A leveled homomorphic encryption method based on conjugate search problem

By adopting a leveled homomorphic encryption method based on conjugate search problem in cloud computing, using new internal product method and key initialization, combined with homomorphic operation of cloud computing platform, the difficulty of decryption and low security in the existing technology are solved. The problem of this achieves safe and efficient data computing and privacy protection in cloud computing.

CN115484018BInactive Publication Date: 2024-11-05TIANJIN UNIVERSITY OF TECHNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210997345.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-08-19
Publication Date
2024-11-05
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing leveled homomorphic encryption method has the problem of user decryption difficulty and long decryption time in cloud computing, and the total homomorphic encryption method based on conjugated search problem is not secure. The attacker can obtain plaintext by finding the ciphertext features.

Method used

Adopt the leveled homomorphic encryption method based on conjugate search problem, by defining new internal product methods and key initialization, homomorphic multiplication and addition operations are performed using cloud computing platform, and the simple ciphertext dimension is operated through Dimension_reduce to ensure the attacker It is difficult to crack cipher text and improve security.

Benefits of technology

The method of safely calculating data on the cloud computing platform is realized, protecting user data privacy, reducing user decryption difficulty, and improving the security of the solution, so that the attacker must solve the one-element high-order equation in a finite domain to obtain plain text.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115484018B_ABST
    Figure CN115484018B_ABST
Patent Text Reader

Abstract

The present invention relates to a leveled homomorphic encryption method based on a conjugate search problem, comprising the following steps: step 1, defining a new inner product method, and in Setup(λ, l, k), initializing a key and various security parameters; step 2, in Encrypt(m, H, t), encrypting the plaintext to be encrypted; step 3, a calculation initiator sends the encrypted ciphertext to a cloud computing party for homomorphic multiplication and homomorphic addition operations; step 4, after the cloud computing platform completes all requirements of the calculation initiator, it uses the Dimension_reduce(C, ab, d, k) operation to perform a simplified operation on the ciphertext dimension, and returns the final result to the calculation initiator; in step 5, the calculation initiator performs a Decrypt(C, sk, H) decryption operation on the ciphertext vector C returned by the cloud computing party to obtain the final calculation result. The present invention can safely calculate data on a cloud computing platform without leaking data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of information security technology and relates to a homomorphic encryption method, in particular to a leveled homomorphic encryption method based on a conjugate search problem. Background Art

[0002] In the network society under the background of cloud computing and big data, more and more data need to be calculated to provide people with better services. The data involved in the calculation may be extremely private data (such as sensitive identity information of patients in the medical system, etc.), but by calculating the data, it plays an important role in analyzing and processing the health status of patients.

[0003] However, most of the current cryptographic systems can only encrypt information but cannot operate on the encrypted information.

[0004] At this time, we need to introduce a homomorphic encryption solution. Homomorphic encryption refers to the encryption of a data m that needs to be protected. 1 ,m 2 ...m n , run the encryption function Enc(m 1 ),Enc(m 2 )...Enc(m n ), the decryption function of the encryption function has the property that satisfies:

[0005] Dec(f(Enc(m 1 ),Enc(m 2 )...,Enc(m n ))=f(m 1 ,m 2 ,...,m n )

[0006] Among them, Enc refers to the encryption method, Dec refers to the decryption method, and f represents the homomorphic operation that needs to be performed.

[0007] Existing homomorphic encryption methods are divided into: fully homomorphic encryption, partially homomorphic encryption and leveled homomorphic encryption; among them, the leveled homomorphic encryption method allows a limited number of homomorphic calculations of multiplication (the limited number can be specified by the user), but addition can be performed any number of times.

[0008] However, the leveled homomorphic encryption method in the prior art has the defects of being difficult for users to decrypt and taking a long time to decrypt.

[0009] Moreover, although the fully homomorphic encryption method based on the conjugate search problem proposed in the prior art has the property of satisfying full homomorphism, it is not safe. The attacker only needs to find the eigenvalue corresponding to the ciphertext to obtain the encrypted plaintext.

[0010] After searching, no literature in the prior art that is identical or similar to the present invention was found. Summary of the invention

[0011] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a leveled homomorphic encryption method based on the conjugate search problem that can safely calculate data on a cloud computing platform without leaking data.

[0012] The present invention solves the practical problem by adopting the following technical solutions:

[0013] A leveled homomorphic encryption method based on a conjugate search problem comprises the following steps:

[0014] Step 1: Define a new inner product method and initialize the key and various security parameters in Setup(λ,l,k);

[0015] Step 2: In Encrypt(m,H,t), encrypt the plaintext to be encrypted;

[0016] Step 3: The computing initiator sends the encrypted ciphertext to the cloud computing party for homomorphic multiplication and homomorphic addition operations;

[0017] Step 4: After the cloud computing platform completes all the requirements of the computing initiator, it uses the Dimension_reduce(C, ab, d, k) operation to simplify the ciphertext dimension and returns the final result to the computing initiator.

[0018] In step 5, the calculation initiator performs a decryption operation Decrypt(C, sk, H) on the ciphertext vector C returned by the cloud computing party to obtain the final calculation result.

[0019] Moreover, a new inner product method is defined in step 1 as follows:

[0020] <a,b> , which means to find the sum of the products of the first i elements in vectors a and b, where i represents the shortest dimension of vectors a and b), which can be expressed mathematically as:

[0021]

[0022] Moreover, in the Setup(λ,l,k) of step 1, the specific steps of initializing the key and various security parameters include:

[0023] 101) First, a number field F used by a scheme is selected;

[0024] 102) Limit the maximum number of homomorphic multiplication operations λ;

[0025] 103) Choose a finitely generated linear group whose elements are square matrices of dimension l.

[0026] 104) Randomly generate a number t in the number field F as the key. If the real number field is selected as the space, the key t should be a real number with an absolute value less than 1;

[0027] 105) Randomly generate an element H in the finite generated linear group. H is a reversible square matrix of dimension l, which is also a key. The key vector is obtained: sk = (1, t, t 2 ,...,t λp ).

[0028] Moreover, the specific steps of step 2 include:

[0029] 201) The computing initiator decomposes the plaintext m to be encrypted into m 1 +m 2 =m, where m 1 ,m 2 From the same plaintext space as m;

[0030] 202) Construct the following l-dimensional encrypted square matrix M, which is an upper triangular square matrix with 2x2 square matrices as blocks, and its lower left corner is a square matrix of 0. Where a, b, ..., k... are any elements from the number field F. Note: M should be a reversible square matrix;

[0031] 203) According to the above The method of generating the square key matrix H generates two reversible matrices K 1 , K 2 ;

[0032] 204) Calculate C based on the key matrix H and t obtained above 0 =H -1 (M-tK 1 -t p K 2 )H,C 1 =H -1 K 1 H,C p =H -1 K 2 H; According to the assumption of the conjugate search problem, at this time, even if the attacker gets C 1 =H -1 K 1 H,C p =H-1 K 2 H, it is also difficult to decipher it into H and K alone 1 , K 2 .

[0033] 205) The output ciphertext vector C = (C 0 ,C 1 ,0,...,C p )

[0034] Among them, C is a p+1-dimensional vector, and each element of C in the ciphertext is an l-dimensional square matrix, where 0 represents a 0 square matrix;

[0035] 206) If the number field selected at this time is a real number field, you can also choose to perform the following operations to reduce the dimension of the ciphertext: Otherwise, go directly to step 3;

[0036] 207) At this time, the calculation initiator should also calculate the infinite norm a of the first row of the key matrix H, H -1 The infinite norm b of the first and second columns and a small number d that is greater than the absolute value of the key t (for example, if t = 0.45, then d = 0.5, or 0.6 or 0.7);

[0037] 208) The calculated parameters ab=a*b, d, the encrypted result ciphertext vector C of each data, the required precision k and the calculation method f are sent to the cloud computing platform.

[0038] Moreover, the specific steps of step 3 include:

[0039] (1) The computing initiator sends the encrypted ciphertext to the cloud computing party for the Mult(C,C') homomorphic multiplication operation, and outputs the new ciphertext obtained by the homomorphic multiplication of the two ciphertexts;

[0040] Where C represents a ciphertext vector, C' represents another ciphertext vector, and the new ciphertext vector C' is output, where C' is calculated as follows:

[0041]

[0042] (2) The computing initiator sends the encrypted ciphertext to the cloud computing party for the homomorphic addition operation Add(C,C'), and outputs the new ciphertext obtained by homomorphic addition of the two ciphertexts;

[0043] Where C represents a ciphertext, C' represents another ciphertext, and the output ciphertext vector C" = C + C'. At this time, we can assume that deg(C)>deg(C'), so the calculation method of C" is as follows:

[0044]

[0045] Where C” is a ciphertext vector with a dimension less than or equal to λp+1.

[0046] Moreover, the specific steps of step 4 include:

[0047] 401) Initialize a cumulative calculation parameter tmp=0

[0048] 402) The cloud computing party selects the infinite norm of the matrix from the last ciphertext matrix of the ciphertext C, for example: C = (C 0 ,C 1 ,...,C n )(n<=λp) The cloud computing party first calculates C n The infinite norm o of

[0049] 403) Calculate tmp = tmp + o*a*b*d n

[0050] 404) If tmp <k / 2

[0051] Then the ciphertext matrix Cn can be deleted and the process returns to step 502), otherwise the process goes to step 505).

[0052] 405) Finally, the obtained ciphertext C' with reduced dimension is returned to the computing initiator.

[0053] Moreover, the specific steps of step 5 include:

[0054] 501) The calculation initiator first calculates using the key matrix H: M = H<sk,C> H -1

[0055] 502) The calculation initiator then calculates m'=M [1,1] +M [1,2] , and finally the precision k of m' is taken to obtain the final calculation result m.

[0056] Advantages and beneficial effects of the present invention:

[0057] 1. The present invention proposes a leveled homomorphic encryption method based on the conjugate search problem, which avoids the defect in the prior art that the attacker only needs to find the eigenvalue corresponding to the ciphertext to obtain the encrypted plaintext by adding additional "noise". The attacker must solve a univariate high-order equation in a finite field to obtain the encrypted plaintext, thereby improving the security of the scheme.

[0058] 2. The present invention can be widely used in the current cloud computing environment, and can solve the problem that users can let the cloud computing platform help users complete the calculations they need without disclosing their own data privacy. After the calculation is completed, the cloud computing platform can neither obtain the data sent by the user nor the calculation results obtained by the user, thereby protecting the user's data privacy security to the greatest extent.

[0059] 3. The present invention allows the user to specify the number domain where the calculation is located. If the specified number domain is a real number domain, the user can also specify the required precision, thereby reducing the difficulty of decryption for the user through an approximate calculation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a processing flow chart of the present invention. DETAILED DESCRIPTION

[0061] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings:

[0062] A leveled homomorphic encryption method based on the conjugate search problem, such as Figure 1 As shown, the following steps are included:

[0063] Step 1: Define a new inner product method and initialize the key and various security parameters in Setup(λ,l,k);

[0064] The definition of a new inner product in step 1 is:

[0065] <a,b> , which means to find the sum of the products of the first i elements in vectors a and b (where i represents the shortest dimension of vectors a and b), which can be expressed mathematically as:

[0066]

[0067] In the step 1, in Setup(λ,l,k), the specific steps of initializing the key and various security parameters include:

[0068] 104) First, a number field F used by a scheme is selected;

[0069] 105) Limit the maximum number of homomorphic multiplication operations λ;

[0070] 106) Choose a finitely generated linear group whose elements are square matrices of dimension l.

[0071] 104) Randomly generate a number t in the number field F as the key. If the real number field is selected as the space, the key t should be a real number with an absolute value less than 1;

[0072] 105) Randomly generate an element H in the finite generated linear group. H is a reversible square matrix of dimension l, which is also a key. The key vector is obtained: sk = (1, t, t 2 ,...,t λp ).

[0073] Step 2: In Encrypt(m,H,t), encrypt the plaintext to be encrypted;

[0074] The specific steps of step 2 include:

[0075] 209) The computing initiator decomposes the plaintext m to be encrypted into m 1 +m 2 =m, where m 1 ,m 2 From the same plaintext space as m;

[0076] 210) Construct the following l-dimensional encrypted matrix M, which is an upper triangular matrix with 2x2 matrices as blocks, and its lower left corner is a 0 matrix. Where a, b, ..., k... are any elements from the number field F. Note: M should be a reversible matrix;

[0077]

[0078] 211) Generate two reversible matrices K according to the generation method of the key matrix H above 1 , K 2 ;

[0079] 212) Calculate C based on the key matrix H and t obtained above 0 =H -1 (M-tK 1 -t p K 2 )H,C 1 =H -1 K 1 H,C p =H -1 K 2 H; According to the assumption of the conjugate search problem, at this time, even if the attacker gets C 1 =H -1 K 1 H,C p =H -1 K 2 H, it is also difficult to decipher it into H and K alone 1 , K 2 .

[0080] 213) The output ciphertext vector C = (C 0 ,C 1 ,0,...,Cp )

[0081] Among them, C is a p+1-dimensional vector, and each element of C in the ciphertext is an l-dimensional square matrix, where 0 represents a 0 square matrix;

[0082] 214) If the number field selected at this time is the real number field, you can also choose to perform the following operations to reduce the dimension of the ciphertext: Otherwise, go directly to step 3;

[0083] 215) At this time, the calculation initiator should also calculate the infinite norm a of the first row of the key matrix H, H -1 The infinite norm b of the first and second columns and a small number d that is greater than the absolute value of the key t (for example, if t = 0.45, then d = 0.5, or 0.6 or 0.7);

[0084] 216) The calculated parameters ab=a*b, d, the encrypted result ciphertext vector C of each data, the required accuracy k and the calculation method f are sent to the cloud computing platform.

[0085] Step 3: The computing initiator sends the encrypted ciphertext to the cloud computing party for homomorphic multiplication and homomorphic addition operations.

[0086] The specific steps of step 3 include:

[0087] (1) The computing initiator sends the encrypted ciphertext to the cloud computing party for the Mult(C,C') homomorphic multiplication operation, and outputs the new ciphertext obtained by the homomorphic multiplication of the two ciphertexts;

[0088] Where C represents a ciphertext vector, C' represents another ciphertext vector, and the new ciphertext vector C' is output, where C' is calculated as follows:

[0089]

[0090] In this embodiment, after one multiplication homomorphic operation, the dimension of the ciphertext vector increases;

[0091] (2) The computing initiator sends the encrypted ciphertext to the cloud computing party for the homomorphic addition operation Add(C,C'), and outputs the new ciphertext obtained by homomorphic addition of the two ciphertexts;

[0092] Where C represents a ciphertext, C' represents another ciphertext, and the output ciphertext vector C'=C+C'. At this time, let deg(C)>deg(C'), so the calculation method of C' is as follows:

[0093]

[0094] Note: C" is a ciphertext vector with a dimension less than or equal to λp+1.

[0095] Step 4: After the cloud computing platform completes all the requirements of the computing initiator, it uses the Dimension_reduce(C, ab, d, k) operation to simplify the ciphertext dimension and returns the final result to the computing initiator.

[0096] In this embodiment, after the cloud computing platform completes all the requirements of the computing initiator, a final ciphertext vector C is obtained. If the selected number field is a real number field, the process goes to step 401), and the Dimension_reduce(C, ab, d, k) operation is performed. Otherwise, if it is a normal finite field calculation, the obtained ciphertext is directly sent to the computing initiator, and the process goes to step 405). The specific steps of Dimension_reduce(C, ab, d, k) are as follows:

[0097] The specific steps of step 4 include:

[0098] 401) Initialize a cumulative calculation parameter tmp=0

[0099] 402) The cloud computing party selects the infinite norm of the matrix from the last ciphertext matrix of the ciphertext C, for example: C = (C 0 ,C 1 ,...,C n )(n<=λp) The cloud computing party first calculates C n The infinite norm o of

[0100] 403) Calculate tmp = tmp + o*a*b*d n

[0101] 404) If tmp <k / 2

[0102] Then the ciphertext matrix Cn can be deleted and the process returns to step 502), otherwise the process goes to step 505).

[0103] 405) Finally, the obtained ciphertext C' with reduced dimension is returned to the computing initiator.

[0104] In step 5, the calculation initiator performs a decryption operation Decrypt(C, sk, H) on the ciphertext vector C returned by the cloud computing party to obtain the final calculation result.

[0105] The specific steps of step 5 include:

[0106] 501) The calculation initiator first calculates using the key matrix H: M = H<sk,C> H -1

[0107] 502) The calculation initiator then calculates m'=M [1,1] +M [1,2] , and finally the precision k of m' is taken to obtain the final calculation result m.

[0108] The present invention is further described below by means of specific examples:

[0109] In this example, assume that user "Zhang San" holds a bunch of data m 01 =81,m 02 =48, user "Zhang San" needs to calculate f=m through the cloud server 01 4 *m 02 4 Assume that server "Li Si" has enough computing capacity to provide services for Zhang San.

[0110] Assume that both parties have selected the (leveled) homomorphic encryption method based on the conjugate search problem of the present invention:

[0111] 1) Setup Zhang San sets the parameters of the homomorphic encryption scheme.

[0112] 101) Select security parameters l=4, λ=8, k=1, p=5, and assume that the number field selected at this time is the real number field;

[0113] 102) Select the key t = 0.008, and then get the key vector sk = (1, t, t 2 ,...,t 40 );

[0114] 103) Select the key matrix H

[0115]

[0116] 104) Calculate the inverse matrix H of H -1

[0117] 2) Encrypt Zhang San calculates the encrypted data according to the encryption function (here m 01 =81 as an example)

[0118] 201) Using the encryption method above, randomly generate an original encrypted matrix M:

[0119]

[0120] 202) Then generate two random matrices K by generating the key matrix H 1 , K 2 :

[0121]

[0122] 203) Calculate m 01 Encrypted ciphertext vector:

[0123] C 01 =(H -1 (M-tK 1 -t 5 K 2 )H,H -1 K 1 H, 0, ..., H -1 K 2 H);

[0124] 204) According to m 01 The encryption method for m 02 Encrypt to get C 02 ;

[0125] 205) Zhang San calculates the infinite norm of the first row of the key matrix H a=13, H -1 Infinity of the first and second columns

[0126] The norm b=0.135 and a small number d=0.01 which is greater than the absolute value of t.

[0127] 206) The ciphertext vector C after encrypting the calculated parameters a*b, d, the required accuracy k=1 and the calculation method f=m 01 4 *m 02 4 Send to John Doe.

[0128] 3) Li Si uses the ciphertext sent by Zhang San and the calculation method to calculate, using Add(C 01 ,C 02 ) Calculate the ciphertext C 01 With ciphertext C 02 Addition, using Mult(C 01 ,C 02 ) Calculate the ciphertext C 01 With C 02 According to f=m 01 4 *m 02 4 , Li Si needs to calculate:

[0129] 301)C 03 =Mult(C 01 ,C 02 )

[0130] 302)C 04 =Mult(C 03 ,C 03)

[0131] 303)C 05 =Mult(C 04 ,C 04 )

[0132] Further, based on the above data, we will get a ciphertext vector C with a dimension of 41 05 .

[0133] 4) Li Si uses Dimension_reduce (C 05 ,13*0.135,0.01,1) Reduce the ciphertext dimension through the data ab,d sent by Zhang San, and return the simplified ciphertext to Zhang San. The above data can reduce the ciphertext dimension to a new ciphertext with less than 41 dimensions.

[0134] 5) Assume that the ciphertext vector sent by Li Si to Zhang San is C. Zhang San uses Decrypt(C,sk,H)

[0135] 501) Calculate M' = H<C,sk> H -1 Get M',

[0136] 502) By calculating M' [1,1] +M' [1,2] And take the precision of k as 1 to get the fourth power of the final calculation result (81*48).

[0137] It should be emphasized that the embodiments of the present invention are illustrative rather than restrictive. Therefore, the present invention includes but is not limited to the embodiments described in the specific implementation modes. Any other implementation modes derived by those skilled in the art based on the technical solutions of the present invention also fall within the scope of protection of the present invention.

Claims

1. A leveled homomorphic encryption method based on the conjugate search problem, characterized in that: The following steps are involved: Step 1: Define a new inner product method and initialize the key and various security parameters in Setup(λ,l,k); Step 2: In Encrypt(m,H,t), encrypt the plaintext to be encrypted; Step 3: The computing initiator sends the encrypted ciphertext to the cloud computing party for homomorphic multiplication and homomorphic addition operations; Step 4: After the cloud computing platform completes all the requirements of the computing initiator, it uses the Dimension_reduce(C, ab, d, k) operation to simplify the ciphertext dimension and returns the final result to the computing initiator. In step 5, the calculation initiator calculates the ciphertext vector C returned by the cloud computing party, performs a decryption operation Decrypt(C,sk,H) on it, and obtains the final calculation result; The definition of a new inner product in step 1 is: <a,b> , which means to find the sum of the products of the first i elements in vectors a and b, where i represents the shortest dimension of vectors a and b, expressed mathematically as: In the step 1, in Setup(λ,l,k), the specific steps of initializing the key and various security parameters include: 101) First, select a number field to be used in the scheme. 102) Limit the maximum number of homomorphic multiplication operations λ; 103) Select a finitely generated linear group whose elements are square matrices of dimension l; 104) Randomly generate a number field The number t in is used as the key. If the real number field is chosen as the space, then the key t should be a real number with an absolute value less than 1. 105) Randomly generate an element H in the finite generated linear group, H is a l-dimensional reversible matrix, which is also a key, and obtain the key vector: sk = (1, t, t 2 ,...,t λp ); The specific steps of step 2 include: 201) The computing initiator decomposes the plaintext m to be encrypted into m1+m2=m, where m1 and m2 are from the same plaintext space as m; 202) Construct the following l-dimensional encrypted matrix M, which is an upper triangular matrix with 2x2 matrices as blocks, and its lower left corner is a 0 matrix; where a, b, ..., k... are any numbers from the number field elements, M is a reversible square matrix; 203) Generate two reversible matrices K1 and K2 according to the generation method of the key matrix H above; 204) According to the key matrix H and t obtained above, C0=H -1 (M-tK1-t p K2)H, C1=H -1 K1H,C p =H - 1 K2H; According to the conjugate search problem assumption, at this time, even if the attacker gets C1=H -1 K1H,C p =H -1 K2H is also difficult to decipher into H, K1, and K2 alone; 205) The output ciphertext vector C = (C0, C1, 0, ..., C p ) Among them, C is a p+1-dimensional vector, and each element of C in the ciphertext is an l-dimensional square matrix, where 0 represents a 0 square matrix; 206) If the number field selected at this time is a real number field, select the following operation to reduce the dimension of the ciphertext: otherwise go directly to step 3; 207) At this time, the calculation initiator should also calculate the infinite norm a of the first row of the key matrix H, H -1 The infinity norm b of the first and second columns and a small number d greater than the absolute value of the key t; 208) Send the calculated parameters ab=a*b, d, the encrypted result ciphertext vector C of each data, the required precision k and the calculation method f to the cloud computing platform; The specific steps of step 3 include: (1) The computing initiator sends the encrypted ciphertext to the cloud computing party for the Mult(C,C') homomorphic multiplication operation, and outputs the new ciphertext obtained by the homomorphic multiplication of the two ciphertexts; Where C represents a ciphertext vector, C' represents another ciphertext vector, and the new ciphertext vector C' is output, where C' is calculated as follows: (2) The computing initiator sends the encrypted ciphertext to the cloud computing party for the homomorphic addition operation Add(C,C'), and outputs the new ciphertext obtained by homomorphic addition of the two ciphertexts; Where C represents a ciphertext, C' represents another ciphertext, and the output ciphertext vector C" = C + C'. At this time, let deg(C) > deg(C'), so the calculation method of C" is as follows: Where C” is a ciphertext vector with a dimension less than or equal to λp+1; The specific steps of step 4 include: 401) Initialize a cumulative calculation parameter tmp=0 402) The cloud computing party selects the infinite norm of the matrix from the last ciphertext matrix of the ciphertext C forward, C=(C0,C1,...,C n )(n<=λp) The cloud computing party first calculates C n The infinite norm o of 403) Calculate tmp = tmp + o*a*b*d n 404) If tmp <k / 2 Then delete the ciphertext matrix Cn and return to step 502), otherwise go to step 505) 405) Finally, the obtained ciphertext C' with reduced dimension is returned to the computing initiator; The specific steps of step 5 include: 501) The calculation initiator first calculates using the key matrix H: M = H<sk,C> H -1 502) The calculation initiator then calculates m'=M [1,1] +M [1,2] , and finally the precision k of m' is taken to obtain the final calculation result m.