PROXY CALCULATION SYSTEM, METHOD, REQUEST DEVICE AND PROGRAM THEREOF

Abstract

Where G and H are cyclic groups, M is an integer of two or more, i=1, . . . , M, f is a homomorphic function of mapping a member xi of group H to group G, Ri and R0 are random variables with a value in group G, ri is a realized value of the random variable Ri,, r0 is a realized value of the random variable R0, and ai is a random number of an integer of 0 or more, a random number generation unit 11 generates random numbers a1, a2, . . . , aM. A sampler 21 is capable of calculating f(x1)r1, f(x2)r2, . . . , f(xM)rM to obtain a calculation result thereof as z1, z2, . . . , zM, respectively. A power calculation unit 12 calculates (z1)a1, (z2)a2, . . . , (zM)aM. An extended randomizable sampler 22 is capable of calculating f(x1a1×x2a2× . . . ×xMaM)r0 to obtain a calculation result z0 thereof. A determination unit 16 determines whether or not (z1)a1×(z2)a2× . . . ×(zM)aM=z0.

Description

TECHNICAL FIELD

The present invention relates to a calculation technique using a computer. In particular, it relates to a technique of performing calculation using a result of calculation performed by another calculator.

BACKGROUND ART

The inventor has proposed a technique of calculating f(x) for an input x using a randomizable sampler and a sampler described below (for example, refer to Non-patent literature 1).

R is defined as a random variable with a value in group G. What returns w^ar′ corresponding to a sample r′ in accordance with the random variable R every time an integer a of 0 or more for w∈G is given is referred to as a randomizable sampler with an error R for w. In the randomizable sampler, distribution of the random variable R is unrelated to the integer a.

In addition, what returns wr′ corresponding to the sample r′ in accordance with the random variable R every time a demand is received in w∈G is referred to as a sampler with an error R for w.

The randomizable sampler and the sampler calculate w^ar′ and wr′, respectively, using a function called black box.

In a case that, when r₁ and r₂ are realized values of the random variable R, an output f(x)^ar₁ of the randomizable sampler with an error R for f(x) is equal to the a-th power (f(x)r₂)^a of an output f(x)r₂ of the sampler with an error R for f(x), r₁ and r₂ are unit elements e_g of group G with high probability. Accordingly, since f(x)r₂=f(x)e_g=f(x) in this case, the output f(x)r₂ of the sampler becomes f(x) intended to be obtained.

PRIOR ART LITERATURE

Non-Patent Literature

Non-patent literature 1: Go Yamamoto, Tetsutaro Kobayashi, “Self-Correctors for Cryptographic Modules”, The 2011 Symposium on Cryptography and Information Security, 2F1-1, 2001.

SUMMARY OF THE INVENTION

Problems to be Solved by the Invention

In the technique described in Background Art, in order to calculate a value of f(x) for certain x, it is necessary to make the randomizable sampler and the sampler calculate at least twice using the black box function. Therefore, when M is an integer of two or more, in order to calculate f(x₁), f(x₂), . . . , f(x_M), respectively, for M items of inputs of x₁, x₂, . . . , x_M, it is necessary to make the randomizable sampler and the sampler calculate at least 2M times using black box. This is because it is necessary to make the randomizable sampler and the sampler calculate at least twice using black box for the respective M items of inputs of x₁, x₂, . . . , x_M.

It is an object of the present invention is to provide a technique of calculating f(x₁), f(x₂), . . . , f(x_M), respectively, for M items of inputs of x₁, x₂, . . . , x_M while reducing the number of times to calculate using black box to less than 2M times.

Means to Solve the Problems

According to an aspect of the present invention, random numbers a₁, a₂, . . . , a_M are generated. It is enabled to calculate f(x₁)r₁, f(x₂)r₂, . . . , f(x_M)r_M to obtain a calculation result thereof as z₁, z₂, . . . , z_M, respectively. (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM is calculated. It is enabled to calculate f(x₁^a1×x₂^a2× . . . ×x_M^aM)r₀ to obtain a calculation result z₀ thereof. It is determined whether or not (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀, where G and H are cyclic groups, M is an integer of two or more, i=1, . . . , M, f is a homomorphic function of mapping a member x_i of group H to group G, R_i and R₀ are random variables with a value in group G, r_i is a realized value of the random variable R_i, r₀ is a realized value of the random variable R₀, and a_i is a random number of an integer of 0 or more.

Effects of the Invention

It is enabled to make the number of times to calculate using black box a minimum of M+1 times.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram illustrating configuration of a proxy calculation system of an embodiment.

FIG. 2 is a block diagram illustrating configuration of a request device and a calculation device of an embodiment.

FIG. 3 is a flowchart showing process in a proxy calculation system of an embodiment.

DETAILED DESCRIPTION OF THE EMBODIMENTS

Descriptions are given below to an embodiment of the present invention with reference to the drawings.

G and H are cyclic groups, M is an integer of two or more, i=1, . . . , M, f is a homomorphic function of mapping a member x_i of group H to group G, R_i and R₀ are random variables with a value in group G, r_i is a realized value of the random variable R_i, r₀ is a realized value of the random variable R₀, and a_i is a random number of an integer of 0 or more.

A random number generation unit 11 of a request device 1 generates random numbers a₁, a₂, . . . , a_M of integers of 0 or more (step S1). Information on the random numbers a₁, a₂, . . . , a_M thus generated is sent to a power calculation unit 12 and a calculation device 2.

A control unit 13 substitutes 1 for t (step S2).

A first request unit 14 of the request device 1 transmits request information to request calculation of f(x₁)r₁, f(x₂)r₂, . . . , f(x_M)r_M to the calculation device 2 (step S3). For example, the request information includes x₁, x₂, . . . , x_M as members of group H. When the calculation device 2 is capable of calculating f(x₁)r₁, f(x₂)r₂, . . . , f(x_M)r_M, the request information does not necessarily include x₁, x₂, . . . , x_M.

A sampler 21 of the calculation device 2 is capable of calculating f(x₁)r₁, f(x₂)r₂, . . . , f(x_M)r_M, and those calculation results are z₁, z₂, . . . , z_M, respectively (step S4). The calculation results z₁, z₂, . . . , z_M are transmitted to the request device 1. That is, the calculation result for f(x₁)r₁ becomes z₁, the calculation result for f(x₂)r₂ becomes z₂, . . . , and the calculation result for f(x_M)r_M becomes z_M.

Herein, to be capable of calculating means to be possible to calculate with probability of not less than non-negligible probability. The non-negligible probability is probability of not less than 1/Φ(k) where a polynomial as a monotone non-decreasing function for a security parameter k is a polynomial Φ(k).

Here, to calculate f(x)r is to calculate a value of a formula defined as f(x)r. As long as it is possible to finally calculate the value of the formula f(x)r, the method of calculation in the middle does not matter. This is similar for calculation of other formulae herein.

The power calculation unit 12 of the request device 1 raises the calculation results z₁, z₂, . . . , z_M to the power of a₁, a₂, . . . , a_M, respectively (step S5). That is, it calculates (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM. In a first memory unit 17, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM are memorized.

Here, where i=1, 2, . . . , M, ai as the superscripts on the right of (z_i) means a_i. In such a manner herein, in a case of expressing as α^βγ where α is a first character, β is a second character, and γ is a numeral, that βγ means β_γ, that is, β with a subscript γ.

A second request unit 15 of the request device 1 transmits request information to request calculation of f(x₁^a1×x₂^a2× . . . ×x_M^aM)r₀ to the calculation device 2 (step S6). For example, the request information includes x₁, x₂, . . . , x_M as members of group H. As long as the calculation device 2 is capable of calculating f(x₁)r₁,f(x₂)r₂, . . . ,f(x_M)r_M, the request information does not necessarily include x₁, x₂, . . . , x_M.

An extended randomizable sampler 22 of the calculation device 2 is capable of calculating f(x₁^a1×x₂^a2× . . . ×x_M^aM)r₀, and the calculation result is z₀ (step S7). The calculation result z₀ is transmitted to the request device 1 and memorized in a second memory unit 18 of the request device 1.

The extended randomizable sampler 22 is to randomize f(x₁×x₂× . . . ×x_M) using random numbers a₁, a₂, . . . , a_M while an error is given based on a random variable R₀ having distribution independent of the random numbers a₁, a₂, . . . , a_M.

The extended randomizable sampler 22 calculates a value of f(x₁^a1×x₂^a2× . . . ×x_M^aM)r₀ using a probability function F as a black box that takes, for example, x₁^a1×x₂^a2× . . . ×x_M^aMh^s as an argument. That is, the extended randomizable sampler 22 calculates a value of, for example, F(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)f(h)^−s to get z₀. Here, h is a generator of group H and s is a random number of an integer of 0 or more. Here, the probability function F is a probability function that outputs a value same as the function f with probability of not less than predetermined probability. That is, F(x_c)=f(x_c) holds for an arbitrary input x_c∈H with probability of not less than predetermined probability.

Where U is uniform distribution of group H and the random variable R₀=F(U)f(U)⁻¹, the distribution of F(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)f(h)^−s becomes equal to the distribution of f(x₁^a1×x₂^a2× . . . ×x_M^aM)R₀. It is because F(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)f(h)^−s=f(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)R₀f(h)^−s=f(x₁^a1×x₂^a2× . . . ×x_M^aM)f(h)^sR₀f(h)^−s=f(x₁^a1×x₂^a2× . . . ×x_M^aM)R₀. In the development of the formula described above, due to the property as a homomorphic function of f and R₀=F(U)f(U)⁻¹=F(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)f(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)⁻¹, a property of F(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)=f(x₁^a1×x₂^a2× . . . ×x_M^aMh^s)R₀ is used.

A determination unit 16 of the request device 1 determines whether or not (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀ using (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM loaded from the first memory unit 17 and z₀ loaded from the second memory unit 18 (step S8).

In a case of (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀, then z₁, z₂, . . . , z_M in that case become f(x₁), f(x₂), . . . , f(x_M), respectively, as described later. Therefore, an output unit 19 outputs z₁, z₂, . . . , z_M in that case as f(x₁), f(x₂), . . . , f(x_M), respectively (step S9).

In a case of (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀, then (f(x₁)r₁)^a1×(f(x₂)r₂)^a2× . . . ×(f(x_M)r_M)^aM=f(x₁^a1×x₂^a2× . . . ×x_M^aM)r₀, and considering that the function f is a homomorphic function, r₁^a1×r₂^a2× . . . ×r_M^aM=r₀. The inventor has found that, in this case, there is a very high possibility that r₁, r₂, . . . , r_M are unit elements e_g of group G.

This is because, firstly, by the principle of proxy calculation described in Non-patent literature 1, determination of whether or not (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀ enables to determine whether or not z₀=f(x₁^a1×x₂^a2× . . . ×x_M^aM) with very high probability. Secondly, since f(x₁^a1×x₂^a2× . . . ×x_M^aM)=f(x₁)^a1×f(x₂)^a2× . . . ×f(x_M)^aM holds, z₀=f(x₁)^a1×f(x₂)^a2× . . . ×f(x_M)^aM holds where z₀=f(x₁^a1×x₂^a2× . . . ×x_M^aM). Thirdly, when repeatedly executing an arbitrary method of calculating z₀=f(x₁^a1×x₂^a2× . . . ×x_M^aM) where a_i is a random number of an integer of 0 or more, it is possible to derive f(x_i) for each i from z₀=f(x₁)^a1×f(x₂)^a2× . . . ×f(x_M)^aM using methods described in Reference Literatures 1 and 2. Fourthly, in a method very same as the method used in the third reason, it is also possible to derive z_i for each i from (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM by using the very value same as the integer a_i used in the third reason. As just described, when (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀=f(x₁^a1×x₂^a2× . . . ×x_M^aM)=f(x₁)^a1×f(x₂)^a2× . . . ×f(x_M)^aM, there is a very high possibility that z_i=f(x_i) for each i. Accordingly, when (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀, there is a very high possibility that r₁, r₂, . . . , r_M are unit elements e_g of group G.

Reference Literature 1: M. Bellare, J. Garay, and T. Rabin, “Fast Batch Verification for Modular Exponentiation and Digital Signatures,” Advances in Cryptology-EUROCRYPT '98, LNCS 1403, pp. 236-250, Springer-Verlag, 1998.

Reference Literature 2: K. Chida and G Yamamoto, “Batch Processing of Interactive Proofs,” Topics in Cryptology-CT-RSA2007, LNCS 4377, pp. 196-207, Springer, 2007.

Therefore, in a case of (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀, then z₁=f(x₁)r₁=f(x₁)e_g=f(x₁), z₂=f(x₂)r₂=f(x₂)e_g=f(x₂), . . . , and z_M=f(x_M)r_M=f(x₁)e_g=f(x_M).

The control unit 13 determines whether or not t=T (step S10). T is a preset integer of not less than 1. As long as t=T, the output unit 19 outputs information of not being capable of calculation, for example, a symbol “⊥” (step S11) and the process is terminated. In a case of not t=T, the control unit 13 increments t by one, that is, substitutes t+1 for t (step S12) and the process goes back to step S3.

Information of not being capable of calculation (in this example, the symbol “⊥”) means that reliability of the calculation device 2 correctly performing calculation is below the standard defined by T. In other words, it means that correct arithmetic operations are not performed repeatedly in T times.

After step S12, similar to the above, the request device 1 causes the calculation device 2 to calculate all or part of z₁, z₂, . . . , z_M (steps S3, S4). The power calculation unit 12 performs the power of all or part of calculated z₁, z₂, . . . , z_M, respectively, using corresponding random numbers a₁, a₂, . . . , a_M (step S5). The powered (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM are memorized in the first memory unit 17.

In the process of step S8, in a case that at least one of z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM are memorized in a plurality in the first memory unit 17 and the second memory unit 18, the determination unit 16 determines whether or not the relationship of (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀ is satisfied, respectively, for all sets (z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM) of z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM configured by selecting each one from z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM, respectively, memorized in the first memory unit 17 and the second memory unit 18 (step S8).

For example, where k₀, k₁, . . . , k_M are integers of 1 or more, k₀ items of z₀, k₁ items of z₁, . . . , k_M items of z_M are memorized in the first memory unit 17 and the second memory unit 18. In this case, there are k₀×k₁×k₂× . . . ×k_M items of a set (z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM) of z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM configured by selecting each one from z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM, respectively, memorized in the first memory unit 17 and the second memory unit 18. Accordingly, the determination unit 16 determines whether or not the relationship of (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀ is satisfied, respectively, for k₀×k₁×k₂× . . . ×k_M items of a set (z₀, (z₁)^a1, (z₂)^a2, . . . , (z_M)^aM).

In a case of satisfying the relationship of (z₁)^a1×(z₂)^a2× . . . ×(z_M)^aM=z₀ in the process of step S8 for the first time, the sampler 21 performs calculation of z₁, z₂, . . . , z_M using black box M times and the extended randomizable sampler 22 performs calculation of z₀ using black box F once, and thus the number of times to refer to black box is M+1 times in total. In such a manner, it is enabled to make the number of times to calculate using black box to be a minimum of M+1 times.

Modifications

The random variable R₀, R₁, . . . , R_M may be same and may also be different.

The random number generation unit 11 generates uniform random numbers, thereby making the safety of the proxy calculation system highest. However, in a case that the intended safety level is not so high, the random number generation unit 11 may also generate random numbers that are not uniform random numbers.

In addition, the request device 1 may also provide a plurality of types of request information to the calculation device 2 in collection to obtain a plurality of corresponding z₀, z₁, z₂, . . . , z_M in collection. This enables to reduce the number of times to communicate between the request device 1 and the calculation device 2.

The communication of data between each unit of the request device 1 and the calculation device 2 may be performed directly and may also be via a memory unit, not shown.

Other than them, the present invention is not limited to the above embodiment. For example, the above various types of process may not only be executed in time series in accordance with the description, but also be executed in parallel or individually in accordance with the capacity of the device that executes the process or necessity.

For example, in a case that at least one of z₀, (z₁)^a1, (z₂)^s2, . . . , (z_M)^aM are memorized in plural in the first memory unit 17 and the second memory unit 18, as illustrated with broken lines in FIG. 3, the determination unit 16 may also perform process in step S8 between steps S5 and S6.

In addition, in a case that the above configuration is achieved using a computer, the process contents of functions that each device should have is described by a program. Then, the program is executed using a computer, thereby achieving the above process function on the computer.

It is enabled to store the program in which the process contents are described in a computer-readable storage medium. The computer-readable storage medium may include, for example, a magnetic storage device, an optical disk, a magneto-optical storage medium, a semiconductor memory, and the like.

Other than those, appropriate modifications are naturally made within the scope of the present invention.

Claims

1. A proxy calculation system comprising: a random number generation unit to generate random numbers a1, a2, . . . , aM;a sampler to be capable of calculating f(x1)r1, f(x2)r2, . . . , f(xM)rM to obtain a calculation result thereof as z1, z2, . . . , zM, respectively;a power calculation unit to calculate (z1)a1, (z2)a2, . . . , (zM)aM;an extended randomizable sampler to be capable of calculating f(x1a1×x2a2× . . . ×xMaM)r0 to obtain a calculation result z0 thereof; anda determination unit to determine whether or not (z1)a1×(z2)a2× . . . ×(zM)aM=z0,where G and H are cyclic groups, M is an integer of two or more, i=1, . . . , M, f is a homomorphic function of mapping a member xi of group H to group G, Ri and R0 are random variables with a value in group G, ri is a realized value of the random variable Ri,, r0 is a realized value of the random variable R0, and ai is a random number of an integer of 0 or more.

2. The proxy calculation system according to claim 1, wherein where a probability function that outputs a value same as the function f with probability of not less than predetermined probability is F, a generator of group H is h, and a random number of an integer of 0 or more is s,the extended randomizable sampler calculates a value of F(x1a1×x2a2× . . . ×xMaMhs)f(h)−s to get a calculation result thereof as z0.

3. A proxy calculation method, where G and H are cyclic groups, M is an integer of two or more, i=1, . . . , M, f is a homomorphic function of mapping a member xi of group H to group G, Ri and R0 are random variables with a value in group G, ri is a realized value of the random variable Ri,, r0 is a realized value of the random variable R0, and ai is a random number of an integer of 0 or more, the method comprising: random number generating that a random number generation unit generates random numbers a1, a2, . . . , aM;sample extracting that a sampler is capable of calculating f(x1)r1, f(x2)r2, . . . , f(xM)rM to obtain a calculation result thereof as z1, z2, . . . , zM, respectively;power calculating that a power calculation unit calculates (z1)a1, (z2)a2, . . . , (zM)aM;extended randomizable sample extracting that an extended randomizable sampler is capable of calculating f(x1a1×x2a2× . . . ×xMaM)r0 to obtain a calculation result z0 thereof; anddetermining that a determination unit determines whether or not (z1)a1×(z2)a2× . . . ×(zM)aM=z0.

4. The proxy calculation method according to claim 3, wherein where a probability function that outputs a value same as the function f with probability of not less than predetermined probability is F, a generator of group H is h, and a random number of an integer of 0 or more is s,the extended randomizable sampler calculates a value of F(x1a1×x2a2× . . . ×xMaMhs)f(h)−s to get a calculation result thereof as z0.

5. A request device the device comprising: a random number generation unit to generate random numbers a1, a2, . . . , aM;a power calculation unit to calculate (z1)a1, (z2)a2, . . . , (zM)aM using a calculation result z1, z2, . . . , zM by a sampler to be capable of calculating f(x1)r1, f(x2)r2, . . . , f(xM)rM;a determination unit to determine whether or not (z1)a1×(z2)a2× . . . ×(zM)aM=z0 using a calculation result of the power calculation unit and a calculation result z0 by an extended randomizable sampler to be capable of calculating f(x1a1×x2a2× . . . ×xMaM)r0,where G and H are cyclic groups, M is an integer of two or more, i=1, . . . , M, f is a homomorphic function of mapping a member xi of group H to group G, Ri and R0 are random variables with a value in group G, ri is a realized value of the random variable Ri,, r0 is a realized value of the random variable R0, and ai is a random number of an integer of 0 or more.

6. The request device according to claim 5, wherein where a probability function that outputs a value same as the function f with probability of not less than predetermined probability is F, a generator of group H is h, and a random number of an integer of 0 or more is s,the extended randomizable sampler calculates a value of F(x1a1×x2a2× . . . ×xMaMhs)f(h)−s to get a calculation result thereof as z0.

7. A program, causing a computer to function as a part of the request device according to claim 5 or 6.

8. A computer-readable storage medium in which a program causing a computer to function as a part of the request device according to claim 5 or 6 is stored.