Calculation unit for executing a cryptographic protocol

Abstract

A calculation unit in which the inverse of an integer modulo a large number is determined provides a series of binary numbers prime with the large number, divides those numbers into two groups at random, and effects the products of the numbers of each group.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The invention relates to a calculation unit for executing at least one part of a cryptographic protocol including determining the inverse of an integer modulo a large number.

The invention relates more particularly to means for determining this kind of inverse quickly.

2. Description of the Prior Art

Some cryptographic protocols, for example those known as the Digital Signature Algorithm (DSA) and the Elliptic Curve DSA (ECDSA), have to calculate the inverse of a random integer modulo a large number repetitively. For more details on such protocols see IEEE publication P1363/D13 and especially sections 6.2.7 and 7.2.7 thereof. The aforementioned large number is denoted N in the remainder of this description.

Two numbers are the inverse of each other if their product is identically equal to 1 modulo said large number. For example, the numbers 3 and 5 are the inverse of each other modulo 14. This is because 3×5=15≡1 modulo 14.

One prior art method of performing this calculation is based on the following equation:y=x⁻¹=x^φ(N)−1  modulo Nin which φ(N) is the number of integers prime with N and less than N. This is Euler's “totient” function.

If N is prime, the following simplification is valid:φ(N)=N−1and consequently:y=x⁻¹=x^N−2

In binary, this calculation requires 3n/2 multiplications of n×n bits where n=log₂ N, in other words of the order of 160 multiplications in practice.

The invention proposes a faster calculation method in which two numbers that are the inverse of each other modulo N are generated at the same time, by a specific calculation, for example based on a random number.

SUMMARY OF THE INVENTION

To be more precise, the invention provides a calculation unit adapted to execute at least one part of a cryptographic protocol including determining the inverse of an integer modulo a large number, which unit includes:

• • means for providing a series of binary numbers that are prime with and less than the large number, a number from the series being the inverse of the product of the others modulo the large number,
  • means for dividing all the numbers of the series into two groups at random, and
  • means for effecting the products of the numbers of each group, the results of such products constituting two of the required numbers that are the inverse of each other.

Two numbers are prime to each other if they have no common divisor.

A preferred embodiment of the calculation unit includes:

• • means for generating a random number,
  • means for representing two of the required numbers initialized to 1 at the beginning of each calculation,
  • means for selecting bits of the random binary number one-by-one and a corresponding number from the series, and
  • means for multiplying the current value of one of the required two numbers by the corresponding number if the bit is equal to 0 or for multiplying the current value of the other of the required two numbers by the corresponding number if the bit is equal to 1, the result of the multiplication constituting the new current value of one or the other of the required two numbers, respectively, and the above operations being effected for all the aforementioned selections, the final current values contained in the means for representing two numbers constituting the required two numbers that are the inverse of each other.

In one embodiment, the means for providing the series of binary numbers include means for storing the numbers constituting that series. Those numbers are precalculated during a preparatory phase that does not have to be executed again each time a pair of numbers that are the inverse of each other modulo N is calculated.

Another embodiment of the means for providing the series include means for generating at least some numbers of the series in real time.

If most of the numbers of the set can be calculated using a fast execution function, time is saved compared to the cited first embodiment in which the numbers are read from a memory.

The precalculation means include means for determining a set of numbers that are prime with the large number, the set constituting numbers of the series, and means for determining the inverse of the product of numbers of the set modulo said large number, the inverse constituting a particular number from the series and completing it.

If all the numbers of the series must be stored, the precalculation means are used once and all the results are stored. On the other hand, if some numbers of the set are obtained by a fast calculation, the precalculation means are used to choose the numbers from the set that can be recalculated each time, based on the use of a fast execution function. It is the choice of numbers from the series that enables an adequate fast execution function to be “found”. When that function has been determined and “implemented”, it is generally sufficient to store, on the one hand, a number from the set, constituting a point of departure for calculating the other numbers of the set and, on the other hand, the inverse of the product of the numbers of said set (modulo the large number). All the elements are then available that are needed to obtain the numbers of the series, essentially by calculation.

For example, the calculation unit can include storage means containing, on the one hand, the smallest number of the set and, on the other hand, the inverse of the product of the numbers of the set modulo the large number, and the generator means can include means for applying a fast execution function successively, starting from the smallest number, to obtain a succession of numbers constituting the set, the series being formed by the numbers and the inverse.

In a preferred example, the fast execution function obtains each number by adding 1 to the preceding number. This type of fast execution function can be used in particular if the large number is prime. If it is not, it is often possible to find a range of numbers in which all the consecutive numbers are prime with the large number and less than it.

The invention will be better understood and other advantages of the invention will become more clearly apparent in the light of the following description of various embodiments of a calculation unit according to the invention, which description is given by way of example only and with reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a flowchart showing various calculation steps for obtaining two numbers that are the inverse of each other modulo a large number.

FIG. 2 is a block diagram of a first embodiment of a calculation unit according to the invention.

FIG. 3 is a block diagram of another embodiment of a calculation unit according to the invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

The functioning of the calculation unit defined above is justified by the following demonstration:

• Let S₀ be the inverse modulo N of the product of a set of k integers S₁ to S_k all of which are prime with said large number N and less than N,
• Let:

$S_0={\left(\prod _{i=1}^k{S}_i\right)}^{-1}\mathrm{modulo}N$ It is shown that S₀ is also prime with N.It is easy to verify that:

$\prod _{i=1}^k{S}_i\equiv 1\mathrm{modulo}N$

Consequently, if the k+1 numbers S_i of this product are divided at random into two groups of numbers and if the respective products Π₁ and Π₂ of the numbers from the two groups are obtained, then the two numbers x and y that result from the products Π₁ and Π₂ are the inverse of each other modulo N.

In other words, the basic principle of the invention leads to choosing a number k and a large number N, for example, and then, in a precalculation phase, determining, on the one hand, a set of k numbers S₁ . . . S_k all less than N and prime with N and, on the other hand, an additional number S₀ that constitutes the inverse of the product of the numbers S₁ . . . S_k modulo N. When this precalculation is effected, a series of k+1 binary numbers is available. These numbers can be stored, for example, to be used whenever necessary, during the execution of a cryptographic calculation to determine two numbers x, y that are the inverse of each other modulo N.

The calculation phase proper can proceed in accordance with the FIG. 1 flowchart.

• • Step E₁ is an initialization phase in which two values x and y are initialized to 1 (which is the neutral element of the multiplication).

Also, a binary number e of k+1 bits is chosen at random and is thus made up of k+1 bits (e₀ . . . e_k) each having the value 0 or 1.

• • Step E₂ initializes to 0 a counter i.
  • Step E₃ is a test to verify if the index i is greater than k. A positive test result indicates that the calculation is finished. A negative test results leads on to step E₅.
  • Step E₄ “obtains” the number S_i. If the numbers S_i are stored, this operation simply consists in reading one of the numbers in a memory. However, as indicated above, it is possible to calculate most of the numbers S_i in real time (with the exception of S₀ which is the subject of a particular precalculation). Consequently, step E₃ can either read a value in a memory or recalculate it using a “fast execution” function.
  • Step E₅ determines the value of the i^th bit of the number e chosen at random. If that bit is equal to 1, the next step is the step E₆. If that bit is equal to 0, the next step is the step E₇.
  • Step E₆ multiples the number S_i by the preceding value of y and the result becomes the new value of y. The next step is then step E₈.
  • Step E₇ multiplies the number S_i by the preceding value of x and the result becomes the new value of x. The next step is step E₈.
  • Step E₈ increments i by one unit and returns to step E₃ to verify if the index i is greater than k.

If the result of test E₄ is positive, the two values x and y represent the required two numbers that are the inverse of each other modulo N.

FIG. 2 is a block diagram showing the means for implementing the algorithm explained by the FIG. 1 flowchart. The block 12 represents the precalculation means and the block 13 represents the calculation means proper.

The precalculation means include a generator Gp for numbers that are prime with said large number N. Those numbers take the form of words on w bits. As the generator Gp generates a succession of numbers prime with N, those numbers S₁ . . . S_i . . . S_k are stored in a memory M that is part of the calculation means 13.

In parallel with this, the numbers S₁-S_k are sent to a circuit Inv which calculates the number S₀. The number S₀ is the inverse of the product of the numbers S₁ to S_k. When this calculation is completed, the value of S₀ is stored in the memory M. At this point the precalculation means can be disconnected from the calculation means 13 because the values of the numbers S₀-S_k are stored in the memory M. The calculation unit includes a random number generator Ga for generating random numbers on k+1 bits and connected to a register 14 for containing a random number e on k+1 bits. At the beginning of the calculation process, the generator Ga writes the random number into the register 14.

The memory M is connected to a selector SE₁ whose output is connected to one input of a multiplier Mu. A selector SE₂ has its output connected to the other input of the multiplier Mu. The output of the multiplier is connected to an input of a router AI which has two outputs connected to respective inputs of two registers 16, 18 in which are written two variables x and y which are the results of products of numbers from the series S₀-S_k. The two outputs of the registers x and y are looped to the two inputs of the selector SE₂. Reading the register 14 bit by bit controls the setting of both the selector SE₂ and the router AI.

To be more precise, the bits e₀ . . . e_i . . . e_k of the register 14 are scrutinized on each iteration and if the bit e_i concerned is equal to 1, then the selector SE₂ connects the output of the register 16 to an input of the multiplier Mu and the router AI connects the output of the multiplier Mu to the input of the register 16. Similarly, if e_i is equal to 0, then the selector SE₂ connects the output of the register 18 to the input of the multiplier Mu and the router AI connects the output of the multiplier Mu to the input of the register 18.

In parallel with this, on each iteration, the selector SE₁ sends the numbers S₀-S_k to the other input of the multiplier Mu in succession, one-by-one.

Consequently, on each iteration, the content of a register 16 or 18 (depending on the value of e_i) is multiplied by a number S_i and the result is written into the same register, overwriting the preceding value. When all the numbers S₀-S_k have been processed, the registers 16 and 18 contain two numbers which are the inverse of each other modulo N.

FIG. 3 shows an embodiment that differs from the FIG. 2 embodiment essentially in how the numbers S_i are obtained. Overall, the precalculation means 12 are identical to those of the FIG. 1 embodiment, i.e. they essentially include a generator for generating numbers that are prime with said large number N and a circuit Inv for generating the number S₀ from the numbers S₁-S_k.

However, in this example the number N is a chosen prime number. Consequently, any series of consecutive numbers S_i is a series of numbers that are prime with N. It is therefore possible to define a fast execution function that simply obtains one of the numbers by adding 1 to the number S_i previously obtained, as follows:S_i+1=S_i+1

In this case, the smallest number S₁ of said set is any chosen number.

There is therefore no need for the set of numbers S_i determined by the precalculation means 12 to be transferred and stored for use by the calculation means 13. It is simply sufficient to store the smallest number S₁ of said set and the number S₀ calculated by the circuit Inv.

Given that these two numbers are stored in two registers 20, 21 of the calculation means 13, it is sufficient to add means 22 for applying a fast execution function, here the function S_i+1=S_i+1, to the calculation means.

Under these conditions, the selector SE₁ is connected to the output of the means 22 and to the outputs of the two registers 20 and 21. In other words, in the series of numbers S₀ . . . S_k, only the numbers S₀ and S₁ are read in the registers and all the others, from S₂ to S_k, are calculated in real time by the means 22. It appears that obtaining the greater part of the numbers S_i by fast calculation of the kind indicated above saves time compared to the preceding embodiment, which entails reading each number in a memory.

Apart from the means connected to the selector SE₁, the remainder of the calculation means 13 is identical to the FIG. 2 embodiment, and for this reason is not described in detail again. Its functioning is also identical provided that the selector SE₁ is capable, on each iteration of the calculation, of “presenting” the successive numbers S_i to the corresponding input of the multiplier Mu.

It should be noted that the FIG. 3 embodiment can also be satisfactory even if the number N is not a prime number. In this case, it is generally possible to determine a succession of k consecutive integers that are all prime with the number N.

This condition will be met in particular if the smallest prime factor of N (denoted p) is greater than S₁+(k−1).

In this case, since p is a prime number, the sequence S₁, S₁+1, . . . S₁+k−1 is formed of k consecutive integers that are prime with p, and therefore prime with N. In this case, the means 20 for applying the fast execution function as already described can be used, i.e. the function:S_i+1=S_i+1

If the above condition is not satisfied, in other words if p≦S₁+(k−1), S₁ is chosen to be prime with p since p is prime and, each time, 1 is added to S_i. A test is executed to determine if the number obtained is prime with p and only numbers that are prime with p are retained. In this case there is obtained, instead of a series of consecutive numbers, a number of groups of consecutive numbers, all prime with N. It is then sufficient to modify accordingly the definition of the fast execution function to be able to recalculate in real time the set of numbers S₂ to S_k, whenever necessary.

More generally, whenever it is possible to define a fast execution function from a judicious selection of the numbers prime with N delivered by the generator Gp, an embodiment of the kind described with reference to FIG. 3 can be used.

It should further be noted that it is advantageous to choose the numbers in said series from smaller numbers that are all prime with and less than said large number N. To give an idea of the order of magnitude concerned, said large number N can be a binary number on 163 bits, the number w can be a binary number on 32 bits and k can be equal to 160.

Claims

1. A calculation device comprising: a cryptographic apparatus configured to execute at least one part of a cryptographic protocol including determining the inverse of an integer modulo a large number N, which apparatus includes:a precalculation circuit comprising a generator for generating a series of binary numbers Sl to Sk that are prime with and less than said large number N, a number from said series being the inverse of the product of the others modulo said large number, anda circuit for precalculating a number S0 defined as

2. The cryptographic device claimed in claim 1, wherein said calculation unit includes: means for generating a random number,a register for storing said generated random number,means for representing two of the required numbers the value of said two required numbers being initialized to 1,means for selecting bits of said random number one-by-one and a corresponding number from said series, andmeans for multiplying the current value of one of the required two numbers by said corresponding number if said bit is equal to 0 or for multiplying the current value of the other of the required two numbers by said corresponding number if said bit is equal to 1, the result of said multiplication constituting the new current value of one or the other of the required two numbers, respectively, and the above operations being effected for all the aforementioned selections, the final current values contained in said multiplying means representing two numbers constituting the required two numbers that are the inverse of each other moduIo the large number N.

3. The cryptographic device claimed in claim 1, wherein said for precalculation means include means for determining a set of numbers that are prime with said large number, said set constituting numbers of said series, and means for determining the inverse of the product of the numbers of said set modulo said large number, said inverse constituting a number from said series.

4. The cryptographic device claimed in claim 1, wherein said means for multiplying said series include means for storing the numbers thereof.

5. The cryptographic device claimed in claim 3, wherein said means for multiplying said series include means for generating some numbers of said series in real time, from the at least one of the binary numbers stored.

6. The cryptographic device claimed in claim 5, wherein said memory contains the smallest number of said set of numbers and the inverse of the product of the numbers of said set modulo said large number, and wherein said generator includes adding means implementing a fast execution function for obtaining a succession of numbers constituting said set successively, starting from said smallest number, said series being formed by the numbers of said succession of numbers and said inverse of the product of the numbers of said set modulo said large number.

7. The cryptographic device claimed in claim 6, wherein said fast execution function obtains each number of said succession of numbers by adding 1 to the preceding number.

8. The cryptographic device claimed in claim 7, wherein said large number is a chosen prime number and said smallest number of said set contained in said storage means is any chosen number.

9. The cryptographic device claimed in claim 7, wherein said large number is any chosen number and said smallest number of said set is the smallest number of a set of consecutive numbers obtained by said fast execution function, said consecutive numbers being chosen in a range of consecutive integers all of which are prime with said large number.

10. The cryptographic device according to claim 1, wherein said cryptographic protocol is a DSA cryptographic protocol.

11. The cryptographic device according to claim 1, wherein said cryptographic protocol is an ECDSA cryptographic protocol.