Secure computation device, secure computation method, and program
The secure calculation system addresses the computation bottleneck in fully homomorphic encryption by using Bernoulli sequences to determine sums and products without bootstrapping, reducing calculation times and key lengths.
Patent Information
- Application Number
- JP2021567179
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2019-12-24
- Filing Date
- 2020-12-08
- Publication Date
- 2024-11-07
- Estimated Expiration
- 2040-12-08
AI Technical Summary
The computation time for secure calculations using fully homomorphic encryption is limited by the bootstrapping process, which requires a significant amount of computation, especially for operations like obtaining sums and products, leading to increased calculation times.
A secure calculation system that uses a homomorphic encryption method with a Bernoulli sequence expressed in various codes to perform operations on encrypted data, eliminating the need for bootstrapping by determining the sum or product values based on random bit strings rather than performing traditional addition or multiplication, thereby reducing noise accumulation.
This approach suppresses the increase in calculation amount and time by avoiding the need for bootstrapping, resulting in reduced key lengths and computation times, enhancing the efficiency of secure calculations.
Smart Images

Figure 0007578992000001 
Figure 0007578992000002 
Figure 0007578992000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a secure computation device, a secure computation method, and a program. This application claims priority based on Japanese Patent Application No. 2019-232752, filed on December 24, 2019, the contents of which are incorporated herein by reference. [Background technology]
[0002] Cloud computing has become popular in recent years. Cloud computing has the advantage that data is stored on the Internet, so users can perform calculations without being tied to a fixed location as long as they are in an environment with Internet access. On the other hand, cloud computing has security risks, such as data leakage, because data is stored on the Internet.
[0003] Therefore, in order to reduce security risks, secret computation may be used. Secret computation is a technology that allows encrypted data to be computed without decrypting it. One of the encryption methods for secret computation is a method that uses encryption called homomorphic encryption. Homomorphic encryption is roughly classified into three encryption methods: single-operation homomorphic encryption, restricted homomorphic encryption, and fully homomorphic encryption (Patent Document 1). Single-operation homomorphic encryption is an encryption method that can perform only one of an addition process to obtain a sum and a multiplication process to obtain a product while data remains encrypted. Restricted-operation homomorphic encryption is an encryption method that can perform addition and multiplication processes while data remains encrypted, but the number of operations is limited. Fully homomorphic encryption is an encryption method that can perform addition and multiplication processes while data remains encrypted, and has no limit on the number of operations. If addition and multiplication processes can be performed without limit on the number of operations, any calculation can be performed, so it is preferable to perform secret computation using fully homomorphic encryption rather than single-operation homomorphic encryption and restricted-operation homomorphic encryption.
[0004] Therefore, bootstrap is conventionally used as a technique for performing secret calculations using fully homomorphic encryption. Bootstrap is a technique for removing the limit on the number of calculations in restricted arithmetic homomorphic encryption. More specifically, it is a technique for reducing, to a certain level, the noise that increases each time an addition is performed on information encrypted by restricted arithmetic homomorphic encryption. Noise is a random element inserted during encryption. The random element also increases due to the addition process. In restricted arithmetic homomorphic encryption, the number of calculations is limited because decryption becomes impossible when the noise exceeds a certain threshold. Bootstrap is a technique for reducing the noise that has increased to just before this threshold to a certain level. [Prior art documents] [Non-patent literature]
[0005] [Non-Patent Document 1] Craig Gentry “A FULLY HOMOMORPHIC ENCRYPTION SCHEME”, A DISSERTATION SUBMITTED TO THE DEPARTMENT OF COMPUTER SCIENCE AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY, 2009 Summary of the Invention [Problem to be solved by the invention]
[0006] However, bootstrap requires a large amount of calculation, and the calculation time required for secure computation using fully homomorphic encryption may be limited by bootstrap. For example, around 2009, the time required for bootstrap may be about 30 minutes while the time required for encryption, multiplication, and decryption is about milliseconds. In recent years, the time required for bootstrap has been shortened compared to the time in 2009, but bootstrap is still a factor that limits the computation time. Thus, in the past, secure computation including a process of obtaining a sum and a process of obtaining a product had a problem that the amount of calculation is significantly larger than that performed on decrypted data.
[0007] In view of the above circumstances, an object of the present invention is to provide a technique for suppressing an increase in the amount of calculation that occurs in secure computation including a process for obtaining a sum and a process for obtaining a product. [Means for solving the problem]
[0008] One aspect of the present invention is a secret computation device that includes a secret computation unit that performs secret computation on encrypted data in which plaintext expressed in a predetermined representation format for probability calculation is encrypted using a homomorphic encryption method, the secret computation including a process of obtaining a sum and a process of obtaining a product, and in obtaining the sum, which is the sum of first encrypted data, which is the first encrypted data, and second encrypted data, which is the second encrypted data, the secret computation unit determines the value of each digit of a bit string representing the sum to be either the value of the corresponding digit of a bit string representing the first encrypted data and represented in the representation format, or the value of the corresponding digit of a bit string representing the second encrypted data and represented in the representation format.
[0009] One aspect of the present invention is the above-mentioned secret calculation device, which includes a random bit string generation unit that generates a random bit string by using a bit string whose digit value is one of two values determined based on a predetermined probability distribution, and the secret calculation unit uses the random bit string to obtain a sum, and when obtaining a sum that is the sum of the first encrypted data and the second encrypted data, determines the value of each digit of the bit string representing the sum to be either the value of the corresponding digit of a bit string representing the first encrypted data and expressed in the expression format, or the value of the corresponding digit of a bit string representing the second encrypted data and expressed in the expression format, based on the value of the corresponding digit of the random bit string.
[0010] One aspect of the present invention is the above-mentioned secure computing device, wherein the bit string expressed in the predetermined representation format is a Bernoulli string represented by a unipolar code.
[0011] One aspect of the present invention is the above-mentioned secure computing device, wherein the bit string expressed in the predetermined representation format is a Bernoulli string expressed in a bipolar code.
[0012] One aspect of the present invention is the above-mentioned secure computing device, wherein the bit string expressed in the predetermined representation format is a Bernoulli string expressed as an inverted bipolar code.
[0013] One aspect of the present invention is the above-mentioned secret calculation device, wherein, in obtaining a product of first encrypted data, which is the first encrypted data, and second encrypted data, which is the second encrypted data, the secret calculation unit obtains the value of each digit of a bit string representing the product by a two-input logical operation in which the value of a corresponding digit of a bit string representing the first encrypted data and expressed in the expression format and the value of a corresponding digit of a bit string representing the second encrypted data and expressed in the expression format are input, the logical operation being in accordance with the expression format.
[0014] One aspect of the present invention is the above-mentioned secure computing device, wherein when the expression format is a unipolar code, the logical operation according to the expression format is a logical AND.
[0015] One aspect of the present invention is the above-mentioned secure computing device, wherein when the expression format is a bipolar code, the logical operation according to the expression format is a negation of an exclusive OR.
[0016] One aspect of the present invention is the above-mentioned secret computation device, wherein, in obtaining a sum which is the sum of the first encrypted data and the second encrypted data, the secret computation unit concatenates a bit string representing the first encrypted data and a bit string representing the second encrypted data, and obtains the result of the concatenation as the sum, and the first encrypted data and the second encrypted data are the result of multiple bits of the plaintext being encrypted into a single cipher.
[0017] One aspect of the present invention is a secret computation method including a secret computation step of performing a secret computation on encrypted data in which plaintext expressed in a predetermined representation format for probability calculation is encrypted using a homomorphic encryption method, the secret computation including a process of obtaining a sum and a process of obtaining a product, and the secret computation step, in obtaining a sum which is the sum of first encrypted data, which is the first encrypted data, and second encrypted data, which is the second encrypted data, determines the value of each digit of a bit string representing the sum to be either the value of the corresponding digit of a bit string representing the first encrypted data and represented in the representation format, or the value of the corresponding digit of a bit string representing the second encrypted data and represented in the representation format.
[0018] One aspect of the present invention is a program for causing a computer to function as the above-mentioned secure computing device. Effect of the Invention
[0019] According to the present invention, it is possible to suppress an increase in the amount of calculation that occurs in secure calculations that include a process for obtaining a sum and a process for obtaining a product. [Brief description of the drawings]
[0020] [Figure 1] FIG. 1 is a diagram showing an example of the functional configuration of a secure computation system 100 according to an embodiment. [Diagram 2] FIG. 2 is a diagram showing an example of a hardware configuration of a terminal device 1 according to an embodiment. [Diagram 3] 2 is a diagram showing an example of the functional configuration of a control unit 11 included in the terminal device 1 in the embodiment. [Figure 4] FIG. 2 is a diagram showing an example of a hardware configuration of a secure computing device 2 in the embodiment. [Diagram 5] FIG. 2 is a diagram showing an example of the functional configuration of a control unit 21 included in the secure computing device 2 in the embodiment. [Figure 6] FIG. 11 is an explanatory diagram illustrating a probabilistic secret product acquisition process according to an embodiment, taking as an example a case where the expression format is unipolar. [Figure 7] FIG. 11 is an explanatory diagram for explaining the probabilistic secret product acquisition process in the embodiment, taking as an example a case where the expression format is bipolar. [Figure 8] FIG. 11 is an explanatory diagram for explaining a confidential sum acquisition process in the embodiment. [Figure 9] 1 is a flowchart showing an example of the flow of processing executed in the secure computation system 100 of the embodiment. [Figure 10] FIG. 11 is a diagram showing an example of an experiment result of a performance evaluation of the secure computation system 100 according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0021] (Embodiment) FIG. 1 is a diagram illustrating an example of a functional configuration of a secure computation system 100 according to an embodiment.
[0022] The secure computation system 100 encrypts a plaintext expressed in a predetermined representation format for probability computation (hereinafter referred to as a "probability-representing plaintext") using the homomorphic encryption method. The secure computation system 100 executes secure computation on the probability-representing plaintext (hereinafter referred to as the "encrypted data") encrypted using the homomorphic encryption method.
[0023] The expression format for probability calculation is a data format that expresses target data using a bit string of a predetermined number of digits that satisfies a probability calculation expression condition (hereinafter referred to as a "probability expression bit string"). The probability calculation expression condition includes a condition that the bit string is a bit string indicating a value whose absolute value is 1 or less. The probability calculation expression condition includes a condition that each bit is a bit string having either a value of 0 or 1. The probability calculation expression condition includes a condition that the value indicated by the bit string is a bit string that indicates the frequency of occurrence of one of the predetermined values, 0 or 1, in the bit string.
[0024] The probability representation bit string is, for example, a Bernoulli string represented by a unipolar code, a Bernoulli string represented by a bipolar code, or a Bernoulli string represented by an inverted bipolar code.
[0025] A Bernoulli sequence represented by a unipolar code is a bit string in which each bit has a value of either 0 or 1, and the frequency of occurrence of one of the predetermined values, 0 or 1, in the bit string is indicated by a real number between 0 and 1. A Bernoulli sequence represented by a unipolar code indicates a larger value the higher the frequency of occurrence of one of the predetermined values, 0 or 1, in the bit string.
[0026] A Bernoulli sequence represented by a bipolar code is a bit string in which each bit has either the value 0 or 1, and is a bit string that indicates the frequency of occurrence of one of the predetermined values, 0 or 1, in the bit string as a real number between (-1) and 1. A Bernoulli sequence represented by a bipolar code indicates a larger value the higher the frequency of occurrence of one of the predetermined values, 0 or 1, in the bit string.
[0027] A Bernoulli sequence represented by an inverted bipolar code is a bit string in which each bit has either the value 0 or 1, and the frequency of occurrence of one of the predetermined values, 0 or 1, in the bit string is indicated by a real number between (-1) and 1. Unlike a Bernoulli sequence represented by a bipolar code, a Bernoulli sequence represented by an inverted bipolar code indicates a smaller value the higher the frequency of occurrence of one of the predetermined values, 0 or 1, in the bit string.
[0028] The homomorphic encryption method may be a homomorphic encryption method in which only addition is defined, a homomorphic encryption method in which only multiplication is defined, or a homomorphic encryption method in which addition and multiplication are defined. The homomorphic encryption method may be, for example, an ideal lattice-based method, an integer-fully homomorphic encryption method, a Ring-LWE-based method, an RSA encryption method, a Pailler encryption method, or an Okamoto-Uchiyama encryption method. The homomorphic encryption method may be, for example, a Goldwasser-Micali encryption method, or an ElGama encryption method. The homomorphic encryption method may be, for example, an elliptic curve ElGama encryption method (EC ElGama encryption method), a Lifted ElGama encryption method, or a Lifted EC ElGama encryption method.
[0029] The secure computation system 100 includes a terminal device 1 and a secure computation device 2. The terminal device 1 executes a representation conversion process on the plaintext. The representation conversion process is a process for converting the format of the plaintext data into a predetermined representation format for probability calculation. The representation conversion process converts the plaintext into a probability representation plaintext. The plaintext before the representation conversion process is executed is expressed in N-ary notation (N is an integer). The plaintext before the representation conversion process is executed is expressed, for example, in binary notation.
[0030] For ease of explanation, hereinafter, when a plaintext is converted into a Bernoulli sequence represented by a unipolar code by a representation conversion process, the representation format is said to be unipolar. Also, hereinafter, when a plaintext is converted into a Bernoulli sequence represented by a bipolar code by a representation conversion process, the representation format is said to be bipolar. Also, hereinafter, when a plaintext is converted into a Bernoulli sequence represented by an inverted bipolar code by a representation conversion process, the representation format is said to be inverted bipolar.
[0031] The terminal device 1 executes encryption processing. The encryption processing is a process of encrypting a probability-expressing plaintext using a predetermined key (hereinafter referred to as an "encryption key") for encrypting data by a homomorphic encryption method. The probability-expressing plaintext is converted into encrypted data by the encryption of the probability-expressing plaintext. The terminal device 1 transmits the encrypted data to the secure computing device 2 via the network 9.
[0032] The terminal device 1 obtains data, which is the calculation result of the secure computation device 2 and is encrypted by the homomorphic encryption method, via the network 9. The terminal device 1 executes a decryption process. The decryption process is a process of decrypting encrypted data indicating the result of the secure computation using a key for encoding (hereinafter referred to as a "decryption key") that is paired with the encryption key.
[0033] The secret computation device 2 receives the encrypted data transmitted by the terminal device 1 via the network 9. The secret computation device 2 performs a predetermined secret computation (hereinafter referred to as a "target secret computation") including a process of obtaining a sum and a process of obtaining a product on the encrypted data. The target secret computation is a computation executed while the probability expression plaintext remains encrypted, and may be any computation including a process of obtaining a sum and a process of obtaining a product. The target secret computation is, for example, machine learning using encrypted data as input data, and is machine learning or inference in machine learning in which a predetermined machine learning model is trained while the probability expression plaintext remains encrypted. The secret computation device 2 transmits the execution result of the target secret computation to the terminal device 1.
[0034] FIG. 2 is a diagram illustrating an example of a hardware configuration of the terminal device 1 according to the embodiment. The terminal device 1 includes a control unit 11 including a processor 91 such as a CPU (Central Processing Unit) and a memory 92 connected by a bus, and executes a program. The terminal device 1 functions as a device including the control unit 11, a communication unit 12, a storage unit 13, an input unit 14, and an output unit 15 by executing the program.
[0035] The control unit 11 controls the operation of each functional unit included in the terminal device 1. The control unit 11 executes, for example, an expression conversion process. The control unit 11 executes, for example, an encryption process. The control unit 11 executes, for example, a decryption process.
[0036] The communication unit 12 includes a communication interface for connecting the device itself to the network 9. The communication unit 12 communicates with the secret computation device 2 via the network 9. The communication unit 12 transmits and receives encrypted data to and from the secret computation device 2, for example, by communicating with the secret computation device 2. The communication unit 12 transmits encrypted data to the secret computation device 2, for example, by communicating with the secret computation device 2. The communication unit 12 receives encrypted data after the target secret computation is performed, for example, by communicating with the secret computation device 2. The communication unit 12 may acquire plaintext transmitted by an external device that transmits plaintext, by communicating with the external device that transmits plaintext, via the network 9.
[0037] The storage unit 13 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 13 stores, for example, a program that executes an expression conversion process. The storage unit 13 stores, for example, a program that executes an encryption process. The storage unit 13 stores, for example, a program that executes a decryption process. The storage unit 13 stores, for example, an encryption key. The storage unit 13 stores, for example, a decryption key. The storage unit 23 stores, for example, encrypted data.
[0038] The input unit 14 includes input devices such as a mouse, a keyboard, a touch panel, etc. The input unit 14 may be configured as an interface that connects these input devices to the input unit 14. The input unit 14 accepts input of, for example, plain text.
[0039] The output unit 15 includes a display device such as a CRT (Cathode Ray Tube) display, a liquid crystal display, or an organic EL (Electro-Luminescence) display. The output unit 15 may be configured as an interface that connects these display devices to the output unit 15. The output unit 15 displays information input to the input unit 14, such as plain text.
[0040] FIG. 3 is a diagram illustrating an example of a functional configuration of the control unit 11 included in the terminal device 1 in the embodiment. The control unit 11 includes a plaintext acquisition unit 110 , an expression conversion unit 120 , an encryption unit 130 , a decryption unit 140 , a communication control unit 150 , an output control unit 160 , and an expression inverse conversion unit 170 .
[0041] The plaintext acquisition unit 110 acquires plaintext via the input unit 14 or the communication unit 12. For ease of explanation, the secure computation system 100 will be described below taking as an example a case where plaintext is input to the input unit 14. The expression conversion unit 120 executes an expression conversion process on the plaintext. By executing the expression conversion process, the plaintext is converted into a probability expression plaintext.
[0042] The encryption unit 130 performs encryption processing on the probability-expression plaintext. By performing the encryption processing, the probability-expression plaintext is converted into encrypted data. The decryption unit 140 decrypts the encrypted data. The decrypted data is plaintext expressed in a predetermined representation format for probability calculation.
[0043] The communication control unit 150 controls the operation of the communication unit 12. The communication control unit 150 controls the operation of the communication unit 12 to transmit and receive encrypted data to and from a communication destination via the communication unit 12.
[0044] The output control unit 160 controls the operation of the output unit 15. The output control unit 160 controls the operation of the output unit 15 to display, for example, the contents input to the input unit .
[0045] The inverse representation converter 170 converts the data decoded by the decoder 140 into plain text expressed in an N-ary representation.
[0046] 4 is a diagram showing an example of a hardware configuration of the secret computing device 2 in the embodiment. The secret computing device 2 includes a control unit 21 including a processor 93 such as a CPU (Central Processing Unit) and a memory 94 connected by a bus, and executes a program. The secret computing device 2 functions as a device including the control unit 21, a communication unit 22, a storage unit 23, an input unit 24, and an output unit 25 by executing the program.
[0047] The control unit 21 controls the operation of each functional unit included in its own device (secure computation device 2). The control unit 21 executes, for example, a target secure computation.
[0048] The communication unit 22 includes a communication interface for connecting the device itself to the network 9. The communication unit 22 communicates with the terminal device 1 via the network 9. The communication unit 22 receives encrypted data transmitted by the terminal device 1 through communication with the terminal device 1, for example. The communication unit 22 also transmits encrypted data after the target confidentiality calculation has been performed to the terminal device 1 through communication with the terminal device 1, for example.
[0049] The storage unit 23 is configured using a storage device such as a magnetic hard disk device or a semiconductor storage device. The storage unit 23 stores, for example, a program of the target confidential calculation. The program of the target confidential calculation is a program encrypted in advance by a homomorphic encryption method. Therefore, values in the target confidential calculation that represent coefficients of a formula in the target confidential calculation, etc., are encrypted data. The storage unit 23 stores, for example, received encrypted data. The storage unit 23 stores, for example, encrypted data in the middle of the target confidential calculation. The storage unit 23 stores, for example, encrypted data after the target confidential calculation has been performed.
[0050] The input unit 24 includes input devices such as a mouse, a keyboard, a touch panel, etc. The input unit 24 may be configured as an interface that connects these input devices to the input unit 24. The input unit 24 receives, for example, an input of a program for the target confidential computation.
[0051] The output unit 25 includes a display device such as a CRT (Cathode Ray Tube) display, a liquid crystal display, or an organic EL (Electro-Luminescence) display. The output unit 25 may be configured as an interface that connects these display devices to the output unit 25. The output unit 25 displays information input to the input unit 24, for example.
[0052] FIG. 5 is a diagram showing an example of a functional configuration of the control unit 21 included in the secure computing device 2 in the embodiment. The control unit 21 includes an encrypted data acquisition unit 210 , a random bit string generation unit 220 , a secure calculation unit 230 , a communication control unit 240 , and an output control unit 250 . The encrypted data acquisition unit 210 acquires the encrypted data transmitted by the terminal device 1 via the communication unit 22 .
[0053] The random bit string generation unit 220 generates a random bit string R. The random bit string R is a bit string in which each digit has one of two values, such as 0 or 1, and the value of each digit is determined based on a predetermined probability distribution. For example, the value of a bit included in the random bit string R is 1 with a 50% probability and 0 with a 50% probability. In such a case, the predetermined probability distribution is a probability distribution in which the probability of 1 appearing is 50% and the probability of 0 appearing is 50%.
[0054] The probability of 0 and 1 appearing does not necessarily have to be the same. For example, 1 may occur with a 40% probability, and 0 with a 60% probability. In such a case, the predetermined probability distribution is a probability distribution in which the probability of 1 appearing is 40% and the probability of 0 appearing is 60%. For ease of explanation, the control unit 21 will be described below using an example in which each bit of the random bit string R indicates either one of the two values 0 or 1, and the bit value is 1 with a 50% probability and 0 with a 50% probability. The random bit string R generated by the random bit string generation unit 220 is input to the secret calculation unit 230.
[0055] The secret calculation unit 230 executes the target secret calculation using the random bit string R on the encrypted data. The secret calculation unit 230 executes a secret product acquisition process as a multiplication process in the target secret calculation. The secret product acquisition process is a process of acquiring a value indicating the product of values indicated by two encrypted data (hereinafter referred to as "product M") while keeping the encrypted data encrypted. The secret product acquisition process is a process of acquiring a value indicating the product of values indicated by two encrypted data, for example, on a probabilistic numerical expression. When the method of homomorphic encryption for generating the encrypted data is a method of homomorphic encryption in which multiplication or addition, or both, are defined in advance, the secret product acquisition process may acquire the product by a method of multiplication or addition defined in advance for the homomorphic encryption. Hereinafter, the secret product acquisition process (hereinafter referred to as "probabilistic secret product acquisition process") for acquiring the product M on a probabilistic numerical expression with two encrypted data as a first input value A and a second input value B will be described in more detail. The number of digits of the bit string representing the first input value A, the number of digits of the bit string representing the second input bit string B, and the number of digits of the bit string representing the product M are the same.
[0056] The probabilistic secret product acquisition process is a process in which the value of each digit of a bit string representing a product M is determined based on the value of the corresponding digit of a bit string representing a first input value A and the value of the corresponding digit of a bit string representing a second input value B. Specifically, the value of each digit of the bit string representing the product M, which is a value output in the probabilistic secret product acquisition process, is the result of a predetermined logical operation of two inputs, the value of the corresponding digit of the bit string representing the first input value A and the value of the corresponding digit of the bit string representing the second input value B being input. The probabilistic secret product acquisition process is a process according to a predetermined representation format for probability calculation. For example, when the representation format is a unipolar code, the logical operation of the two inputs is a logical AND. For example, when the representation format is a bipolar code, the logical operation of the two inputs is a negation of an exclusive OR.
[0057] FIG. 6 is an explanatory diagram for explaining the probabilistic secret product acquisition process in the embodiment, taking as an example a case where the expression format is unipolar. 6 shows that when a first input value A and a second input value B are input as two pieces of encrypted data, a product M is obtained using a process represented by an AND gate. Specifically, in the probabilistic private product obtaining process, when the representation format is unipolar, a product bit string M L The value of each digit of the first input bit string A L and the corresponding digit value of the second input bit string B L The product bit string M in Figure 6 is a logical product of the corresponding digits of L is a unipolar bit string that represents the product M. L is a unipolar bit string that represents the first input value A. L is a unipolar bit string that represents the second input value B. L , the second input bit string B L and the product bit string M L The number of digits is the same.
[0058] In Fig. 6, the first input value A is specifically (1 / 2). In Fig. 6, the first input bit string A, which is a bit string in the unipolar representation format representing the first input value A, LSpecifically, it is a four-digit bit string with the first digit being 0, the second digit being 0, the third digit being 1, and the fourth digit being 1.
[0059] In FIG. 6, the second input value B is specifically (1 / 2). In FIG. 6, the second input bit string B, which is a bit string in the unipolar representation format representing the second input value B, L Specifically, it is a four-digit bit string with the first digit being 0, the second digit being 1, the third digit being 1, and the fourth digit being 0.
[0060] In FIG. 6, the product M is specifically (1 / 4). In FIG. 6, the product bit string M L Specifically, it is a four-digit bit string with the first digit being 0, the second digit being 0, the third digit being 1, and the fourth digit being 0.
[0061] The process of obtaining the probabilistic secret product in Fig. 6 will be described in detail. In Fig. 6, for example, the product bit string M L The first digit value VM1 is the first input bit string A L The first digit value VA1 and the first input bit string B L In FIG. 6, for example, the product bit string M L The third digit VM3 is the first input bit string A L The third digit value VA3 and the first input bit string B L This is the logical AND of the third digit VB3.
[0062] FIG. 7 is an explanatory diagram for explaining the probabilistic secret product acquisition process in the embodiment, taking as an example a case where the representation format is bipolar. 7 shows that when a first input value A and a second input value B are input as two pieces of encrypted data, a product M is obtained using a process represented by an XNOR gate. Specifically, in the probabilistic private product obtaining process, when the representation format is bipolar, the product bit string M L The value of each digit of the first input bit string A L and the corresponding digit value of the second input bit string B L The product bit string M in Figure 7 is the negation of the exclusive OR with the corresponding digit value. Lis a bipolar bit string that represents the product M. L is a bipolar bit string that expresses the first input value A. L is a bipolar bit string that represents the second input value B. L , the second input bit string B L and the product bit string M L The number of digits is the same.
[0063] In Fig. 7, the first input value A is specifically 1. In Fig. 7, a first input bit string A, which is a bit string in a bipolar representation format representing the first input value A, L Specifically, it is a four-digit bit string with the first digit being 1, the second digit being 1, the third digit being 1, and the fourth digit being 1.
[0064] 7, the second input value B is specifically (-1 / 2). In FIG. 7, the second input bit string B, which is a bit string in the bipolar representation format representing the second input value B, L Specifically, it is a four-digit bit string with the first digit being 0, the second digit being 0, the third digit being 1, and the fourth digit being 0.
[0065] In FIG. 7, the product M is specifically (−1 / 2). In FIG. 7, the product bit string M L Specifically, it is a four-digit bit string with the first digit being 0, the second digit being 0, the third digit being 1, and the fourth digit being 0.
[0066] The process of obtaining the probabilistic secret product in Fig. 7 will be described in detail. In Fig. 7, for example, the product bit string M L The first digit value VM1 is the first input bit string A L The first digit value VA1 and the first input bit string B L In FIG. 7, for example, the product bit string M L The third digit VM3 is the first input bit string A L The third digit value VA3 and the first input bit string B L This is the negation of the exclusive OR with the third digit value VB3. This concludes the description of the process of obtaining a product through secure computation performed by the secure computation unit 230.
[0067] The secret calculation unit 230 executes the following secret sum acquisition process as an addition process in the target secret calculation, instead of an addition process in an N-ary notation. The secret sum acquisition process does not depend on the encryption scheme used.
[0068] The private sum acquisition process is a process of acquiring a value (hereinafter referred to as "sum S") indicating the sum of two encrypted data in a probabilistic numerical representation. In the private sum acquisition process, the two encrypted data and the sum S are expressed as a bit string of binary bits. A random bit string R is used in the private sum acquisition process. For ease of explanation, the private sum acquisition process will be explained below using an example in which the two encrypted data input in the private sum acquisition process are a first input value A and a second input value B.
[0069] In the secret sum acquisition process, the number of digits of the bit string representing the first input value A, the number of digits of the bit string representing the second input value B, and the number of digits of the bit string representing the sum S are the same. In the secret sum acquisition process, the number of digits of the random bit string R is equal to or greater than the number of digits of the bit string representing the first input value A, the number of digits of the bit string representing the second input value B, and the number of digits of the bit string representing the sum S.
[0070] In the secret sum acquisition process, the value of a bit in the random bit string R indicates either one of the bit strings representing the first input value A or the bit string representing the second input value B. More specifically, the value of each digit in the random bit string R in the secret sum acquisition process indicates the following: That is, it indicates either that the value of the corresponding digit in the bit string representing the first input value A is used to acquire the value of the corresponding digit in the sum S, or that the value of the corresponding digit in the bit string representing the second input value B is used to acquire the value of the corresponding digit in the sum S.
[0071] A bit of the random bit string with a value of, for example, 1 indicates that the corresponding value of the bit string representing the first input value A is used to obtain the corresponding digit value of the sum S. A bit of the random bit string with a value of, for example, 0 indicates that the corresponding value of the bit string representing the second input value B is used to obtain the corresponding digit value of the sum S.
[0072] For ease of explanation, the following describes the secret sum acquisition process using an example where the bit value of the random bit string R is 1 as the first input value A indication number and 0 as the second input value B indication number. 1 as the first input value A indication number means that the value indicating that the corresponding value of the bit string representing the first input value A is used to acquire the value of the corresponding digit of the sum S is 1. 0 as the second input value B indication number means that the value indicating that the corresponding value of the bit string representing the second input value B is used to acquire the value of the corresponding digit of the sum S is 0.
[0073] In the secret sum acquisition process, the value of each digit of the bit string representing the sum S is determined based on the value of the corresponding digit of the bit string representing the first input value A, the value of the corresponding digit of the bit string representing the second input value B, and the value of the corresponding digit of the random bit string R. Specifically, in the secret sum acquisition process, the value of each digit of the bit string representing the sum S is determined to be the value of the digit of the bit string indicated by the value of the corresponding digit of the random bit string R, out of the value of the corresponding digit of the bit string representing the first input value A and the value of the corresponding digit of the bit string representing the second input value B.
[0074] Therefore, the sum bit string S obtained as a result of the secret sum acquisition process is L The value of each digit of the sum bit string S is determined based on the value of the corresponding digit of the random bit string R, either the value of the corresponding digit of the bit string representing the first input value A or the value of the corresponding digit of the second input value B. L is a bit string representing the sum S.
[0075] FIG. 8 is an explanatory diagram illustrating the confidential sum acquisition process in the embodiment. First input bit string A in FIG. 8 Lis a bipolar bit string that expresses the first input value A. L is a bipolar bit string that represents the second input value B. L , the second input bit string B L , sum bit string S L and the number of digits of the random bit string R is the same.
[0076] In Fig. 8, the first input value A is specifically (1 / 2). In Fig. 8, the first input bit string A, which is a bit string in the bipolar representation format representing the first input value A, L Specifically, it is a four-digit bit string with the first digit being 1, the second digit being 0, the third digit being 1, and the fourth digit being 1.
[0077] In FIG. 8, the second input value B is specifically (−1 / 2). In FIG. 8, the second input bit string B, which is a bit string in the bipolar representation format representing the second input value B, L Specifically, it is a four-digit bit string with the first digit being 0, the second digit being 0, the third digit being 1, and the fourth digit being 0.
[0078] In FIG. 8, the sum S is specifically 0. In FIG. 8, the sum bit string S, which is a bit string in a bipolar representation format representing the sum S, L Specifically, it is a four-digit bit string with the first digit being 0, the second digit being 1, the third digit being 0, and the fourth digit being 0.
[0079] 8, the random bit string R is a four-digit bit string with the first digit being 1, the second digit being 0, the third digit being 1, and the fourth digit being 0. In this case, the value represented by the random bit string R is (1 / 2).
[0080] The secret sum acquisition process in Fig. 8 will be described in detail. In Fig. 8, for example, the sum bit string S L Since the value of the first digit VS1 of the random bit string A is 1, the value of the first digit VR1 of the random bit string A is 1. L In FIG. 8, for example, the sum bit string S LSince the value of the fourth digit VS4 of the random bit string VR4 is 0, the second input bit string B L The fourth digit is the value of VB4.
[0081] Returning to the description of Fig. 5, the communication control unit 240 controls the operation of the communication unit 22. The communication control unit 240 controls the operation of the communication unit 22 to transmit and receive encrypted data to and from a communication destination via the communication unit 22. The communication control unit 240 transmits encrypted data, which is a result of the secure calculation in the secure calculation unit 230, to the terminal device 1 via the communication unit 22, for example.
[0082] The output control unit 250 may control the operation of the output unit 25 to output encrypted data that is the result of the secure calculation in the secure calculation unit 230.
[0083] FIG. 9 is a flowchart showing an example of the flow of processing executed in the secure computation system 100 of the embodiment.
[0084] A plaintext is input to the terminal device 1 via the input unit 14 (step S101). Next, the plaintext acquisition unit 110 acquires the plaintext input in step S101 (step S102). Next, the expression conversion unit 120 executes an expression conversion process on the plaintext acquired by the plaintext acquisition unit 110 (step S103). By executing the expression conversion process, the plaintext is converted into a probability expression plaintext. After step S103, the encryption unit 130 encrypts the probability expression plaintext and converts it into encrypted data (step S104). Next, the communication control unit 150 transmits the encrypted data to the secret computing device 2 via the communication unit 12 (step S105).
[0085] Next, the encrypted data acquisition unit 210 acquires the encrypted data transmitted by the terminal device 1 via the communication unit 22 (step S106). Next, the secure calculation unit 230 executes the secure calculation (step S107). In step S107, a secure product acquisition process or a secure sum acquisition process is executed according to the target secure calculation. When executing the secure sum acquisition process in the secure calculation, first, the random bit string generation unit 220 generates a random bit string R.
[0086] Next, a secret sum acquisition process is executed using the generated random bit string R. After step S107, the communication control unit 240 transmits encrypted data indicating the calculation result of the secret calculation to the terminal device 1 via the communication unit 22 (step S108). Next, the decryption unit 140 acquires the encrypted data indicating the calculation result of the secret calculation via the communication unit 12 and decrypts it (step S109). Next, the expression inverse conversion unit 170 converts the decrypted data into plain text (step S110).
[0087] The thus configured secure computation system 100 obtains the sum of two positive or negative values using the random bit string R in secure computation for data encrypted by the homomorphic encryption method. More specifically, when obtaining the sum S, the value of each digit of the bit string representing the sum S is obtained by executing the following process. The secure computation system 100 determines, based on the random bit string R, either the value of the corresponding digit of the bit string representing the first input value A and expressed in a predetermined expression format in probability computation, or the value of the corresponding digit of the bit string representing the second input value B and expressed in a predetermined expression format in probability computation. The determined value is the value of the corresponding digit of the bit string representing the sum S. In this way, the secure computation system 100 obtains the sum S by selecting a value representing the sum, rather than performing addition.
[0088] Incidentally, in restricted homomorphic encryption and fully homomorphic encryption, random elements are added to data during encryption. When the process of obtaining the sum and the process of obtaining the product in homomorphic encryption are performed by N-ary addition, the added random value (hereinafter referred to as "noise") increases each time the process of obtaining the sum and the process of obtaining the product are performed. On the other hand, in the secure computation system 100, in the process of obtaining the sum or the product of two positive or negative values, N-ary addition or multiplication is not performed, but the sum value or the product value is obtained by selecting bits of a bit string representing a value. Therefore, in the secure computation system 100, even when the process of obtaining the sum is performed, noise does not increase. More precisely, noise does not occur in the first place in the secure computation system 100. Since noise does not increase in the secure computation system 100 in this way, the secure computation system 100 does not require a bootstrap process.
[0089] In addition, the secure computation system 100 can execute a process of obtaining a product and a process of obtaining a sum. The process of obtaining a product corresponds to multiplication. The process of obtaining a sum corresponds to addition. Therefore, the secure computation system 100 can execute secure computation including a process of obtaining a sum and a process of obtaining a product, and can suppress an increase in the amount of calculation occurring in the secure computation.
[0090] The increase in the amount of calculation is suppressed by the secure computation system 100 because the bootstrap process is unnecessary as described above. Therefore, the secure computation method executed in the secure computation system 100 is a method different from HEAAN (Homomoriphic Encryption for Arithmetic of Approximate Numbers). HEAAN is a method for approximating values.
[0091] (Experimental Results) As described above, the secure computation system 100 does not require the bootstrap process. Therefore, the secure computation system 100 can reduce the key length, which is an index representing the amount of computation in secure computation, to 10^(-5) to 10^(-6) times the key length in fully homomorphic encryption using bootstrap. This is shown in the experimental results described below.
[0092] <Experimental Results> The performance of the secure computation system 100 and the performance of other secure computation methods will be described using experimental results. The performance of the secure computation is, for example, a key length. In the experiment, the secure computation system 100 performed secure computation using an IBP (Inverted Bipolar) code under the encryption method of EC ElGamal encryption.
[0093] One of the secret calculations by the other secret calculation method is a secret calculation on data encrypted by the additive homomorphic encryption (Additive HE) method by the EC ElGamal encryption method. Another of the secret calculations by the other secret calculation method is a secret calculation on data encrypted by the restricted homomorphic encryption (SHE) method by the Ring-LWE-based encryption encryption method. Another of the secret calculations by the other secret calculation method is a secret calculation on data encrypted by the Torus-based Fully Homomorphic Encryption (TFHE) method. The performance related to the secret calculation is, for example, the key length.
[0094] SHE is an encryption method in which the upper limit of the number of times multiplication can be performed is predetermined according to a parameter. The parameter of SHE is, for example, the size of the plaintext space. The parameter of SHE may be the size of the ciphertext space or the degree of a polynomial.
[0095] Fig. 10 is a diagram showing an example of an experiment result of a performance evaluation regarding secure computation of the secure computation system 100 of the embodiment. Fig. 10 shows the Number of Mult. and the key length as the performance regarding secure computation. The Number of Mult. indicates the upper limit of the number of multiplications that can be performed in SHE using the used SHE parameters. The used SHE parameters are the parameters of SHE used in the experiment below. N / A means that multiplication cannot be performed.
[0096] Fig. 10 shows the performance of secure computation (hereinafter referred to as "AHE computation") for data encrypted by additive homomorphic encryption (Additive HE) based on the EC ElGamal encryption scheme. The results of the AHE computation shown in Fig. 10 are those described in Reference 1. According to Reference 1, as shown in Fig. 10, the Number of Mult. in the AHE computation is N / A, and the key length in the AHE computation is 256 bits.
[0097] Reference 1: Ilaria Chillotti, Nicolas Gama, Maria Georgieva, and Malika Izabachene, “TFHE: Fast Fully Homomorphic Encryption Over the Torus”, Journal of Cryptology, 25, April, 2019
[0098] FIG. 10 shows the performance of a secure computation (hereinafter referred to as "SHE computation") for data encrypted by the restricted homomorphic encryption (SHE) method using the Ring-LWE-based encryption encryption method. The result of the SHE computation shown in FIG. 10 is the result described in Reference 1. According to Reference 1, as shown in FIG. 10, the Number of Mult. in the SHE computation is 1 or more and less than 15. According to Reference 1, as shown in FIG. 10, the shortest key length in the SHE computation is 20 kilobits and the longest key length is 13 megabits. According to Reference 1, the result shown in FIG. 10 is the result of the SHE computation corresponding to the case of t=2 in Tab. 1. Corresponding to the case of t=2 in Tab. 1 means that it corresponds to the case where the plaintext is given as either 0 or 1.
[0099] FIG. 10 shows the performance of secure computation (hereinafter referred to as "TFHE computation") for data encrypted by the fully homomorphic encryption (TFHE) method. The results of the TFHE computation shown in FIG. 10 are those described in Reference 1. According to Reference 1, as shown in FIG. 10, the Number of Mult. in the TFHE computation is infinite, and the key length is 16 megabits. According to Reference 1, as shown in FIG. 10, in the TFHE computation, it took 13 milliseconds to perform bootstrap processing for each bit.
[0100] FIG. 10 shows the performance (hereinafter referred to as "secure computation performance") of the secure computation performed by the secure computation system 100, obtained as a result of an experiment (hereinafter referred to as "secure computation performance evaluation experiment") in which secure computation is performed by the secure computation system 100. In FIG. 10, the secure computation performance obtained as a result of the secure computation performance evaluation experiment is shown in enclosure D101. FIG. 10 shows that the Number of Mult. was infinite and the key length was 256 bits for the secure computation performance obtained as a result of the secure computation performance evaluation experiment. FIG. 10 shows that the secure computation performance evaluation experiment included noise generated by probability calculation when decrypting plaintext.
[0101] (Modification) The terminal device 1 and the secret computing device 2 may be implemented using a plurality of information processing devices communicably connected via a network. In this case, each functional unit of the terminal device 1 and the secret computing device 2 may be distributed and implemented in a plurality of information processing devices. For example, the random bit string generating unit 220, the encrypted data acquiring unit 210, the secret computing unit 230, and the output control unit 250 may be implemented in different information processing devices.
[0102] The terminal device 1 and the secure computing device 2 do not necessarily need to be implemented as different devices. For example, the terminal device 1 and the secure computing device 2 may be implemented as a single device having both functions.
[0103] All or part of the functions of the terminal device 1 and the secure computing device 2 may be realized using hardware such as an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array). The program may be recorded on a computer-readable recording medium. The computer-readable recording medium is, for example, a portable medium such as a flexible disk, a magneto-optical disk, a ROM, or a CD-ROM, or a storage device such as a hard disk built into a computer system. The program may be transmitted via a telecommunication line.
[0104] Note that the first input value A in Fig. 8 is an example of first encrypted data. The second input value B in Fig. 8 is an example of second encrypted data. Note that obtaining a value indicating a product in a probabilistic numerical expression means obtaining a product in the probabilistic numerical expression. Note that obtaining a value indicating a sum in a probabilistic numerical expression means obtaining a sum in the probabilistic numerical expression.
[0105] In addition, when encrypting the probability-expressing plaintext obtained by the expression conversion unit 120, the encryption unit 130 does not necessarily have to encrypt each bit of the probability-expressing plaintext into one cipher individually. The encryption unit 130 may encrypt multiple bits together into one cipher.
[0106] Encrypting multiple bits of a probability expression plaintext together into one cipher means, for example, regarding U bits (U is an integer equal to or greater than 2) of the probability expression plaintext as a U-dimensional vector, and encrypting the U-dimensional vector into one cipher. Note that regarding U bits as a U-dimensional vector means using the U bits as a single piece of information for information processing. Note that encrypting multiple bits of a probability expression plaintext together into one cipher means, for example, regarding the U bits of the probability expression plaintext as a U-order polynomial, and encrypting the U-dimensional polynomial into one cipher. Note that regarding U bits as a U-dimensional polynomial means using the U bits as a single piece of information for information processing.
[0107] When U bits are encrypted together into one cipher in this way, the amount of calculation required for encryption by the encryption unit 130 is 1 / N of the amount of calculation required when each bit of the probability expression plaintext is encrypted individually into one cipher. Therefore, by encrypting multiple bits of the probability expression plaintext together into one cipher, the amount of calculation required for encryption is reduced.
[0108] Furthermore, when U bits are collectively encrypted into one cipher in this way, the amount of calculation in the secure calculation unit 230 is also 1 / N. Therefore, by collectively encrypting multiple bits of the probability expression plaintext into one cipher, the amount of calculation required for secure calculation is reduced.
[0109] Hereinafter, the process of encrypting multiple bits of a probability-representing plaintext into one cipher is called packing. The process of obtaining the sum of the encrypted data generated by the packing process is performed by concatenated stochastic addition, which will be described below, instead of the process shown in FIG.
[0110] <Concatenated stochastic addition> Concatenated stochastic addition may be performed in the secure computation unit 230. Concatenated stochastic addition is a process for obtaining a sum S, which is a value indicating the sum of two encrypted data in a probabilistic numerical representation, but unlike the process shown in FIG. 8, it is a process that does not use a random bit string R.
[0111] In concatenated stochastic addition, when obtaining a value indicating the sum of V bits of encrypted data and W bits of encrypted data, a process of concatenating V bits of encrypted data and W bits of encrypted data (hereinafter referred to as a "concatenation process") is performed. Specifically, the concatenation process is a process of concatenating V bits of encrypted data and W bits of encrypted data to generate (V+W) bits of data.
[0112] For example, when concatenating encrypted data "1011" and encrypted data "0101", the concatenation process generates 8-bit data "10110101" using the encrypted data "1011" and "0101".
[0113] For example, when concatenating encrypted data "0100" and encrypted data "1111", the concatenation process generates 8-bit data "01001111" using the encrypted data "0100" and "1111".
[0114] In this way, the concatenation process is a process in which, for two encrypted data to be added together, the bit string representing one encrypted data is added to the end of the bit string representing the other encrypted data.
[0115] In concatenated stochastic addition, the data generated by the concatenation process is the sum S. Therefore, the secure computation unit 230 obtains the result of the concatenation as the sum.
[0116] Each piece of data represented by a probabilistic numerical expression can also be considered as information representing the probability of occurrence of 0 or 1. Therefore, for example, when a concatenation process is executed to concatenate 4-bit information Q and 4-bit information P, the probability of occurrence of each of 0 and 1 in the data resulting from the execution is the average of the probability of occurrence in information Q and the probability of occurrence in information P. Since the average corresponds to the process of obtaining the sum of the numerator, the concatenation process corresponds to the process of obtaining the sum of two pieces of data.
[0117] When concatenated stochastic addition is performed in the secure calculation unit 230, the process for obtaining the product may be any process capable of obtaining the product, for example, a stochastic secure product obtaining process.
[0118] Note that the encrypted data used in concatenated stochastic addition is data generated by packing processing, and is the result of multiple bits of plaintext being encrypted into one cipher.
[0119] The process of concatenated stochastic addition for obtaining the sum of the first packed data and the second packed data is an example of a process for determining the value of each digit of a bit string representing the sum to be either the value of the corresponding digit of the first packing bit string or the value of the corresponding digit of the second packing bit string. The first packed data is encrypted data generated by the packing process. It is therefore an example of first encrypted data. The second packed data is encrypted data generated by the packing process. It is therefore an example of second encrypted data. The first packing bit string is a bit string representing the first packed data and is expressed in an expression format of a probabilistic numerical expression. The second packing bit string is a bit string representing the second packed data and is expressed in an expression format of a probabilistic numerical expression.
[0120] Although an embodiment of the present invention has been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and designs that do not deviate from the gist of the present invention are also included. [Explanation of symbols]
[0121] 1... terminal device, 2... secure computation device, 9... network, 100... secure computation system, 11... control unit, 12... communication unit, 13... storage unit, 14... input unit, 15... output unit, 91... processor, 92... memory, 110... plaintext acquisition unit, 120... representation conversion unit, 130... encryption unit, 140... decryption unit, 150... communication control unit, 160... output control unit, 170... representation inverse conversion unit, 21... control unit, 22... communication unit, 23... storage unit, 24... input unit, 25... output unit, 93... processor, 94... memory, 210... encrypted data acquisition unit, 220... random bit string generation unit, 230... secure computation unit, 240... communication control unit, 250... output control unit
Claims
1. a secret calculation unit that performs secret calculation on encrypted data obtained by encrypting plaintext represented in a predetermined representation format for probability calculation using a homomorphic encryption method; Equipped with The secure computation includes a process of obtaining a sum and a process of obtaining a product, the secret calculation unit, in obtaining a sum which is a sum of first encrypted data which is the first encrypted data and second encrypted data which is the second encrypted data, determines a value of each digit of a bit string which represents the sum to be either a value of a corresponding digit of a bit string which represents the first encrypted data and is expressed in the expression format, or a value of a corresponding digit of a bit string which represents the second encrypted data and is expressed in the expression format. Secret computing device.
2. a random bit string generation unit that generates a random bit string, the random bit string being a bit string in which each digit has one of two values determined based on a predetermined probability distribution; Equipped with the secret calculation unit uses the random bit sequence when obtaining the sum, and when obtaining the sum that is the sum of the first encrypted data and the second encrypted data, determines, based on a value of a corresponding digit of the random bit sequence, a value of a corresponding digit of a bit sequence that represents the sum to be either a value of a corresponding digit of a bit sequence that represents the first encrypted data and is expressed in the expression format, or a value of a corresponding digit of a bit sequence that represents the second encrypted data and is expressed in the expression format. The secure computing device according to claim 1 .
3. The secure computing device according to claim 2 , wherein the bit string expressed in the predetermined representation format is a Bernoulli string expressed in a unipolar code.
4. The secure computing device according to claim 2 , wherein the bit string expressed in the predetermined representation format is a Bernoulli string expressed in a bipolar code.
5. The secure computing device according to claim 2 , wherein the bit string expressed in the predetermined representation format is a Bernoulli string expressed as an inverted bipolar code.
6. the secret calculation unit, in obtaining a product of first encrypted data, which is the first encrypted data, and second encrypted data, which is the second encrypted data, obtains a value of each digit of a bit string representing the product by a two-input logical operation in accordance with the expression format, using as input a value of a corresponding digit of a bit string representing the first encrypted data and expressed in the expression format, and a value of a corresponding digit of a bit string representing the second encrypted data and expressed in the expression format; The secure computing device according to claim 2 .
7. When the expression format is a unipolar code, the logical operation according to the expression format is a logical AND. The secure computing device according to claim 6.
8. When the expression format is a bipolar code, the logical operation according to the expression format is a negation of an exclusive OR. The secure computing device according to claim 6.
9. the secret calculation unit, in obtaining the sum that is the sum of the first encrypted data and the second encrypted data, concatenates a bit string representing the first encrypted data and a bit string representing the second encrypted data, and obtains the result of the concatenation as the sum; the first encrypted data and the second encrypted data are the result of multiple bits of the plaintext being encrypted into one cipher; The secure computing device according to claim 1 .
10. A secret calculation step of performing a secret calculation on encrypted data obtained by encrypting plaintext represented in a predetermined representation format for probability calculation by a computer using a homomorphic encryption method; having The secure computation includes a process of obtaining a sum and a process of obtaining a product, the secret calculation step, in obtaining a sum which is a sum of first encrypted data which is the first encrypted data and second encrypted data which is the second encrypted data, determines a value of each digit of a bit string which represents the sum to be either a value of a corresponding digit of a bit string which represents the first encrypted data and is expressed in the expression format, or a value of a corresponding digit of a bit string which represents the second encrypted data and is expressed in the expression format; Confidential calculation method.
11. A program for causing a computer to function as the secure computing device according to any one of claims 1 to 9.
Citation Information
Patent Citations
Homomorphic Encryption with Optimized Encoding
US20170134157A1
