encrypted data makes it vulnerable to man-in-the-middle attacks, including tampering and replacement, rendering the encrypted results unusable and compromising system availability. Summary of the Invention

[0004] The present invention aims to provide a verifiable encrypted data fusion method to solve the problems existing in the above-mentioned traditional encryption technology.

[0005] The present invention provides a verifiable dense-state data fusion method, comprising: S100, construct a dense-state data fusion model; S200, based on the dense data fusion model, performs dense data fusion: S201, Key initialization phase; S202, Data encryption and authentication code generation stage; S203, computational dense-state data fusion stage; S204, the stage of decrypting fused encrypted data.

[0006] Furthermore, the encrypted data fusion model supports both point-to-point decentralized fusion computing and cloud-based fusion computing. Point-to-point decentralized fusion computing is executed in real time by the computing party and does not store encrypted data from any party. Cloud-based fusion computing supports encrypted storage from all parties and initiates asynchronous fusion processing by the computing party.

[0007] Furthermore, step S201 includes: (1-1) Data encryption key initialization: (1-1-1) Each participant generates an initial encryption key: n senders each hold a unique private key sk, composed of elements uniformly sampled from a polynomial quotient ring. i i = 1, 2, ..., n; corresponding public key pk i (pk i =(p0, p1)=(-p sk i +e0, p), where p and a are common parameters uniformly sampled from the polynomial quotient ring, and e0 is an element sampled Gaussian from the polynomial quotient ring; The receiver holds the private key sk, which was generated in the same way as each sender. r PK with public key r ; (1-1-2) Each participating party generates a secret shared public key pk based on the secret sharing method. s (pk s =(Σpk i ,p)); Each participant uses its own private key sk i (i=1,2,…,n), generate the corresponding computational public key H. 0,i H 1,i (i=1,2,…,n), H 0,i =-a r i +sk i P+e 0,i H 1,i =a sk i +e 1,i Where P is the modulus on the polynomial quotient ring, and r i 'a' is a common parameter uniformly sampled from the polynomial quotient ring, and 'e' is a common parameter. 0,i e 1,i These are elements sampled from the Gaussian ring of the polynomial quotient; the public key H will be calculated. 0,i H 1,i Sending the public keys H0 and H1 to the key management aggregation point, where H0 = ΣH 0,i H1=ΣH 1,i, i=1,2,…,n; (1-1-3) Each participating party receives the public keys H0 and H1 collected by the computing party; and uses the private key sk i And calculate H using public keys H0 and H1 0,i '、H 1,i ' , H 0,i '=sk i H0+e 2,i H 1,i '=(r i -sk i H1+e 3,i , where e 2,i e 3,i These are elements sampled from the Gaussian ring of the polynomial quotient; the result H is then calculated again. 0,i '、H 1,i 'Send to key management to collect H0' and H1', ​​H0'=ΣH 0,i ',H1'=ΣH 1,i '; and further calculate the re-linearized public key rlk(rlk=(H0'+H1',H1)), which is used to reduce the size of the ciphertext after the secret state calculation; (1-1-4) Each sender collaboratively generates a re-linearized public key rlk; first, each participant uses its own private key sk i Generate the corresponding computational public key H 0,i H 1,i H 0,i =-a r i +sk i P+e 0,i H 1,i =a sk i +e 1,i Where P is the modulus on the polynomial quotient ring, r i 'a' is a common parameter uniformly sampled from the polynomial quotient ring, and 'e' is a common parameter. 0,i e 1,i These are elements sampled from a Gaussian ring of polynomials; the key management center aggregates and calculates H0 and H1, and then each participant uses their private key sk. i And H is calculated from H0 and H1 0,i '、H 1,i The key management center recalculates H0' and H1', ​​and finally, all parties calculate and linearize the public key rlk (rlk=(r0, r1)=(H0'+H1', H1)=(-a sk+sk sk P+e1,a), where P is the modulus on the polynomial quotient ring; p and a are common parameters uniformly sampled from the polynomial quotient ring, and e1 is an element sampled from the polynomial quotient ring by Gaussian sampling, used to reduce the size of the ciphertext after dense state calculation. (1-2) Data verification key initialization: Each participant obtains the same authentication key k from the key management center based on the same pseudo-random function F.

[0008] Furthermore, step S202 includes: (2-1) Data encryption: (2-1-1) Each sender encodes the plaintext into an element Δ on the polynomial quotient ring. m ( Ecd (z,Δ)→Δ m ), where 1 / Δ represents the precision that the plaintext needs to retain, so multiplying by Δ in plaintext encoding controls the precision; (2-1-2) Each sender uses the shared public key pk from the initialization phase. s The plaintext encoding result Δ m Encrypt to ciphertext c(c= (c i,0 c i,1 )= (Δ m +v i pk s +e 2,i ,v i p1+e 3,i ) mod q L , where v i It is an element uniformly sampled on the polynomial quotient ring, where p1 corresponds to the public key pk. i (pk i =(p0, p1)=(-p sk i +e0,p), i=1,2,…,n,e 2,i e 3,i These are elements sampled from the Gaussian polynomial quotient ring that satisfy c i,0 +c i,1 s=Δ m +e(mod q L ); (2-2) Authentication code generation: Each sender has its own data c i Calculate data labels Further calculate homomorphic random numbers Construct message authentication code ; Simultaneously construct a first-order polynomial of the encrypted authentication code while encrypting data. , Encrypt the result With authentication code Send them together to the calculation party.

[0009] Furthermore, step S203 includes: (3-1) Data Fusion: The computer collects the ciphertext from each sender and directly performs ciphertext computation, including simultaneous addition / multiplication homomorphic computation on the ciphertext data and the verification code: Let the ciphertext data include c1=(c 1,0 c 1,1 c2=(c 2,0 c 2,1 ),…,c n =(c n,0 c n,1 (corresponding to plaintexts m1 and m2, ..., m) n Add / multiply each pair of elements n-1 times to get the result c. f ; (3-1-1) Performing addition calculations yields c. f =c add (c add =c1+c2+…+c n =(c 1,0 +c 2,0 +…+c 2,n c 1,1 +c 2,1 +…+c 2,n ) mod q L Satisfying c0+c1 s=Δ (m1+m2+…+m n )+e(mod q L )); s is the receiver's key; (3-1-2) Performing multiplication yields c f =c mult Select the first two encrypted data c1 and c2, c mult =c1 c2=(c 1,0 c 1,1 ) (c 2,0 c 2,1 )=(d0, d1, d2)=(c 1,0 c 2,0 c 1,0 c 2,1 +c 1,1 c2,0 c 1,1 c 2,1 The result is then relinearized using the relinearization public key rlk from the initialization phase to obtain c. mult The calculation process is as follows: c mult '=(b mult ,a mult )=(d0+「P 1 d2 r0”, d1+“P 1 d2 r1」))=(d0+P 1 d2 (-a sk+e1+P sk 2 ), d1+P 1 d2 a) , b mult +a mult sk=Δ 2 m mult +e mult =Δ 2 m1m2+e mult Where sk is the receiver's key, satisfying sk = Σsk i ; (3-1-3) Using modular reduction techniques to reduce c mult 'noise e mult To improve plaintext accuracy, c is obtained by reducing the size while keeping the magnification factor Δ constant. f The calculation is as follows: c f= RS(c mult ')=c''=(b3,a3)=「 1 c mult '」=( 1 b mult , 1 a mult (mod q) i-1 It satisfies b³ + a³ sk= m mult +e(modq i-1 ); After completing the fusion of two dense-state data, continue with steps (3-1-2) to (3-1-3) to fuse the fusion result with the next dense-state data. A total of n-1 dense-state data fusions are completed for n data. (3-2) Authentication code integration: Perform addition / multiplication operations on the encrypted authentication code: It is performed along with the dense state calculation; When addition is performed, it is equivalent to adding the coefficients of a polynomial: ; When multiplication is performed, it is equivalent to performing a convolution on the corresponding coefficients: .

[0010] Furthermore, step S204 includes: The computational method will use the dense-state fusion computation results (c) f (A, B) are returned to the recipient; (4-1) Verification code: The receiver receives data c f and its verification code First, determine If the condition is met, then further calculate the data label. random numbers ,verify Check if the condition is met; if it is, authentication is successful; otherwise, it fails. (4-2) Recipient decryption: The receiver for c f = (b f , a f Decryption: b f +a f sk=Δ m f + e (mod q) i ) to obtain the encoded plaintext Δ m f Then decode to obtain plaintext Dcd(m) f ; Δ)=z; sk is the receiver's key, satisfying sk=Σsk i .

[0011] Furthermore, in each stage, the verification process and the encrypted data process are executed in parallel.

[0012] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: This invention allows organizations or individuals to control their data privacy, supports distributed collaborative computation directly on encrypted text, prevents malicious behavior by participants, and enables verifiable encrypted text fusion computation among a group of participants, ensuring data security and the authenticity and trustworthiness of information, while reducing the cost of cross-domain and cross-departmental collaboration. Its computational features mainly include distributed collaborative computation, input privacy, and robustness. 1. Distributed collaborative computing: The decentralized computing method gives secure multi-party computing the advantage of being resistant to interference.

[0013] 2. Input privacy: The information exposed during the execution of the protocol, apart from the part that can be deduced from the calculation results, cannot lead to the private input of any participant.

[0014] 3. Robustness (computational correctness): In data fusion computation, it can verify man-in-the-middle attacks and prevent malicious actions by computation participants from causing other honest participants to output incorrect results.

[0015] In summary, this invention proposes a method for fusion of encrypted data applicable to both cloud-based and peer-to-peer scenarios, designs a verification method for encrypted data fusion, and achieves "usable but invisible" encrypted data, thereby improving the availability of encrypted data. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of a verifiable dense-state data fusion model in an embodiment of the present invention.

[0018] Figure 2 This is a flowchart of a verifiable dense-state data fusion method in an embodiment of the present invention.

[0019] Figure 3 This is a flowchart of the data encryption and re-linearization key initialization process in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0021] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0022] Example This embodiment proposes a verifiable encrypted data fusion method, including a key initialization stage, a data encryption and authentication code generation stage, an encrypted data fusion calculation stage, and a fused encrypted data decryption stage. Specifically, as follows (in each stage, the verification process and the encrypted data process are executed in parallel): S100, Construct a dense-state data fusion model: like Figure 1 As shown, the encrypted data fusion model supports point-to-point decentralized fusion computing and cloud-based fusion computing. Point-to-point decentralized fusion computing is executed in real time by the computing party and does not store encrypted data from any party. Cloud-based fusion computing supports encrypted storage from all parties and initiates asynchronous fusion processing by the computing party. Corresponding to the encrypted data fusion process, the encrypted verification is also based on the homomorphic computation concept. It uses a one-way function to extract data tags and randomly processes these tags, constructing a polynomial to support subsequent homomorphic authentication code calculation. The specific process is as follows: (1) Each sender generates its own private key based on the key generation algorithm, generates a data fusion encryption key based on the secret sharing method, and generates an authentication key independently.

[0023] (2) Each sender encrypts its own data based on the homomorphic encryption algorithm to generate data ciphertext, and generates a homomorphic MAC authentication code for the encrypted data.

[0024] (3) The computing party performs fusion calculation on the encrypted data and performs corresponding homomorphic MAC authentication code calculation.

[0025] (4) The computing party converts the encrypted fusion calculation result into ciphertext that the receiver can decrypt, and sends it to the receiver together with the homomorphic MAC authentication code.

[0026] (5) After the receiver verifies the MAC, it decrypts the plaintext result.

[0027] S200, based on the dense-state data fusion model, performs dense-state data fusion, such as... Figure 2 As shown: S201, Key Initialization Phase: (1-1) Data encryption key initialization (1-1-1) Each participant generates an initial encryption key. n senders each hold a unique private key sk, composed of elements uniformly sampled from a polynomial quotient ring. i i = 1, 2, ..., n; corresponding public key pk i (pk i =(p0, p1)=(-p sk i +e0, p), i=1,2,…,n, where p and a are common parameters uniformly sampled from the polynomial quotient ring, and e0 is an element sampled Gaussian from the polynomial quotient ring.

[0028] The receiver holds the private key sk, which was generated in the same way as each sender. r PK with public key r .

[0029] (1-1-2) Each participating party generates a secret shared public key pk based on the secret sharing method. s (pk s =(Σpk i ,p)).

[0030] Each participant uses its own private key sk i (i=1,2,…,n), generate the corresponding computational public key H. 0,i H 1,i (i=1,2,…,n), H 0,i =-a r i +sk i P+e 0,i H 1,i =a sk i +e 1,i Where P is the modulus on the polynomial quotient ring, and r i 'a' is a common parameter uniformly sampled from the polynomial quotient ring, and 'e' is a common parameter. 0,i e 1,i These are elements sampled from the Gaussian ring of the polynomial quotient; the public key H will be calculated. 0,i H 1,i Sending the keys to the key management (task management) aggregation point yields public keys H0 and H1 (H0 = ΣH). 0,i H1=ΣH 1,i (i=1,2,…,n).

[0031] (1-1-3) Each participating party receives the public keys H0 and H1 collected by the computation party. They then use the private key sk. i And calculate H using public keys H0 and H1 0,i '、H 1,i ' (H 0,i '=sk i H0+e 2,i H 1,i '=(r i -sk i H1+e 3,i , where e 2,i e 3,i (These are elements sampled from the Gaussian ring of the polynomial quotient), and the result H is calculated again. 0,i '、H 1,i 'Send to key management to collect H0' and H1' (H0'=ΣH 0,i ',H1'=ΣH 1,i The public key rlk (rlk=(H0'+H1',H1)) is further calculated and used to reduce the size of the ciphertext after the secret state calculation.

[0032] (1-1-4) Each sender collaboratively generates a relinearization public key rlk (RelinearizationKey). First, each participant uses its own private key sk. i Generate the corresponding computational public key H 0,i H 1,i , (H 0,i =-a r i +sk i P+e 0,i H 1,i =a sk i +e 1,i Where P is the modulus on the polynomial quotient ring, r i 'a' is a common parameter uniformly sampled from the polynomial quotient ring, and 'e' is a common parameter. 0,i e 1,i These are elements sampled from the Gaussian region on the polynomial quotient ring. The key management center further aggregates and calculates H0 and H1 (H0 = ΣH). 0,i H1=ΣH 1,i Then each participant uses their private key sk i And H is calculated from H0 and H1 0,i '、H 1,i '(H 0,i '=sk i H0+e 2,i H 1,I '=(r i -sk i H1+e 3,i The key management center then calculates H0' and H1' again (H0'=ΣH). 0,i ',H1'=ΣH 1,iFinally, all parties compute and linearize the public key rlk(rlk=(r0, r1)=(H0'+H1', H1)=(-a sk+sk sk P+e1,a), where P is the modulus on the polynomial quotient ring. Expanding the modulus space by P reduces the noise. p and a are common parameters uniformly sampled from the polynomial quotient ring, and e1 is an element sampled Gaussian from the polynomial quotient ring, used to reduce the size of the ciphertext after dense-state calculation.

[0033] (1-2) Data verification key initialization Each participant obtains the same authentication key k from the key management center based on the same pseudo-random function F.

[0034] S202, Data Encryption and Authentication Code Generation Stage: (2-1) Data Encryption (2-1-1) Each sender encodes the plaintext into an element Δ on the polynomial quotient ring. m ( Ecd (z,Δ)→Δ m ), where 1 / Δ represents the precision that the plaintext needs to retain, so multiplying by Δ in plaintext encoding controls the precision.

[0035] (2-1-2) Each sender uses the shared public key pk from the initialization phase. s The plaintext encoding result Δ m Encrypt to ciphertext c(c= (c i,0 c i,1 )= (Δ m +v i pk s +e 2,i ,v i p1+e 3,i ) mod q L , where v i It is an element (private key) uniformly sampled on the polynomial quotient ring, p1 corresponding to the public key pk. i (pk i =(p0, p1)=(-p sk i +e0, p), i=1,2,…,n,e 2,i e 3,i These are elements sampled from the Gaussian polynomial quotient ring that satisfy c i,0 +c i,1 s=Δ m +e(mod q L ).

[0036] (2-2) Authentication Code Generation Each sender has its own data c i Calculate data labels Further calculate homomorphic random numbers Construct message authentication code .

[0037] Simultaneously construct a first-order polynomial of the encrypted authentication code while encrypting data. , Encrypt the result With authentication code Send them together to the calculation party.

[0038] S203, Computation of dense-state data fusion stage: In cloud-based asynchronous scenarios, the computing party is the cloud center; in peer-to-peer real-time scenarios, it can act as an intermediary proxy to perform fusion computation. It uses the computation public key rlk, the ciphertext c, and the action f to complete homomorphic computation and obtain the ciphertext computation result c. f= c add or c mult (Result of addition / multiplication), such as Figure 3 As shown.

[0039] (3-1) Data Fusion The computer collects the ciphertext from each sender and directly performs ciphertext computation, including simultaneous addition / multiplication homomorphic computation on the ciphertext data and the verification code: Let the ciphertext data include c1=(c 1,0 c 1,1 c2=(c 2,0 c 2,1 ),…,c n =(c n,0 c n,1 (corresponding to plaintexts m1 and m2, ..., m) n Add / multiply each pair of elements n-1 times to get the result c. f .

[0040] (3-1-1) Performing addition calculations yields c. f =c add (c add =c1+c2+…+c n =(c 1,0 +c 2,0 +…+c 2,n c 1,1 +c 2,1 +…+c 2,n ) mod qL Satisfying c0+c1 s=Δ (m1+m2+…+m n )+e(mod q L )). s is the receiver's key.

[0041] (3-1-2) Performing multiplication yields c f =c mult Select the first two encrypted data c1 and c2, c mult =c1 c2=(c 1,0 c 1,1 ) (c 2,0 c 2,1 )=(d0, d1, d2)=(c 1,0 c 2,0 c 1,0 c 2,1 +c 1,1 c 2,0 c 1,1 c 2,1 Since ciphertext multiplication results in ciphertext size expansion, the result needs to be re-linearized using the re-linearization public key rlk from the initialization phase to obtain c. mult The calculation process is as follows: c mult '=(b mult ,a mult )=(d0+「P 1 d2 r0”, d1+“P 1 d2 r1」))=(d0+P 1 d2 (-a sk+e1+P sk 2 ), d1+P 1 d2 a) , b mult +a mult sk=Δ 2 m mult +e mult =Δ 2 m1m2+e mult Where sk is the receiver's key, satisfying sk = Σsk i .

[0042] (3-1-3) Using modular reduction (Rescaling) to reduce c mult 'noise e mult To improve plaintext accuracy, c is obtained by reducing the size while keeping the magnification factor Δ constant. f The calculation is as follows: c f= RS(c mult ')=c''=(b3,a3)=「 1 c mult '」=( 1 b mult , 1 a mult (mod q) i-1 ), (after homomorphic multiplication, q i Become q i-1 And satisfy q i / q i-1 = , This indicates the precision required to retain the number of decimal places. It satisfies b³ + a³ sk= m mult +e(modq i-1 ).

[0043] Having completed the fusion of two dense-state data, continue with steps (3-1-2) to (3-1-3) to fuse the fusion result with the next dense-state data. A total of n-1 dense-state data fusions are completed for n data.

[0044] (3-2) Authentication Code Integration Perform addition / multiplication operations on the encrypted authentication code: It is executed along with the dense state calculation.

[0045] When addition is performed, it is equivalent to adding the coefficients of a polynomial: .

[0046] When multiplication is performed, it is equivalent to performing a convolution on the corresponding coefficients: .

[0047] S204, Decryption stage of fused encrypted data: The computational method will use the dense-state fusion computation results (c)f (A, B) are returned to the recipient.

[0048] (4-1) Authentication code verification The receiver receives data c f and its verification code First, determine If the condition is met, then further calculate the data label. random numbers ,verify Check if the condition is met. If it is met, the authentication will pass; otherwise, it will fail.

[0049] (4-2) Recipient decryption The receiver for c f = (b f , a f Decryption: b f +a f sk=Δ m f + e (mod q) i ) to obtain the encoded plaintext Δ m f Then decode to obtain plaintext Dcd(m) f ; Δ)=z. sk is the receiver's key, satisfying sk=Σsk i .

[0050] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A verifiable dense-state data fusion method, characterized in that, include: S100, construct a dense-state data fusion model; S200 performs dense data fusion based on the dense data fusion model: S201, Key initialization phase; S202, Data encryption and authentication code generation stage; S203, computational dense-state data fusion stage; S204, the stage of decrypting fused encrypted data; Step S201 includes: (1-1) Data encryption key initialization: (1-1-1) Each participant generates an initial encryption key: n senders each hold a unique private key sk, composed of elements uniformly sampled from a polynomial quotient ring. i i = 1, 2, ..., n; corresponding public key pk i Among them, pk i =(p0, p1)=(-p*sk) i +e0, p), where p and a are common parameters uniformly sampled from the polynomial quotient ring, and e0 is an element sampled Gaussian from the polynomial quotient ring; The receiver holds the private key sk, which was generated in the same way as each sender. r PK with public key r ; (1-1-2) Each participating party generates a secret shared public key pk based on the secret sharing method. s Among them, pk s =(Σpk i ,p); Each participant uses its own private key sk i (i=1,2,…,n), generate the corresponding computational public key H. 0,i H 1,i (i=1,2,…,n), H 0,i =-a*r i +sk i *P+e 0,i H 1,i =a*sk i +e 1,i Where P is the modulus on the polynomial quotient ring, and r i 'a' is a common parameter uniformly sampled from the polynomial quotient ring, and 'e' is a common parameter. 0,i e 1,i These are elements sampled from the Gaussian ring of the polynomial quotient; the public key H will be calculated. 0,i H 1,i Sending the public keys H0 and H1 to the key management aggregation point, where H0 = ΣH 0,i H1=ΣH 1,i , i=1,2,…,n; (1-1-3) Each participating party receives the public keys H0 and H1 collected by the computing party; and uses the private key sk i And calculate H using public keys H0 and H1 0,i '、H 1,i ' , H 0,i '=sk i *H0+e 2,i H 1,i '=(r i -sk i H1+e 3,i , where e 2,i e 3,i These are elements sampled from the Gaussian ring of the polynomial quotient; the result H is then calculated again. 0,i '、H 1,i 'Send to key management to collect H0' and H1', ​​H0'=ΣH 0,i ',H1'=ΣH 1,i '; and further calculate the re-linearized public key rlk, where rlk=(H0'+H1',H1), which is used to reduce the size of the ciphertext after the secret state calculation; (1-1-4) Each sender collaboratively generates a re-linearized public key rlk; first, each participant uses its own private key sk i Generate the corresponding computational public key H 0,i H 1,i H 0,i =-a*r i +sk i *P+e 0,i H 1,i =a*sk i +e 1,i Where P is the modulus on the polynomial quotient ring, r i 'a' is a common parameter uniformly sampled from the polynomial quotient ring, and 'e' is a common parameter. 0,i e 1,i These are elements sampled from a Gaussian ring of polynomials; the key management center aggregates and calculates H0 and H1, and then each participant uses their private key sk. i And H is calculated from H0 and H1 0,i '、H 1,i The key management center recalculates H0' and H1', ​​and finally all parties calculate and linearize the public key rlk, where rlk=(r0, r1)=(H0'+H1', H1)=(-a*sk+sk*sk*P+e1, a), where P is the modulus on the polynomial quotient ring; p and a are common parameters uniformly sampled from the polynomial quotient ring, and e1 is an element sampled from the polynomial quotient ring by Gaussian sampling, used to reduce the size of the ciphertext after the secret state calculation. (1-2) Data verification key initialization: Each participant obtains the same authentication key k from the key management center based on the same pseudo-random function F; Step S202 includes: (2-1) Data encryption: (2-1-1) Each sender encodes the plaintext into an element Δ* on the polynomial quotient ring. m ,in, Ecd (z,Δ)→Δ* m 1 / Δ represents the precision that the plaintext needs to retain, so multiplying by Δ in plaintext encoding controls the precision; (2-1-2) Each sender uses the shared public key pk from the initialization phase. s The plaintext encoding result Δ* m Encrypt to ciphertext c(c=(c i,0 c i,1 )= (Δ* m +v i *pk s +e 2,i ,v i *p1+e 3,i ) mod q L , where v i It is an element uniformly sampled on the polynomial quotient ring, where p1 corresponds to the public key pk. i Among them, pk i =(p0, p1)=(-p*sk) i +e0, p), i=1,2,…,n,e 2,i e 3,i These are elements sampled from the Gaussian polynomial quotient ring that satisfy c i,0 +c i,1 *s=Δ* m +e(mod q L ); (2-2) Authentication code generation: Each sender has its own data c i Calculate data labels Further calculate homomorphic random numbers Construct message authentication code ; Simultaneously construct a first-order polynomial of the encrypted authentication code while encrypting data. , Encrypt the result With authentication code Send them together to the calculation party; Step S203 includes: (3-1) Data Fusion: The computer collects the ciphertext from each sender and directly performs ciphertext computation, including simultaneous addition / multiplication homomorphic computation on the ciphertext data and the verification code: Let the ciphertext data include c1=(c 1,0 c 1,1 c2=(c 2,0 c 2,1 ),…,c n =(c n,0 c n,1 (corresponding to plaintexts m1 and m2, ..., m) n Add / multiply each pair of elements n-1 times to get the result c. f ; (3-1-1) Performing addition calculations yields c. f =c add , where c add =c1+c2+…+c n =(c 1,0 +c 2,0 +…+c 2,n c 1,1 +c 2,1 +…+c 2,n ) mod q L The condition c0 + c1*s = Δ*(m1 + m2 + … + m) is satisfied. n )+e(mod q L ); s is the receiver's key; (3-1-2) Performing multiplication yields c. f =c mult Select the first two encrypted data c1 and c2, c mult =c1*c2=(c 1,0 c 1,1 )*(c 2,0 c 2,1 )=(d0, d1, d2)=(c 1,0 c 2,0 c 1,0 c 2,1 +c 1,1 c 2,0 c 1,1 c 2,1 The result is then relinearized using the relinearization public key rlk from the initialization phase to obtain c. mult The calculation process is as follows: c mult '=(b mult ,a mult )=(d0+「P 1 *d2*r0”, d1+“P 1 *d2*r1」))=(d0+P 1 *d2*(-a*sk+e1+P*sk 2 ), d1+P 1 *d2*a), we have b mult +a mult *sk=Δ 2 *m mult +e mult =Δ 2 *m1m2+e mult Where sk is the receiver's key, satisfying sk = Σsk i ; (3-1-3) Using modular reduction techniques to reduce c mult 'noise e mult To improve plaintext accuracy, c is obtained by reducing the size while keeping the magnification factor Δ constant. f The calculation is as follows: c f= RS(c mult ')=c''=(b3,a3)=「D -1 *c mult '」=(D -1 *b mult ,D -1 *a mult )(mod q i-1 ); mult +e(modq i-1 ); After completing the fusion of two dense-state data, continue with steps (3-1-2) to (3-1-3) to fuse the fusion result with the next dense-state data. A total of n-1 dense-state data fusions are completed for n data. (3-2) Authentication code integration: Perform addition / multiplication operations on the encrypted authentication code: It is performed along with the dense state calculation; When addition is performed, it is equivalent to adding the coefficients of a polynomial: ; When multiplication is performed, it is equivalent to performing a convolution on the corresponding coefficients: 。 2. The verifiable dense-state data fusion method according to claim 1, characterized in that, Step S204 includes: The computational method will use the dense-state fusion computation results (c) f (A, B) are returned to the recipient; (4-1) Verification code: The receiver receives data c f and its verification code First, determine If the condition is met, then further calculate the data label. random numbers ,verify Check if the condition is met; if it is, authentication is successful; otherwise, it fails. (4-2) Recipient decryption: The receiver for c f = (b f , a f Decryption: b f +a f *sk=Δ*m f + e (mod q) i ) to obtain the encoded plaintext Δ*m f Then decode to obtain plaintext Dcd(m) f ; Δ)=z; sk is the receiver's key, satisfying sk=Σsk i .

3. The verifiable dense-state data fusion method according to claim 1, characterized in that, The encrypted data fusion model supports both point-to-point decentralized fusion computing and cloud-based fusion computing. Point-to-point decentralized fusion computing is executed in real time by the computing party and does not store encrypted data from any party. Cloud-based fusion computing supports encrypted storage from all parties and initiates asynchronous fusion processing by the computing party.

4. The verifiable dense-state data fusion method according to any one of claims 1-3, characterized in that, In each stage, the verification process and the encrypted data process are executed in parallel.

Citation Information

Patent Citations

  • Secure distributed key generation for multiparty homomorphic encryption

    US20210399874A1