English
Tiếng Việt
(en)Elliptic polynomial cryptography with secret key embedding is a method that allows for the encryption of messages through elliptic polynomial cryptography and, particularly, with the embedding of secret keys in the message bit string. The method of performing elliptic polynomial cryptography is based on the elliptic polynomial discrete logarithm problem. It is well known that an elliptic polynomial discrete logarithm problem is a computationally “difficult” or “hard” problem.
1.ApplicationNumber: US-65690110-A
1.PublishNumber: US-2011200187-A1
2.Date Publish: 20110818
3.Inventor: GHOUTI LAHOUARI
IBRAHIM MOHAMMAD K.
GUTUB ADNAN A.
4.Inventor Harmonized: GHOUTI LAHOUARI(SA)
IBRAHIM MOHAMMAD K(GB)
GUTUB ADNAN A(SA)
5.Country: US
6.Claims:
(en)Elliptic polynomial cryptography with secret key embedding is a method that allows for the encryption of messages through elliptic polynomial cryptography and, particularly, with the embedding of secret keys in the message bit string. The method of performing elliptic polynomial cryptography is based on the elliptic polynomial discrete logarithm problem. It is well known that an elliptic polynomial discrete logarithm problem is a computationally “difficult” or “hard” problem.
7.Description:
(en)BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to computerized cryptographic systems and methods for encrypting communications in a computer network or electronic communications system, and particularly to a method of performing elliptic polynomial cryptography with secret key embedding.
2. Description of the Related Art
In recent years, the Internet community has experienced explosive and exponential growth. Given the vast and increasing magnitude of this community, both in terms of the number of individual users and web sites, and the sharply reduced costs associated with electronically communicating information, such as e-mail messages and electronic files, between one user and another, as well as between any individual client computer and a web server, electronic communication, rather than more traditional postal mail, is rapidly becoming a medium of choice for communicating information. The Internet, however, is a publicly accessible network, and is thus not secure. The Internet has been, and increasingly continues to be, a target of a wide variety of attacks from various individuals and organizations intent on eavesdropping, intercepting and/or otherwise compromising or even corrupting message traffic flowing on the Internet, or further illicitly penetrating sites connected to the Internet.
Encryption by itself provides no guarantee that an enciphered message cannot or has not been compromised during transmission or storage by a third party. Encryption does not assure integrity due to the fact that an encrypted message could be intercepted and changed, even though it may be, in any instance, practically impossible, to cryptanalyze. In this regard, the third party could intercept, or otherwise improperly access, a ciphertext message, then substitute a predefined illicit ciphertext block(s), which that party, or someone else acting in concert with that party, has specifically devised for a corresponding block(s) in the message. The intruding party could thereafter transmit the resulting message with the substituted ciphertext block(s) to the destination, all without the knowledge of the eventual recipient of the message.
The field of detecting altered communication is not confined to Internet messages. With the burgeoning use of stand-alone personal computers, individuals or businesses often store confidential information within the computer, with a desire to safeguard that information from illicit access and alteration by third parties. Password controlled access, which is commonly used to restrict access to a given computer and/or a specific file stored thereon, provides a certain, but rather rudimentary, form of file protection. Once password protection is circumvented, a third party can access a stored file and then change it, with the owner of the file then being completely oblivious to any such change.
Methods of adapting discrete-logarithm based algorithms to the setting of elliptic polynomials are known. However, finding discrete logarithms in this kind of group is particularly difficult. Thus, elliptic polynomial-based crypto algorithms can be implemented using much smaller numbers than in a finite-field setting of comparable cryptographic strength. Therefore, the use of elliptic polynomial cryptography is an improvement over finite-field based public-key cryptography.
In practice, an elliptic curve group over a finite field F is formed by choosing a pair of a and b coefficients, which are elements within F. The group consists of a finite set of points P(x,y) that satisfy the elliptic curve equation F(x,y)=y 2 −x 3 −ax−b=0, together with a point at infinity, O. The coordinates of the point, x and y, are elements of F represented in N-bit strings. In the following, a point is either written as a capital letter (e.g., point P) or as a pair in terms of the affine coordinates; i.e. (x,y).
The elliptic curve cryptosystem relies upon the difficulty of the elliptic curve discrete logarithm problem (ECDLP) to provide its effectiveness as a cryptosystem. Using multiplicative notation, the problem can be described as: given points B and Q in the group, find a number k such that B k =Q; where k is the discrete logarithm of Q to the base B. Using additive notation, the problem becomes: given two points B and Q in the group, find a number k such that kB=Q.
In an elliptic curve cryptosystem, the large integer k is kept private and is often referred to as the secret key. The point Q together with the base point B are made public and are referred to as the public key. The security of the system, thus, relies upon the difficulty of deriving the secret k, knowing the public points B and Q. The main factor that determines the security strength of such a system is the size of its underlying finite field. In a real cryptographic application, the underlying field is made so large that it is computationally infeasible to determine k in a straightforward way by computing all the multiples of B until Q is found.
At the heart of elliptic curve geometric arithmetic is scalar multiplication, which computes kB by adding together k copies of the point B. Scalar multiplication is performed through a combination of point-doubling and point-addition operations. The point-addition operations add two distinct points together and the point-doubling operations add two copies of a point together. To compute, for example, B=(2*(2*(2B)))+2B=Q, it would take three point-doublings and two point-additions.
Addition of two points on an elliptic curve is calculated as follows: when a straight line is drawn through the two points, the straight line intersects the elliptic curve at a third point. The point symmetric to this third intersecting point with respect to the x-axis is defined as a point resulting from the addition. Doubling a point on an elliptic curve is calculated as follows: when a tangent line is drawn at a point on an elliptic curve, the tangent line intersects the elliptic curve at another point. The point symmetric to this intersecting point with respect to the x-axis is defined as a point resulting from the doubling.
Table 1 illustrates the addition rules for adding two points (x 1 , y 1 ) and (x 2 , y 2 ); i.e., (x 3 , y 3 )=(x 1 , y 1 )+(x 2 , y 2 ):
TABLE 1
Summary of Addition Rules: (x 3 , y 3 ) = (x 1 , y 1 ) + (x 2 , y 2 )
General Equations
x 3 = m 2 − x 2 − x 1
y 3 = m (x 3 − x 1 ) + y 1
Point Addition
m
=
y
2
-
y
1
x
2
-
x
1
Point Doubling
m
=
3
x
1
2
-
a
2
y
1
(x 3 , y 3 ) = 2(x 1 ,y 1 )
(x 2 , y 2 ) = −(x 1 , y 1 )
(x 2 , y 2 ) = O
(x 3 , y 3 ) = (x 1 , y 1 ) + (−(x 2 , y 2 )) = O
(x 3 , y 3 ) = (x 1 , y 1 ) + O = (x 1 , y 1 )−(x 1 , y 1 ) = (x 1 , −y 1 )
For elliptic curve encryption and decryption, given a message point (x m , y m ), a base point (x B , y B ), and a given key, k, the cipher point (x C , y C ) is obtained using the equation (x C , y C )=(x m , y m )+k(x B , y B ).
There are two basics steps in the computation of the above equations. The first step is to find the scalar multiplication of the base point with the key, k(x B , y B ). The resulting point is then added to the message point, (x m , y m ) to obtain the cipher point. At the receiver, the message point is recovered from the cipher point, which is usually transmitted, along with the shared key and the base point (x m , y m )=(x C , y C )−k(x B , y B ).
As noted above, the x-coordinate, x m , is represented as an N-bit string. However, not all of the N-bits are used to carry information about the data of the secret message. Assuming that the number of bits of the x-coordinate, x m , that do not carry data is L, then the extra bits L are used to ensure that message data, when embedded into the x-coordinate, will lead to an x m value which satisfies the elliptic curve equation (1). Typically, if the first guess of x m is not on a curve, then the second or third try will be.
Thus, the number of bits used to carry the bits of the message data is (N−L). If the secret data is a K-bit string, then the number of elliptic curve points needed to encrypt the K-bit data is
⌈
K
N
-
L
⌉
.
It is important to note that the y-coordinate, y m , of the message point carries no data bits.
An attack method, referred to as power analysis exists, in which the secret information is decrypted on the basis of leaked information. An attack method in which change in voltage is measured in cryptographic processing using secret information, such as DES (Data Encryption Standard) or the like, such that the process of the cryptographic processing is obtained, and the secret information is inferred on the basis of the obtained process is known.
As one of the measures against power analysis attack on elliptic curve cryptosystems, a method using randomized projective coordinates is known. This is a measure against an attack method of observing whether a specific value appears or not in scalar multiplication calculations, and inferring a scalar value from the observed result. By multiplication with a random value, the appearance of such a specific value is prevented from being inferred.
In the above-described elliptic curve cryptosystem, attack by power analysis, such as DPA or the like, was not taken into consideration. Therefore, in order to relieve an attack by power analysis, extra calculation has to be carried out using secret information in order to weaken the dependence of the process of the cryptographic processing and the secret information on each other. Thus, time required for the cryptographic processing increases so that cryptographic processing efficiency is lowered.
With the development of information communication networks, cryptographic techniques have been indispensable elements for the concealment or authentication of electronic information. Efficiency in terms of computation time is a necessary consideration, along with the security of the cryptographic techniques. The elliptic curve discrete logarithm problem is so difficult that elliptic curve cryptosystems can make key lengths shorter than that in Rivest-Shamir-Adleman (RSA) cryptosystems, basing their security on the difficulty of factorization into prime factors. Thus, the elliptic curve cryptosystems offer comparatively high-speed cryptographic processing with optimal security. However, the processing speed is not always high enough to satisfy smart cards, for example, which have restricted throughput or servers that have to carry out large volumes of cryptographic processing.
The pair of equations for m in Table 1 are referred to as “slope equations”. Computation of a slope equation in finite fields requires one finite field division. Alternatively, the slope computation can be computed using one finite field inversion and one finite field multiplication. Finite field division and finite field inversion are costly in terms of computational time because they require extensive CPU cycles for the manipulation of two elements of a finite field with a large order. Presently, it is commonly accepted that a point-doubling and a point-addition operation each require one inversion, two multiplications, a square, and several additions. At present, there are techniques to compute finite field division and finite field inversion, and techniques to trade time-intensive inversions for multiplications through performance of the operations in projective coordinates.
In cases where field inversions are significantly more time intensive than multiplication, it is efficient to utilize projective coordinates. An elliptic curve projective point (X, Y, Z) in conventional projective (or homogeneous) coordinates satisfies the homogeneous Weierstrass equation: {tilde over (F)}(X, Y, Z)=Y 2 Z−aXZ 2 −bZ 3 =0, and, when Z≠0, it corresponds to the affine point
(
x
,
y
)
=
(
X
Z
,
Y
Z
)
.
Other projective representations lead to more efficient implementations of the group operation, such as, for example, the Jacobian representations, where the triplets (X, Y, Z) correspond to the affine coordinates
(
x
,
y
)
=
(
X
Z
2
,
Y
Z
3
)
whenever Z≠0. This is equivalent to using a Jacobian elliptic curve equation that is of the form {tilde over (F)} j (X, Y, Z)=Y 2 −X 3 −aXZ 4 −bZ 6 =0.
Another commonly used projection is the Chudnovsky-Jacobian coordinate projection. In general terms, the relationship between the affine coordinates and the projection coordinates can be written as
(
x
,
y
)
=
(
X
Z
i
,
Y
Z
j
)
where the values of i and j depend on the choice of the projective coordinates. For example, for homogeneous coordinates, i=1 and j=1.
The use of projective coordinates circumvents the need for division in the computation of each point addition and point doubling during the calculation of scalar multiplication. Thus, finite field division can be avoided in the calculation of scalar multiplication,
k
(
X
B
Z
B
i
,
Y
B
Z
B
j
)
,
when using projective coordinates.
The last addition for the computation of the cipher point,
(
X
C
Z
C
i
,
Y
C
Z
C
j
)
;
i.e., the addition of the two points
(
X
m
Z
m
i
,
Y
m
Z
m
j
)
and
k
(
X
B
Z
B
i
,
Y
B
Z
B
j
)
,
can also be carried out in the chosen projection coordinate:
(
X
C
Z
C
i
,
Y
C
Z
C
j
)
=
(
X
m
Z
m
i
,
Y
m
1
Z
m
j
)
+
(
X
B
Z
B
i
,
Y
B
Z
B
j
)
.
It should be noted that Z m =1.
However, one division (or one inversion and one multiplication) must still be carried out in order to calculate
x
C
=
X
C
Z
C
i
,
since only the affine x-coordinate of the cipher point, x C , is sent by the sender.
Thus, the encryption of (N−L) bits of the secret message using elliptic curve encryption requires at least one division when using projective coordinates. Similarly, the decryption of a single message encrypted using elliptic curve cryptography also requires at least one division when using projective coordinates.
Thus, elliptic polynomial cryptography with secret key embedding solving the aforementioned problems is desired.
SUMMARY OF THE INVENTION
The elliptic polynomial cryptography with secret key embedding includes a method that allows for the encryption of messages through elliptic polynomial cryptography and, particularly, with the embedding of secret keys in the message bit string. The method of performing elliptic polynomial cryptography is based on the elliptic polynomial discrete logarithm problem. It is well known that an elliptic polynomial discrete logarithm problem is a computationally “difficult” or “hard” problem.
The method includes the following steps: (a) defining a maximum block size that can be embedded into (nx+1) x-coordinates and ny y-coordinates, wherein n is an integer, and setting the maximum block size to be (nx+ny+1)N bits, wherein N is an integer; (b) a sending correspondent and a receiving correspondent agree upon the values of nx and ny, and further agree on a set of coefficients a 1k , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F, wherein F represents a finite field where the field's elements can be represented in N-bits, the sending and receiving correspondents further agreeing on a set of random numbers k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l , where l=1, . . . N p , the set of random numbers being shared secret keys for communication, wherein N p ≧1, the sending and receiving correspondents further agreeing on a base point on an elliptic polynomial (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ) ∈EC xn+ny+2 , wherein EC nx+ny+2 represents the elliptic polynomial.
The sending correspondent then performs the following steps: (c) establishing an integer index j such that a first block is defined by j=0, and embedding j th block of a secret message bit string into an elliptic curve message point (x 0,m,j , x 1,m,j , . . . , x nx,m,j , y 0,m,j , y 1,m,j , . . . , y ny,m,j ); (d) embedding the j th block shared secret key bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into a set of elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), where l=1, . . . N p , and wherein the set of elliptic curve key points satisfy the elliptic polynomial at the elliptic curve message point; (e) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) + k s,j,l ( xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (f) calculating a cipher point of the j th data block (x c,0,j , x c,1,j , . . . , x c,nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
p
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
y
r
y
,
1
,
j
,
l
,
…
,
y
r
y
,
ny
,
j
,
l
)
;
(g) sending appropriate bits of the x-coordinates and the y-coordinates of the cipher point (x c,0,j , x c,1,j , . . . , x c,nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) to the receiving correspondent; (h) generating a new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p for encryption of a following block of the message data bit string using at least one random number generators and the set of keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l for l=1, . . . N p ; and (i) returning to step (c) to encrypt the following block of the message data bit string.
The receiving correspondent then performs the following steps: (j) embedding the h th block of the shared secret message bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into the elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), wherein the set of elliptic curve key points satisfy the elliptic polynomial; (k) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,1 , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) + k s,j,l ( xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (l) calculating the message point of the j th data block (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
p
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
y
r
y
,
1
,
j
,
l
,
…
,
y
r
y
,
ny
,
j
,
l
)
;
(m) recovering the secret messages bit string for the j th block from the appropriate x-coordinates and y-coordinates of the message point (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ); (n) generating the new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p for the encryption of the following block of the message data bit string, using at least one random number generator and the keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l ; and (o) returning to step (j) to encrypt the following block of the message data bit string.
These and other features of the present invention will become readily apparent upon further review of the following specification and drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
The sole drawing Figure is a block diagram illustrating system components for elliptic polynomial cryptography with secret key embedding according to the present invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Elliptic polynomial cryptography with secret key embedding provides elliptic polynomial cryptographic methods based on the elliptic polynomial discrete logarithm problem. It is well known that an elliptic polynomial discrete logarithm problem is a computationally “difficult” or “hard” problem.
The secret key embedding methods to be described below use elliptic polynomials in their generation, where different elliptic polynomials are used for different blocks of the same plaintext. Particularly, the password protocols use an elliptic polynomial with more than one independent x-coordinate. More specifically, a set of elliptic polynomial points are used which satisfy an elliptic polynomial equation with more than one independent x-coordinate which is defined over a finite field F having the following properties: One of the variables (the y-coordinate) has a maximum degree of two, and appears on its own in only one of the monomials; the other variables (the x-coordinates) have a maximum degree of three, and each must appear in at least one of the monomials with a degree of three; and all monomials which contain x-coordinates must have a total degree of three.
The group of points of the elliptic polynomial with the above form is defined over additions in the extended dimensional space, and, as will be described in detail below, the method makes use of elliptic polynomials where different elliptic polynomials are used for different blocks of the same plaintext.
The particular advantage of using elliptic polynomial cryptography with more than one x-coordinate is that additional x-coordinates are used to embed extra message data bits in a single elliptic point that satisfies the elliptic polynomial equation. Given that nx additional x-coordinates are used, with nx being greater than or equal to one, a resulting elliptic point has (nx+1) x-coordinates, where all coordinates are elements of the finite field F. The number of points which satisfy an elliptic polynomial equation with nx additional x-coordinates defined over F and which can be used in the corresponding cryptosystem is increased by a factor of (#F) nx , where # denotes the size of a field.
Through the use of this particular method, security is increased through the usage of different elliptic polynomials for different message blocks during the secret key embedding. Further, each elliptic polynomial used for each message block is selected at random, preferably using an initial value and a random number generator.
Given the form of the elliptic polynomial equation described above, the elliptic polynomial and its twist are isomorphic with respect to one another. The method uses an embedding technique, to be described in greater detail below, which allows the embedding of a bit string into the x-coordinates of an elliptic polynomial point in a deterministic and non-iterative manner when the elliptic polynomial has the above described form. This embedding method overcomes the disadvantage of the time overhead of the iterative embedding methods used in existing elliptic polynomial.
The difficulty of using conventional elliptic polynomial cryptography for secret key embedding typically lies in the iterative and non-deterministic method needed to embed a bit string into an elliptic polynomial point, which has the drawback of the number of iterations needed being different for different bit strings which are being embedded. As a consequence, different calculation times are required for different blocks of bit strings. Such a data-dependant generation time is not suitable for secret key embedding methods, which require data independent encryption time. Further, with regard to iterative and non-deterministic methods in conventional elliptic polynomial cryptography, given an elliptic polynomial defined over a finite field that needs N-bits for the representation of its elements, only ((nx+ny+1)N−L) bits of the message data bits can be embedded in any elliptic polynomial point.
The isomorphic relationship between an elliptic polynomial and its twist, which is obtained as a result of the given form of the elliptic polynomial equation, ensures that any bit string whose equivalent binary value is an element of the underlying finite field has a bijective relationship between the bit string and a point which is either on the elliptic polynomial or its twist. This bijective relationship allows for the development of the secret key embedding to be described below.
In the conventional approach to elliptic polynomial cryptography, the security of the resulting cryptosystem relies on breaking the elliptic polynomial discrete logarithm problem, which can be summarized as: given the points k(x 0,B , x 1,B , . . . , x nx,B , y B ) and (x 0,B , x 1,B , . . . , x nx,B , y B ), find the scalar k.
Further, projective coordinates are used at the sending and receiving entities in order to eliminate inversion or division during each point addition and doubling operation of the scalar multiplication. It should be noted that all of the elliptic polynomial cryptography-based method with secret key embedding disclosed herein are scalable.
In the following, with regard to elliptic polynomials, the “degree” of a variable u i is simply the exponent i. A polynomial is defined as the sum of several terms, which are herein referred to as “monomials”, and the total degree of a monomial u i v j w k is given by (i+j+k). Further, in the following, the symbol ∈ denotes set membership.
One form of the subject elliptic polynomial equation with more than one x-coordinate and one or more y-coordinates is defined as follows: the elliptic polynomial is a polynomial with more than two independent variables such that the maximum total degree of any monomial in the polynomial is three; at least two or more of the variables, termed the x-coordinates, have a maximum degree of three, and each must appear in at least one of the monomials with a degree of three; and at least one or more variables, termed the y-coordinates, have a maximum degree of two, and each must appear in at least one of the monomials with a degree of two.
Letting S nx represent the set of numbers from 0 to nx (i.e., S nx ={0, . . . , nx}), and letting S ny represents the set of numbers from 0 to ny (i.e., S ny ={0, . . . , ny}), and further setting (nx+ny)≧1, then, given a finite field, F, the following equation defined over F is one example of the polynomial described above:
∑
k
∈
S
ny
a
1
k
y
k
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
y
l
+
∑
k
∈
S
ny
a
3
k
y
k
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
c
1
kli
y
k
y
l
x
i
+
∑
k
∈
S
ny
,
l
∈
S
nx
c
2
kl
y
k
x
l
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
c
3
kli
y
k
x
l
x
i
=
∑
l
∈
S
nx
b
1
l
x
l
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
2
x
k
+
∑
l
,
k
∈
S
nx
b
3
lk
x
l
x
k
+
∑
k
∈
S
nx
b
4
k
x
k
+
b
c
,
(
1
)
where a 1l , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F.
Two possible examples of equation (1) are y 0 2 =x 0 3 +x 1 3 +x 0 x 1 and y 0 2 +x 0 x 1 y 0 +y 0 =x 0 3 +x 1 3 +x 0 2 x 1 +x 0 x 1 2 +x 0 x 1 +x 0 +x 1 .
With regard to the use of the elliptic polynomial equation in the addition of points of an elliptic polynomial with more than one x-coordinate and one or more y-coordinates, we may examine specific coefficients a 1k , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F for F, wherein a set of points EC nx+ny+2 is defined as the (nx+ny+2)-tuple (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ), where x i , y k ∈ F, i ∈ S nx and k ∈ S ny . This set of points comprises solutions of F, although excluding the point (0, 0, . . . , 0) and the point at infinity, (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ).
The rules for conventional elliptic polynomial point addition may be adopted to define an additive binary operation, “+”, over EC nx+ny+2 , i.e., for all
(x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and
(x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 ,
the sum:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )+( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )
is also
(x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 ) ∈ EC nx+ny+2 .
As will be described in greater detail below, (EC nx+ny+2 , +) forms a pseudo-group (p-group) over addition that satisfies the following axioms:
(i) There exists a set (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 such that
( x 0 , x 1 , . . . , x nx ,y 0 , y 1 , . . . , y ny )+( x 0,I , x 1,I , . . . , x nx,I , y o,I , y 1,I , . . . , y ny,i ) =( x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny )
for all (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 ;
(ii) for every set (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 , there exists an inverse set, −(x 0 , x 1 , . . . , x nx , y 0 y 1 , . . . , y ny ) ∈ EC nx+ny+2 , such
( x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny )−( x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny )=that ( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I )
(iii) the additive binary operation in (EC nx+ny+2 , +) is commutative, and the p-group (EC nx+ny+2 , +) forms a group over addition when:
(iv) the additive binary operation in (EC nx+ny+2 , +) is associative.
Prior to a more detailed analysis of the above axioms, the concept of point equivalence must be further developed. Mappings can be used to indicate that an elliptic point represented using (nx+1) x-coordinates and (ny+1) y-coordinates, (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ), is equivalent to one or more elliptic points that satisfy the same elliptic polynomial equation, including the equivalence of an elliptic point to itself.
Points that are equivalent to one another can be substituted for each other at random, or according to certain rules during point addition and/or point doubling operations. For example, the addition of two points
(x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) and
(x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) is given by:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) +( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ).
If the point (x″ 0,1 , x″ 1,1 , . . . , x″ nx,1 , y″ 0,1 , y″ 1,1 , . . . , y″ ny,1 ) is equivalent to the point (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ), then the former can be substituted for (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) in the above equation in order to obtain:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x″ 0,1 , x″ 1,1 , . . . , x nx,1 , y″ 0,1 , y″ 1,1 , . . . , y″ ny,1 ) +( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ).
Mappings that are used to define equivalences can be based on certain properties that exist in elliptic polynomial equations, such as symmetry between variables. As an example, we consider the point (x 0 , x 1 , y 0 ) that satisfies the equation y 0 2 =x 0 3 +x 1 3 +x 0 x 1 . The equivalent of this point may be defined as (x 1 , x 0 , −y 0 ).
With regard to the addition rules for (EC nx+ny+2 , +), the addition operation of two points (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and (x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 , otherwise expressed as:
(x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )+( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )
is calculated in the following manner: First, a straight line is drawn which passes through the two points to be added. The straight line intersects EC nx+ny+2 at a third point, which we denote (x′ 0,3 , x′ 1,3 , . . . , x′ nx,3 , y′ 0,3 , y′ 1,3 , . . . , y′ ny,3 ) ∈ EC nx+ny+2 . The sum point is defined as
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=−( x′ 0,3 , x′ 1,3 , . . . , x′ nx,3 , y′ 0,3 , y′ 1,3 , . . . , y′ ny,3 ).
From the above definition of the addition rule, addition over EC nx+ny+2 is commutative, that is:
( x 0,1 , x 1,1 , . . . , x nx,1 , y 0.1 , y 1,1 , . . . , y ny,1 )+( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )=( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )+( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )
for all (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and for all (x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 . This community satisfies axiom (iii) above.
There are two primary cases that need to be considered for the computation of point addition for (EC nx+ny+2 , +): (A) for at least one j ∈ S nx , x j,1 ≠x j,2 ; and (B) for all j ∈ S nx , x j,1 =x j,2 =x j,o . Case B includes three sub-cases:
i. for all k ∈ S ny y k,1 =y k,2 that is:
( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )=( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )′
which corresponds to point doubling;
ii. for k ∈ S ny & k≠0, y k,1 =y k,2 , and where y 0,1 & y 0,2 are the roots of the following quadratic equation in y 0 :
a
10
y
0
2
+
∑
k
∈
S
ny
,
k
≠
0
a
1
k
y
k
,
1
2
+
y
0
{
∑
k
∈
S
ny
,
k
≠
0
a
2
k
0
y
k
,
1
+
∑
l
∈
S
ny
,
l
≠
0
a
20
l
y
l
,
1
}
+
∑
k
,
l
∈
S
ny
,
l
≠
k
,
l
&
k
≠
0
a
2
kl
y
k
,
1
y
l
,
1
+
a
30
y
0
+
∑
k
∈
S
ny
,
k
≠
0
a
3
k
y
k
,
1
+
y
0
2
∑
i
∈
S
nx
c
100
i
x
i
,
1
+
y
0
{
∑
k
∈
S
ny
,
i
∈
S
nx
c
1
k
0
i
y
k
,
1
x
i
,
1
+
∑
l
∈
S
ny
,
i
∈
S
nx
c
10
li
y
l
,
1
x
i
,
1
}
+
∑
k
,
l
∈
S
ny
,
l
&
k
≠
0
,
i
∈
S
nx
c
1
kli
y
k
,
1
y
l
,
1
x
i
,
1
+
y
0
∑
l
∈
S
nx
c
20
l
x
l
,
1
+
∑
k
∈
S
ny
,
k
≠
0
,
l
∈
S
nx
c
2
kl
y
k
,
1
x
l
,
1
+
y
0
∑
l
,
i
∈
S
nx
c
30
li
x
l
,
1
x
i
,
1
+
∑
k
∈
S
ny
,
k
≠
0
,
l
,
i
∈
S
nx
c
3
kli
y
k
,
1
x
l
,
1
x
i
,
1
=
∑
l
∈
S
nx
b
1
l
x
l
,
1
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
,
1
2
x
k
,
1
+
∑
l
,
k
∈
S
nx
b
3
lk
x
l
,
1
x
k
,
1
+
∑
k
∈
S
nx
b
4
k
x
k
,
1
+
b
c
,
which corresponds to point inverse; and
iii. all other conditions except those in Cases B.i and B.ii. This case occurs only when ny is greater than or equal to one.
For Case A, for at least one j ∈ S nx x j,1 ≠x j,2 , a straight line in (nx+ny+2)-dimensional space is defined by
y
k
-
y
k
,
1
y
k
,
2
-
y
k
,
1
=
x
j
-
x
j
,
1
x
j
,
2
-
x
j
,
1
,
k ∈ S ny and j ∈ S nx and
x
i
-
x
i
,
1
x
i
,
2
-
x
i
,
1
=
x
j
-
x
j
,
1
x
j
,
2
-
x
j
,
1
,
i≠j, i ∈ S nx .
For this case, y k =m yk x j +c yk , where
m
yk
=
y
k
,
2
-
y
k
,
1
x
j
,
2
-
x
j
,
1
and c yk =y k,1 −x j,1 m yk . Further, x i =m xi x j +c xi , where
m
xi
=
x
i
,
2
-
x
i
,
1
x
j
,
2
-
x
j
,
1
and c xi =x i,1 −x j,1 m xi . Equation (1) can then be re-written as
∑
k
∈
S
ny
a
1
k
y
k
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
y
l
+
∑
k
∈
S
ny
a
3
k
y
k
+
x
j
∑
k
,
l
∈
S
ny
c
1
klj
y
k
y
l
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
1
kli
y
k
y
l
x
i
+
x
j
∑
k
∈
S
ny
c
2
kj
y
k
+
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
c
2
kl
y
k
x
l
+
x
j
2
∑
k
∈
S
ny
c
3
kjj
y
k
+
x
j
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
c
3
klj
y
k
x
l
+
x
j
∑
k
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
3
kji
y
k
x
i
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
,
l
&
i
≠
j
c
3
kli
y
k
x
l
x
i
=
b
1
j
x
j
3
+
∑
l
∈
S
nx
,
l
≠
j
b
1
l
x
l
3
+
x
j
2
∑
k
∈
S
nx
,
k
≠
j
b
2
jk
x
k
+
x
j
∑
l
∈
S
nx
,
l
≠
j
b
2
lj
x
l
2
+
∑
l
,
k
∈
S
nx
,
l
,
k
≠
j
,
l
≠
k
b
2
lk
x
l
2
x
k
+
b
3
jj
x
j
2
+
x
j
∑
k
∈
S
nx
,
k
≠
j
b
3
jk
x
k
+
x
j
∑
l
∈
S
nx
,
l
≠
j
b
3
lj
x
l
+
∑
l
,
k
∈
S
nx
,
l
,
k
≠
j
b
3
lk
x
l
x
k
+
b
4
j
x
j
+
∑
k
∈
S
nx
,
k
≠
j
b
4
k
x
k
+
b
c
,
and substitution of the above into the rewritten equation (1) for y k , k ∈ S ny and x i , i ∈ S nx & i≠j, results in:
∑
k
∈
S
ny
a
1
k
(
m
yk
x
j
+
c
yk
)
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
(
m
yk
x
j
+
c
yk
)
(
m
yl
x
j
+
c
yl
)
+
∑
k
∈
S
ny
a
3
k
(
m
yk
x
j
+
c
yk
)
+
x
j
∑
k
,
l
∈
S
ny
c
1
klj
(
m
yk
x
j
+
c
yk
)
(
m
yl
x
j
+
c
yl
)
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
1
kli
(
m
yk
x
j
+
c
yk
)
(
m
yl
x
j
+
c
yl
)
(
m
xi
x
j
+
c
xi
)
+
x
j
∑
k
∈
S
ny
c
2
kj
(
m
yk
x
j
+
c
yk
)
+
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
2
2
kl
(
m
yk
x
j
+
c
yk
)
(
m
xl
x
j
+
c
xl
)
+
x
j
2
∑
k
∈
S
ny
c
3
kjj
(
m
yk
x
j
+
c
yk
)
+
x
j
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
c
3
klj
(
m
yk
x
j
+
c
yk
)
(
m
xl
x
j
+
c
xl
)
+
x
j
∑
k
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
3
kji
(
m
yk
x
j
+
c
yk
)
(
m
xi
x
j
+
c
xi
)
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
,
l
&
i
≠
j
c
3
kli
(
m
yk
x
j
+
c
yk
)
(
m
xl
x
j
+
c
xl
)
(
m
xi
x
j
+
c
xi
)
=
b
1
j
x
j
3
+
∑
l
∈
S
nx
,
l
≠
j
b
1
l
(
m
xl
x
j
+
c
l
)
3
+
x
j
2
∑
k
∈
S
nx
,
k
≠
j
b
2
jk
(
m
xk
x
j
+
c
xk
)
+
x
j
∑
l
∈
S
nx
,
l
≠
j
b
2
lj
(
m
xl
x
j
+
c
xl
)
2
+
∑
l
,
k
∈
S
nx
,
l
&
k
≠
j
,
l
≠
k
b
2
lk
(
m
xl
x
j
+
c
xl
)
2
(
m
xk
x
j
+
c
xk
)
+
b
3
jj
x
j
2
+
x
j
∑
k
∈
S
nx
,
k
≠
j
b
3
jk
(
m
xk
x
j
+
c
xk
)
+
x
j
∑
l
,
k
∈
S
nx
,
l
≠
j
b
3
lj
(
m
xl
x
j
+
c
xl
)
+
∑
l
,
k
∈
S
nx
,
l
&
k
≠
j
b
3
lk
(
m
xl
x
j
+
c
xl
)
(
m
xk
x
j
+
c
xk
)
+
b
4
j
x
j
+
∑
k
∈
S
nx
,
k
≠
j
a
6
k
(
m
xk
x
j
+
c
xk
)
+
b
c
Expanding the terms in the above equation leads to a cubic equation of the form x j , C 3 x j 3 +C 2 x j 2 +C 1 x j +C 0 =0, where C 3 , C 2 , C 1 & C 0 are obtained from the above equation.
Assuming C 3 ≠0, the above cubic equation in x j has three roots x j,1 x j,2 , & x j,3 and can be written as (x j −x j,1 )(x j −x j,2 )(x j −x j,3 )=0. Normalizing by the coefficient of x 3 and equating the coefficients of x 2 in the resulting equation with that in (x j −x j,1 )(x j −x j,2 )(x j −x j,3 )=0, one obtains a solution for x′ j,3 :
x
j
,
3
′
=
-
C
2
C
3
-
x
j
,
1
-
x
j
,
2
.
(
2
)
The values of y′ k,3 , k ∈ S ny , and x′ i,3 , i ∈ S nx & i≠j, may be similarly obtained from equations for x j =x′ j,3 .
For cases where C 3 =0, C 3 x j 3 +C 2 x j 2 +C 1 x j =C 0 =0 becomes a quadratic equation. Such quadratic equations may be used in the definition of point equivalences.
With regard to Case B for all j ∈ S nx , x j,1 =x j,2 , the three sub-cases are considered below. In all cases, x j,0 is defined as x j,0 =x j,1 =x j,2 , j ∈ S nx .
For Case B.i., all k ∈ S ny , y k,1 =y k,2 , which corresponds to point doubling. In this case,
( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )=( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , . . . , y ny,2 ).
Letting:
( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) =( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )
the sum is written as
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o )+( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o ) (3)
There are several ways of defining the addition in this case. Three possible rules are described below. Case B.i.1: Letting S nx,Lx denote a subset of S nx with Lx elements, i.e., S nx,Lx ⊂ S nx ; letting S ny,Ly denote a subset of S ny with Ly elements and which does not include the element 0; i.e. S ny,Ly ⊂ S ny and 0∉ S ny,Ly ; setting the value of Lx and Ly as at least one, then the straight line in this case can be defined as a tangent to the point (x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o ) defined in a sub-dimensional space with coordinates y n and x m with n ∈ S ny,Ly and M ∈ S nx,Lx .
In this case, the gradients m yn and m xm of the straight line to be used in equation (2) are essentially the first derivatives of y n and x m , n ∈ S ny,Ly and m ∈ S nx,Lx , for F with respect to x j , j ∈ S nx,Lx , i.e.,
m
yn
=
y
n
x
j
and
m
xn
=
x
m
x
j
.
Using these derivatives for the values of the gradients,
m
yn
=
y
n
x
j
,
where n ∈ S ny,Ly , and
m
yn
=
x
m
x
j
,
where m ∈ S nx,Lx , in equation (2) and noting that it is assumed that
x
m
x
j
=
0
,
for m ∈ (S nx −S nx,Lx ) and
y
n
x
j
=
0
,
for n ∈ (S ny −S ny,Lx ) then a solution for x′ j,3 may be obtained.
The values of y′ n,3 for n ∈ S ny and x′ m,3 , for m ∈ S nx & m≠j, can further be obtained for x j =x′ j,3 . The choice of the x m -coordinates, m ∈ S nx,Lx , and y n -coordinates, n ∈ S ny,Ly , which can be used to compute the tangent of the straight line in this case may be chosen at random or according to a pre-defined rule. Further, a different choice of the x m -coordinates, m ∈ S nx,Lx and y n -coordinates, n ∈ S ny,Ly , may be made when one needs to compute successive point doublings, such as that needed in scalar multiplication.
With regard to the next case, Case B.i.2, the second possible way of defining the addition of a point with itself is to apply a sequence of the point doublings according to the rule defined above in Case B.i.1, where the rule is applied with a different selection of the x-coordinate(s) and y-coordinates(s) in each step of this sequence.
In the third sub-case, Case B.i.3, a point is substituted with one of its equivalents. Letting (x 0,oe , x 1,oe , . . . , x nx,oe , y 0,oe , y 1,oe , . . . , y ny,oe ) represent the equivalent point of (x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o ) then equation (3) may be written as:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o )+( x 0,oe , x 1,oe , . . . , x nx,oe , y 0,oe , y 1,oe , . . . , y ny,oe ).
With regard to Case B.ii, for k ∈ S ny & k≠0, y k,1 =y k,2 , and where y 0,1 & y 0,2 are the roots of the quadratic equation in y 0 , this case corresponds to generation of the point inverse.
Letting y k,1 =y k,2 =y k,o for k ∈ S ny & k≠0, then any two points, such as the point (x 0,o , x 1,o , . . . , x nx,o , y 0,1 y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 and the point (x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 , are in the hyper-plane with x i =x i,o , i ∈ S nx and y k =y k,o , k ∈ S ny & k≠0 . Thus, any straight line joining these two points such that (x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,o , . . . , y ny,o )≠(x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,o , . . . , y ny,o ) is also in this hyper-plane.
Substituting the values of x 0,o , x 1,o , . . . , x nx,o , y 1,o , . . . , & y ny,o in an elliptic polynomial equation with multiple x-coordinates and multiple y-coordinates, a quadratic equation for y 0 is obtained, as given above. Thus, y 0 has only two solutions y 0,1 & y 0,2 .
Thus, a line joining points (x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 and (x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 does not intersect with EC nx+ny+2 at a third point.
A line that joins these two points is assumed to intersect with EC nx+ny+2 at infinity (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 . This point at infinity is used to define both the inverse of a point in EC nx+ny+2 and the identity point. According to the addition rule defined above, one can write:
( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny )+( x 0 , x 1 , . . . , x nx , y 0,2 , y 1 , . . . , y ny ) =( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) (4)
since the third point of intersection of such lines is assumed to be at infinity, (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 . Thus, this equation defines a unique inverse for any point (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 , namely:
−( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny )=( x 0 , x 1 , . . . , x nx , y 0,2 , y 1 , . . . , y ny ).
Thus, equation (4) can be written as:
( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny )−( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny ) =( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) (5).
Further, a line joining the point at infinity (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 and a point (x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 will intersect with EC nx+ny+2 at (x 0 , x 1 , . . . , x nx , y 0,2 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 . Thus, from the addition rule defined above,
( x 0 , x 1 , . . . , x nx , y 0 , y 1 , y 2 , . . . , y ny )+( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) =( x 0 , x 1 , x nx , y 0 , y 1 , . . . , y ny ) (6)
Equation (5) satisfies axiom (ii) while equation (6) satisfies axiom (i), defined above.
Case B.iii applies for all other conditions except those in cases B.i and B.ii. This case occurs only when ny is greater than or equal to one. Given two points (x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and (x 0,o , x 1,o , . . . , x nx,o y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 that do not satisfy the conditions of cases B.i and B.ii above, the sum point is written as
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y y 1,1 , . . . , y ny,1 )+( x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,2 , . . . , y ny,2 )
There are several possible rules to find the sum point in this case. Three possible methods are given below:
1) Using three point doublings and one point addition,
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=4( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 ); −2( x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,2 , . . . , y ny,2 )
2) using one point doublings and three point additions,
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=(2( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 ) +( x 0,o , x 1,o , . . . , x nx,o , x 0,2 , y 1,2 , . . . , y ny,2 ))−; and ( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 )
3) using point equivalence,
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 )+( x 0,oe , x 1,oe , . . . , x nx,oe , y 0,2e , y 1,2e , . . . , y ny,2e ),
where (x 0,oe , x 1,oe , . . . , x nx,oe , y 0,2e , . . . , y 1,2e , . . . , y ny,2e ) is assumed to be the equivalent point of (x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,2 , . . . , y ny,2 ).
It should be noted that the above methods for defining the sum point are not the only ones that can be defined and are provided for exemplary purposes only. The choice of method used to obtain the sum point in this case should depend on the computation complexity of point addition and point doubling.
With regard to associativity, one way of proving associativity of (EC nx+ny+2 , +) is as follows: Given particular elliptic polynomial equations (i.e., for particular coefficients a 1l , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k , b c ∈ F) defined over a finite F, if it can be shown by computation that any point Q ∈ EC nx+ny+2 (and any of its equivalent points) can be uniquely written as k Q P ∈ EC nx+ny+2 , where P is the generator point of (EC nx+ny+2 , +), then the corresponding EC nx+ny+2 groups based on such polynomials are associative. This is because any three points Q, R, S ∈ EC nx+ny+2 (or any of their equivalent points) can be written as k Q P, k R P, k S P ∈ EC nx+ny+2 , respectively, thus their sum (Q+R+S)=(k Q P+k R P+k S P)=(k Q +k R +k S )P can be carried out in any order.
The following elliptic polynomial equation with nx=1 and ny=0 is used to show an example of the equations in Case A used in point addition: y 0 2 =x 0 3 +x 1 3 +x 0 x 1 . Choosing x j =x 0 , and substituting y k =m yk x j +c yk from Case A above for y 0 , and the corresponding equation x i =m xi x j +c xi for x 1 , one obtains (m y0 x 0 +c y0 ) 2 =x 0 3 +(m xi x 0 +c x1 ) 3 +x 0 (m x1 x 0 +c x1 ).
Expanding this equation yields the equation
m y0 2 x 0 2 +2 m y0 c y0 x 0 +c y0 2 =x 0 3 +m x1 3 x 0 3 +3 m m1 2 c x1 x 0 2 +3 m m1 c x1 2 x 0 +c x1 3 +m x1 x 0 2 +c x1 x 0
or
(1 +m x1 3 ) x 0 3 +(3 m x1 2 c x1 +m x1 −m y0 2 ) x 0 2 +(3 m x1 c x1 2 +c x1 −2 m y0 c y0 ) x 0 +c x1 3 −c y0 2 =0.
From equation (2), the solution for x′ 0,3 in this case is obtained:
x
0
,
3
′
=
-
(
3
m
xl
2
c
x
1
+
m
x
1
-
m
y
0
2
)
(
1
+
m
x
1
3
)
-
x
j
,
1
-
x
j
,
2
.
Similarly, one can obtain the values of y 0,3 and x′ 1,3 for x 0 =x′ 0,3 .
It should be noted that when m x1 =−1, the coefficient of the cubic term in the above is zero; i.e. C 3 =0. In this case, the resulting quadratic equation can be used in the definition of point equivalences for the points that satisfy the elliptic polynomial equation.
Each of the equations for point addition and point doublings derived for cases A and B above require modular inversion or division. In cases where field inversions or divisions are significantly more expensive (in terms of computational time and energy) than multiplication, projective coordinates are used to remove the requirement for field inversion or division from these equations.
Several projective coordinates can be utilized. In the preferred embodiment, the Jacobean projective coordinate system is used. As an example, we examine:
x
i
=
X
i
V
2
for
i
∈
S
nx
;
and
;
and
(
7
)
y
k
=
Y
k
V
3
for
k
∈
S
ny
.
(
8
)
Using Jacobian projection yields:
∑
k
∈
S
ny
a
1
k
Y
k
2
V
6
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
Y
k
V
3
Y
l
V
3
+
∑
k
∈
S
ny
a
3
k
Y
k
V
3
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
c
1
kli
Y
k
V
3
Y
l
V
3
X
i
V
2
+
∑
k
∈
S
ny
,
l
∈
S
nx
c
2
kl
Y
k
V
3
X
l
V
2
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
c
3
kli
Y
k
V
3
X
l
V
2
X
i
V
2
=
∑
l
∈
S
nx
b
1
l
X
l
3
V
6
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
X
l
2
V
4
X
k
V
2
+
∑
l
,
k
∈
S
nx
b
3
lk
X
l
V
2
X
k
V
2
+
∑
k
∈
S
nx
b
4
k
X
k
V
2
+
b
c
(
9
)
which can be rewritten as:
∑
k
∈
S
ny
a
1
k
Y
k
2
V
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
Y
k
Y
l
V
2
+
∑
k
∈
S
ny
a
3
k
Y
k
v
5
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
c
1
kli
Y
k
Y
l
X
i
+
∑
k
∈
S
ny
,
l
∈
S
nx
c
2
kl
Y
k
X
l
V
3
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
c
3
kli
Y
k
X
l
X
i
V
=
∑
l
∈
S
nx
b
1
l
X
l
3
V
2
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
X
l
2
X
k
V
2
+
∑
l
,
k
∈
S
nx
b
3
lk
X
l
X
k
V
4
+
∑
k
∈
S
nx
b
4
k
X
k
V
6
+
b
c
V
8
.
(
10
)
In the following, the points (X 0 , X 1 , . . . , X nx , Y 0 , Y 1 , . . . , Y ny , V) are assumed to satisfy equation (10). When V≠0, the projected point (X 0 , X 1 , . . . , X nx , Y 0 , Y 1 , . . . , Y ny , V) corresponds to the point:
(
x
0
,
x
1
,
…
,
x
nx
,
y
0
,
y
1
,
…
,
y
ny
)
=
(
X
0
V
2
,
X
1
V
2
,
…
,
X
nx
V
2
,
Y
0
V
3
,
Y
1
V
3
,
…
,
Y
ny
V
3
)
,
which satisfies equation (1).
Using Jacobean projective coordinates, equation (10) can be written as:
(
X
0
,
3
V
3
2
,
X
1
,
3
V
3
2
,
…
,
X
nx
,
3
V
3
2
,
Y
0
,
3
V
3
3
,
Y
1
,
3
V
3
3
,
…
,
Y
ny
,
3
V
3
3
)
=
(
X
0
,
1
V
1
2
,
X
1
,
1
V
1
2
,
…
,
X
nx
,
1
V
1
2
,
Y
0
,
1
V
1
3
,
Y
1
,
1
V
1
3
,
…
,
Y
ny
,
1
V
1
3
)
+
(
X
0
,
2
V
2
2
,
X
1
,
2
V
2
2
,
…
,
X
nx
,
2
V
2
2
,
Y
0
,
2
V
2
3
,
Y
1
,
2
V
2
3
,
…
,
Y
ny
,
2
V
2
3
)
.
(
11
)
By using Jacobian projective coordinates in the equations of Cases A and B above, and by an appropriate choice of the value of V 3 , it can be shown that point doubling and point addition can be computed without the need for field inversion or division.
In order to examine the embedding method, the twist of an elliptic polynomial equation needs to be defined. Given an elliptic polynomial with (nx+1) x-coordinates and (ny+1) y-coordinates of the form described above:
y
0
2
+
∑
k
∈
S
ny
a
1
k
y
k
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
y
l
=
∑
l
∈
S
nx
b
1
l
x
l
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
2
x
k
,
(
12
)
where a 1l , a 2kl , b 1l , b 2lk ∈ F.
Given certain values for the x-coordinates x 0,o , x 1,o , . . . , x nx,o and y-coordinates y 1,o , . . . , y ny,o , respectively, that are elements of the finite field, F, these values are substituted into the elliptic polynomial equation (1) in order to obtain a quadratic equation in y 0 :
y
0
2
=
-
∑
k
∈
S
ny
a
1
k
y
k
,
o
2
-
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
,
o
y
l
,
o
+
∑
l
∈
S
nx
b
1
l
x
l
,
o
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
,
o
2
x
k
,
o
=
T
.
If a solution of the above quadratic equation (i.e., y 0 2 =T) is an element of the finite field F, the point (x 0,o , x 1,o , . . . , x nx,o , y 0 , y 1,o , . . . , y ny,o ) is said to satisfy the given elliptic polynomial equation. If a solution of the above quadratic equation is not an element of the finite field F, the point (x 0,o , x 1,o , . . . , x nx,o , y 0 , y 1,o , . . . , y ny,o ) is said to satisfy the twist of the given elliptic curve equation. The inventive embedding method is based on the isomorphic relationship between a curve and its twist as described in the following theorem:
An elliptic polynomial equation of the form given above is isomorphic to its twist if:
1) there are mathematical mappings that can be defined on the values of x 0 , x 1 , . . . , x nx , y 1 , . . . , y ny (i.e., φ x (x i ) and φ y (y i )) such that any point (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) that satisfies such an elliptic polynomial equation can be mapped into another point (φ x (x 0 ), φ x (x 1 ), . . . , φ x (x xn ), φ y (y 0 ), φ y (y 1 ), . . . , φ y (y ny )) that satisfies the twist of the same elliptic polynomial equation; and 2) the mapping between the points (x o , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) and (φ x (x 0 ), φ x (x 1 ), . . . , φ x (x xn ), φ y (y 0 ), φ y (y 1 ), . . . , φ y (y ny )), is unique, i.e., a one-to-one correspondence.
The proof of this theorem is as follows. Re-writing equation (12) as:
y
0
2
=
-
∑
k
∈
S
n
y
a
1
k
y
k
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
y
l
+
∑
l
∈
S
n
x
b
1
l
x
l
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
2
x
k
,
(
13
)
,
and letting the right-hand side of equation (13) be denoted as T, then:
T
=
-
∑
k
∈
S
n
y
a
1
k
y
k
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
y
l
+
∑
l
∈
S
n
x
b
1
l
x
l
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
2
x
k
.
(
14
)
Thus, any value of x 0 , x 1 , . . . , x nx , y 1 , . . . , y ny will lead to a value of T ∈ F(p). T could be quadratic residue or non-quadratic residue. If T is quadratic residue, then equation (14) is written as:
T
q
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
y
l
,
q
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
2
x
k
,
q
(
15
)
where x 0,q , x 1,q , . . . , x nx,q , y 1,q , . . . , y ny,q ∈ F denotes the values of x 0 , x 1 , . . . , x nx , y 1 , . . . , y ny that result in a quadratic residue value of T, which is hereafter denoted as T q .
If T is non-quadratic residue, then equation (14) is written as:
T
q
_
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
_
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
_
y
l
,
q
_
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
_
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
_
2
x
k
,
q
_
(
16
)
where x 0, q , x 1, q , . . . , x nx, q , y 1, q , . . . , y ny, q ∈ F denotes the values of x 0 , x 1 , . . . , y 1 , . . . , y ny that result in a non-quadratic residue value of T, denoted as T q .
Letting g be any non-quadratic residue number in F (i.e., g ∈ F(p) & √{square root over (g)}∉ F(p)), then multiplying equation (15) with g 3 yields:
g
3
T
q
=
-
g
3
∑
k
∈
S
n
y
a
1
k
y
k
,
q
2
-
g
3
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
y
l
,
q
+
g
3
∑
l
∈
S
n
x
b
1
l
x
l
,
q
3
+
g
3
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
2
x
k
,
q
,
which can be re-written as:
g
3
T
q
=
-
∑
k
∈
S
n
y
a
1
k
(
g
3
y
k
,
q
)
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
g
3
(
g
3
y
k
,
q
)
(
g
3
y
l
,
q
)
+
∑
l
∈
S
n
x
b
1
l
(
g
x
l
,
q
)
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
(
g
x
l
,
q
)
2
(
g
x
k
,
q
)
.
(
17
)
It should be noted that if g is non-quadratic residue, then g 3 is also non-quadratic residue. Further, the result of multiplying a quadratic residue number by a non-quadratic residue number is a non-quadratic residue number. Thus, g 3 T q is non-quadratic residue.
By comparing the terms of equations (16) and (17), we obtain the following mappings:
x i, q =gx i,q ; (18);
y i , q =√{square root over (g 3 )}y i,q ; (19); and
T q =g 3 T q (20).
The mappings between the variables x i,q and x i, q in equation (18), y i,q and y i, q in equation (19), and T q and T q in equation (20) are all bijective, i.e., there is a one-to-one correspondence from basic finite field arithmetic. As a consequence, the mappings between the (nx+ny+2)-tuple (x 0,q , x 1,q , . . . , x nx,q , T q , y 1,q , . . . , y ny,q ) and the (nx+ny+2)-tuple (x 0, q , x 1, q , . . . , x nx, q , T q , y 1, q , . . . , y ny, q ) are also bijective.
Therefore, for every solution of equation (15), there is an isomorphic solution that satisfies equation (16), and since the mappings of the coordinates of one to the other are given in equations (18)-(20), these two solutions are isomorphic with respect to each other.
Since T q is quadratic residue, this expression can be written as:
T q =y 0 2 . (21)
Thus, from equation (20), T q can be written as:
T q =g 3 y 0 2 (22).
Using equations (21) and (22), equations (15) and (16) can be written as:
y
0
2
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
y
l
,
q
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
2
x
k
,
q
;
(
23
)
and
g
3
y
0
2
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
_
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
_
y
l
,
q
_
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
_
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
_
2
x
k
,
q
_
.
(
24
)
Since any solution of equation (15) has an isomorphic solution that satisfies equation (16), it follows that the solution of equation (23), denoted as (x 0,q , x 1,q , . . . , x nx,q , y 0 , y 1,q , . . . , y ny,q ), has an isomorphic solution that satisfies equation (24), denoted as
(
g
x
0
,
q
,
g
x
1
,
q
,
…
,
g
x
n
x
,
q
,
g
3
2
y
0
,
g
3
2
y
1
,
q
,
…
,
g
3
2
y
n
y
,
q
)
.
The solutions of equation (23) form the points (s 0,q , x 1,q , . . . , x nx,q , y 0 , y 1,q , . . . , y ny,q ) that satisfy an elliptic polynomial. Similarly, the solutions of equation (24) form the points
(
g
x
0
,
q
,
g
x
1
,
q
,
…
,
g
x
n
x
,
q
,
g
3
2
y
0
,
g
3
2
y
1
,
q
,
…
,
g
3
2
y
n
y
,
q
)
that satisfy its twist. This proves the above theorem.
An example of a mapping of the solutions of equation (23) defined over F(p), where p=3 mod 4, to the solutions of its twist is implemented by using −x i for the x-coordinates and −y i 2 for the y-coordinates.
The isomorphism between an elliptic polynomial and its twist, discussed above, is exploited for the embedding of the bit sting of a shared secret key into the appropriate x and y coordinates of an elliptic polynomial point without the need for an iterative search for a quadratic residue value of the corresponding y 0 -coordinate, which usually requires several iterations, where the number of iterations needed is different for different bit strings which are being embedded.
Assuming F=F(p) and that the secret key is an M-bit string such that (nx+ny+1)N>M>N−1, where N is the number of bits needed to represent the elements of F(p), then the secret key bit string is divided into (nx+ny+1) bit-strings k x,0 , k x,1 , . . . , k x,nx , k y,1 , . . . , k k,ny . The value of the bit-strings k x,0 , k x,1 , . . . , k x,nx , k y,1 , . . . , k k,ny must be less than p. In the preferred embodiment of embedding the (nx+ny+1) bit-strings k x,0 , k x,1 , . . . , k x,nx , k y,1 , . . . , k k,ny , the embedding is as follows.
First, assign the value of the bit string of k x,0 , k x,1 , . . . , k x,nx to x 0,k , x 1,k , . . . , x nx,k . Next, assign the value of the bit string of k y,1 , . . . , k k,ny to y 1,k , . . . , y ny,k . Then, compute:
T
=
-
∑
i
∈
S
n
y
a
1
i
y
i
,
k
2
-
∑
i
,
l
∈
S
n
y
,
l
≠
i
a
2
i
l
y
i
,
k
y
l
,
k
+
∑
l
∈
S
n
x
b
1
l
x
l
,
k
3
+
∑
l
,
i
∈
S
n
x
,
l
≠
i
b
2
l
i
x
l
,
k
2
x
i
,
k
.
Finally, use the Legendre test to see if T has a square root. If T has a square root, assign one of the roots to y 0 ; otherwise, the x-coordinates and y-coordinates of the elliptic polynomial point with the embedded shared secret key bit string are given by gx i,k and
g
3
2
y
i
,
k
,
respectively, where g is non-quadratic residue in F.
It should be noted that p is usually predetermined prior to encryption, so that the value of g can also be predetermined. Further, the receiver can identify whether the point (x, 0,k , x 1,k , . . . , x nx,k , y 0,k , . . . , y ny,k ) or the point
(
g
x
0
,
k
,
g
x
1
,
k
,
…
,
g
x
n
x
,
k
,
g
3
2
y
0
,
k
,
g
3
2
y
1
,
k
,
…
,
g
3
2
y
n
y
,
k
)
is the elliptic polynomial point with the embedded secret key bit strings without any additional information. Additionally, any non-quadratic value in F(p) can be used for g. For efficiency, g is chosen to be −1 for p≡3 mod 4 and g is chosen to be 2 for p≡1 mod 4 .
The same deterministic and non-iterative method described above can be used to embed a secret message bit string into an elliptic polynomial point in a deterministic and non-iterative manner. Assuming F=F(p) and that the message is an M-bit string such that (nx+ny+1)N>M>N−1, where N is the number of bits needed to represent the elements of F(p), then the message bit string is divided into (nx+ny+1) bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 . . . , m k,ny . The value of the bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 , . . . , m k,ny must be less than p. As the previous embodiment, the embedding of the (nx+ny+1) bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 , . . . , m k,ny can be accomplished out as follows.
First, assign the value of the bit string of m x,0 , m x,1 , . . . , m x,nx to x 0,m , x 1,m , . . . , x nx,m . Next, assign the value of the bit string of m y,1 , . . . , m k,ny to y 1,m , . . . , y ny,m . Then compute:
T
=
-
∑
i
∈
S
n
y
a
1
i
y
i
,
m
2
-
∑
i
,
l
∈
S
n
y
,
l
≠
i
a
2
i
l
y
i
,
m
y
l
,
m
+
∑
l
∈
S
n
x
b
1
l
x
l
,
m
3
+
∑
l
,
i
∈
S
n
x
,
l
≠
i
b
2
l
i
x
l
,
m
2
x
i
,
m
.
Finally, use the Legendre test to see if T has a square root. If T has a square root, then assign one of the roots to y 0 , otherwise the x-coordinates and y-coordinates of the elliptic polynomial point with the embedded shared secret key bit string are given by gx i,m and
g
3
2
y
i
,
m
,
respectively.
It should be noted that p is usually predetermined prior to encryption; thus, the value of g can also be predetermined. Further, when using the above method, the strings m x,0 , m x,1 , . . . , m x,nx and m y,1 , . . . , m k,ny can be recovered directly from x 0,m , x 1,m , . . . , x nx,m and y 1,m , . . . , y ny,m , respectively. An extra bit is needed to identify whether (x 0,m , x 1,m , . . . , x nx,m , y 0,m , y 1,m , . . . , y 1,m , . . . , y ny,m ) or
(
g
x
0
,
m
,
g
x
1
,
m
,
…
,
g
x
n
x
,
m
,
g
3
2
y
0
,
m
,
g
3
2
y
1
,
m
,
…
,
g
3
2
y
n
y
,
m
)
is used at the sending correspondent. Additionally, any non-quadratic value in F(p) can be used for g. For efficiency, g is chosen to be −1 for p≡3 mod 4 and is chosen to be 2 for p≡1 mod 4. Further, at the receiver, the process is reversed. In the case of g=2, a division by two is carried out. It should noted that dividing x i,m by two is computed using one modulo addition, because:
x i,m /2=(( x i,m −( x i,m )mod 2)/2)+( x i,m )mod 2*(1/2)mod p; (i)
(x i,m )mod 2 is the least significant bit of x i,m ; (ii) and
(1/2)mod p =( p+ 1)/2. (iii)
In the following, it is assumed that F=F(p). In conventional elliptic polynomial cryptography, the cipher point is computed as:
( x 0,c , x 1,c , . . . , x nx,c , y 0,c , y 1,c , . . . , y ny,c )=( x 0,m , x 1,m , . . . , x nx,m , y 0,m , y 1,m , . . . , y ny,m )+(x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )
where (x 0,m , x 1,m , . . . , x nx,m , y 0,m , y 1,m , . . . , y ny,m ) is the message point and (x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk ) is referred to as “the encrypting point”, which is typically computed as the scalar multiplication of the shared secret key k and a base point (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ), which is known publicly, such that:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )= k ( x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B )
where all of the coordinates are elements of the underlying field F(p), and which are presented in N-bits.
In the present method, the underlying elliptic polynomial cryptography is based on the principle of blinding the base point by making the base point dependant on the shared secret key. In a first embodiment, the shared secret key is represented using nN-bits, where n is greater than or equal to (nx+ny+2). The nN bits of the shared secret key could be given as one block, or may alternatively be generated using a smaller bit-string and a random number generator.
When n=2(nx+ny+2), various techniques may be used to generate the encrypting point (x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk ). In the preferred embodiment, which utilizes one blinded base point, (nx+ny+1)N-bits of the shared secret key bit-string k1 are embedded into the x-coordinates and the y-coordinates of an elliptic polynomial point to obtain a blinded base point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ). The other (nx+ny+2)N-bits of the key are used as a scalar value k2, which is then multiplied by the blinded base point to obtain the encrypting point:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )= k 2( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ).
In an alternative embodiment, one blinded base point and one public base point are utilized. In this embodiment, (nx+ny+1)N-bits of the shared secret key bit-string k1 are embedded into the x-coordinates and the y-coordinates of an elliptic curve point to obtain a blinded base point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ). The other N-bits of the key are used as a scalar value k2 that is multiplied by the public base point (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ), with the encrypting point being obtained by:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )=( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+ k 2( x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B )
In another alternative embodiment, two blinded base points are utilized. In this embodiment, the two blinded points are generated from the 2(nx+ny+2)N-bits of the shared secret key bit-string. One blinded point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ) is generated from (nx+ny+1)N-bits out of the (nx+ny+2)N-bits of the shared secret key bit-string, while the other blinded base point (x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ) is obtained from a different (nx+ny+1)N-bits combination of the 2(nx+ny+2)N-bits of the shared secret key bit-string. Part of the key used to generate the point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ), k 1 , is used as a scalar and multiplied by the point (x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ) to obtain the scalar multiplication k1(x 0,k2 , x 1,k2 , . . . , x nx,k2 y 0,k2 , y 1,k2 , . . . , y ny,k2 ). The other part of the key used to generate the point (x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ), k 2 is used as a scalar and multiplied by the point (x 0,k1 , x 1,k1 , . . . , y 0,k1 , y 1,k1 , . . . , y ny,k1 ) to obtain the scalar multiplication k2(x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ). The encrypting point can be selected using a first scalar multiplication and addition:
( x 0,bk , x 1,bk , . . . , x nx,bk y 0,bk , y 1,bk , . . . , y ny,bk )= k 2( x 0,k1 , x 1,k1 , . . . , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+ k 1( x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 )
or, alternatively, a scalar multiplication and addition as:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )=( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+ k 1( x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 )
or, in a further alternative, a scalar multiplication and addition given by:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )= k 2( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+( x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ).
In a first embodiment, the elliptic polynomial encryption method includes the following steps (where it is assumed that the data bit stream is encrypted in blocks of M-bits where (nx+ny+1)N>M>N−1, where N is the number of bits needed to represent the elements of f(p): (a) defining a maximum block size that can be embedded into (nx+1) x-coordinates and ny y-coordinates, wherein n is an integer, and setting the maximum block size to be (nx+ny+1)N bits, wherein N is an integer; (b) a sending correspondent and a receiving correspondent agree upon the values of nx and ny, and further agree on a set of coefficients a 1k , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F, wherein F represents a finite field where the field's elements can be represented in N-bits, the sending and receiving correspondents further agreeing on a set of random numbers k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l , where l=1, . . . N p , the set of random numbers being shared secret keys for communication, wherein N p ≧1, the sending and receiving correspondents further agreeing on a base point on an elliptic polynomial (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ) ∈ EC xn+ny+2 , wherein EC nx+ny+2 represents the elliptic polynomial.
The sending correspondent then performs the following steps: (c) establishing an integer index /such that a first block is defined by j=0, and embedding a j th block of a secret message bit string into an elliptic curve message point (x 0,m,j , x 1,m,j , . . . , x nx,m,j , y 0,m,j , y 1,m,j , . . . , y ny,m,j ); (d) embedding the, j th block shared secret key bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into a set of elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), where l=1, . . . N p , and wherein the set of elliptic curve key points satisfy the elliptic polynomial at the elliptic curve message point; (e) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr ny,1,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l )+ k s,j,l (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (f) calculating a cipher point of the j th data block (x c,0,j , x c,1,j , . . . , x nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
P
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
yr
y
,
1
,
j
,
l
,
…
,
yr
y
,
ny
,
j
,
l
)
;
(g) sending appropriate bits of the x-coordinates and the y-coordinates of the cipher point ( c,0,j , x c,1,j , . . . , x c,nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) to the receiving correspondent, together with any other information needed to recover the message point without sacrificing security; (h) generating a new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p for encryption of a following block of the message data bit string using at least one random number generators and the set of keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l for l=1, . . . N p ; and (i) returning to step c) to encrypt the following block of the message data bit string.
The receiving correspondent then performs the following steps: (j) embedding the j th block of the shared secret message bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into the elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), wherein the set of elliptic curve key points satisfy the elliptic polynomial, which may be received from the sending correspondent; (k) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+k r,j,l (xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l )+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l )+ k s,j,l ( xk x,0,j,l , xr x,1,j,l , . . . , xk x,nx,h,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (l) calculating the message point of the j th data block (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
P
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
yr
y
,
1
,
j
,
l
,
…
,
yr
y
,
ny
,
j
,
l
)
together with any other information sent by the sending correspondent needed to recover the message point without sacrificing security; (m) recovering the secret messages bit string for the j th block from the appropriate x-coordinates and y-coordinates of the message point (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ), and any information received from the sending correspondent; (n) generating the new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p which are needed for the encryption of the next block of the message data bit string, using at least one random number generator and the keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l ; and (o) returning to step (j) to encrypt the following block of the message data bit string.
A special case of the above method is when k r,j,l =k s,j,l =1 for l=1, . . . , N P . In this case, no scalar multiplication is needed to calculate the points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P , which can be computed using one point addition:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) (xk x,0,j,l , xk x,1,j,l , . . . , xk x,na,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ).
An alternative embodiment of the method includes using a non-linear process in generating the points: (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P , with the keys k r,j,i =k s,j,l =1, and then using the points: (xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) and (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ) for l=1, . . . , N P . (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P , are computed as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr′ x,0,j−1,l , xr′ x,1,j−1,l , . . . , xr′ x,nx,j−1,l , yr′ y,0,j−1,l , yr′ y,1,j−1,l , . . . , yr′ y,ny,j−1,l )+ k s,j,l ( xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
where,
xr″ x,i,j,l =ƒ x ( xr x,i,j−1,l , yr z,i,j−1,l |l= 1, . . . , N P ) and
yr″ x,i,j,l =ƒ y ( xr x,i,j−1,l , yr z,i,j−1,l |l= 1, . . . , N P ),
and if yr″ y,0,j−1,l is quadratic residue, then:
( xr′ x,0,j−1,l , xr′ x,1,j−1,l , . . . , xr′ x,nx,j−1,l , yr′ y,0,j−1,l , yr′ y,1,j−1,l , . . . , yr′ y,ny,j−1,l )=( xr″ x,0,j−1,l , xr″ x,1,j−1,l , . . . , xr″ x,nx,j−1,l , yr″ y,0,j−1,l , yr″ y,1,j−1,l , . . . , yr″ y,ny,j−1,l )
otherwise:
(
xr
x
,
0
,
j
-
1
,
l
′
,
xr
x
,
1
,
j
-
1
,
l
′
,
…
,
xr
x
,
nx
,
j
-
1
,
l
′
,
yr
y
,
0
,
j
-
1
,
l
′
,
yr
y
,
1
,
j
-
1
,
l
′
,
…
,
yr
y
,
ny
,
j
-
1
,
l
′
)
=
(
gxr
x
,
0
,
j
-
1
,
l
″
,
gxr
x
,
1
,
j
-
1
,
l
″
,
…
,
gxr
x
,
nx
,
j
-
1
,
l
″
,
g
3
2
yr
y
,
0
,
j
-
1
,
l
″
,
g
3
2
yr
y
,
1
,
j
-
1
,
l
″
,
…
,
g
3
2
yr
y
,
ny
,
j
-
1
,
l
″
)
where ƒ x (.) and ƒ y (.) are any non-linear functions.
Both the transmitting correspondent and the receiving correspondent perform the same non-linear function in generating the points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P .
As noted above, the methods include data embedding. In order to embed a message bit string into a point (x, √{square root over (α)}y) which satisfies either an elliptic curve equation y 2 =x 3 ax+b or its twist, α y 2 =x 3 +ax+b, the message bit string is first divided into N-bit strings and the i th block is denoted as m i . Following this, the value of the bit string of m i is assigned to x m i , and the values of x m i are substituted and the value of t m i is computed using t m i =x m i 3 +ax m i +b.
If t m i is quadratic residue, then y m i =√{square root over (t m i )} and the point is given as (x m i , y m i ). However, if t m i is non-quadratic residue, then
y
m
i
=
t
m
i
α
_
and the point is given as (x m i , √{square root over ({tilde over (α)}y m i ). The message point is then denoted as (x m i , √{square root over (α m i )}y m i ), where the point is on the elliptic curve if α m i =1, and the point is on the twist if α m i = α .
Alternatively, data embedding may be performed as follows: (a) dividing the message bit string into (nx+ny+1) bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 , . . . , m k,ny ; (b) assigning the value of the bit string of m x,0 , m x,1 , . . . , m x,nx to x 0,m , x 1,m , . . . , x nx,m ; (c) assigning the value of the bit string of to m y,1 , . . . , m k,ny to y 1,m , . . . , y ny,m ; (d) computing
T
(
s
-
1
)
=
∑
l
∈
S
nx
b
1
l
x
l
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
2
x
k
;
and (e) performing a Legendre test to determine if T has a square root, wherein if T has a square root, then assigning the square root to y 0 , and if T does not have a square root, then the x-coordinates and y-coordinates of the elliptic polynomial point with the embedded shared secret key bit string are selected as gx i,m and
g
2
3
y
i
,
m
,
respectively, where g is non-quadratic residue in F.
The Legendre Symbol is used to test whether an element of F(p) has a square root or not, i.e., whether an element is quadratic residue or not. The Legendre Symbol and test are as follows. Given an element of a finite field F(p), such as d, the Legendre symbol is defined as (d/p). In order to test whether d is quadratic residue or not, the Legendre symbol, (d/p), is computed such that:
(
d
p
)
=
{
+
1
if
x
is
quadratic
residue
0
if
x
≡
0
mod
F
(
p
)
-
1
otherwise
.
In the above, the password protocols use the scalar multiplication k m (x Pu , y Pu ). It should be noted that, in order to find a collision means, that there are two message bits strings m and m′ such that their integer values k m and k m′ will lead to k m k(x B , y B )≡k m′ k(x B , y B ). This collision implies that integers can be found such that k m k−k m ,k=l*#EC, where #EC is the order of the group (EC,+). This is equivalent to solving the elliptic curve discrete logarithm problem. This also applies to finding a collision for the points on the twist of an elliptic curve, k m k(x TB , √{square root over ( α y TB )=k m′ k(x TB , √{square root over ( α y TB ).
Thus, security of the password protocols depends on the security of the underlying elliptic polynomial cryptography. The security of elliptic polynomial cryptosystems is assessed by both the effect on the solution of the elliptic curve discrete logarithmic problem (ECDLP) and power analysis attacks.
It is well known that the elliptic curve discrete logarithm problem (ECDLP) is apparently intractable for non-singular elliptic curves. The ECDLP problem can be stated as follows: given an elliptic curve defined over F that needs N-bits for the representation of its elements, an elliptic curve point (x P , y P ) ∈ EC, defined in affine coordinates, and a point (x Q , y Q ) 531 EC, defined in affine coordinates, determine the integer k,0≦k≦#F, such that (x Q , y Q )=k(x P , y P ), provided that such an integer exists. In the below, it is assumed that such an integer exists.
The most well known attack used against the ECDLP is the Pollard ρ-method, which has a complexity of O(√{square root over (πK)}/2), where K is the order of the underlying group, and the complexity is measured in terms of an elliptic curve point addition.
Since the underlying cryptographic problems used in the above block cipher chaining methods is the discrete logarithm problem, which is a known difficult mathematical problem, it is expected that the security of the above methods are more secure than prior art ciphers which are not based on such a mathematically difficult problem.
Projective coordinate can also be used by the sending correspondent and the receiving correspondent to embed extra message data bits in the projective coordinate wherein the addition of the corresponding elliptic points is defined in (nx+ny+3) dimensional space where there are (nx+1) x-coordinates, (ny+1) y-coordinates and one projective coordinate.
The equations for the addition rule can be obtained by using the elliptic polynomial equation with (nx+1) x-coordinates and (nx+1) y-coordinates in projective coordinates and substituting a straight line equation to obtain a cubic equation in terms of one of the x-coordinates. This cubic equation can be used to identify the third point of intersection between a straight line and the elliptic polynomial in (nx+ny+3) dimensions given two other intersection points. This third point of intersection is used to identify the sum of the given two points.
It should be understood that the calculations may be performed by any suitable computer system, such as that diagrammatically shown in the sole drawing Figure. Data is entered into system 100 via any suitable type of user interface 116 , and may be stored in memory 112 , which may be any suitable type of computer readable and programmable memory. Calculations are performed by processor 114 , which may be any suitable type of computer processor and may be displayed to the user on display 118 , which may be any suitable type of computer display.
Processor 114 may be associated with, or incorporated into, any suitable type of computing device, for example, a personal computer or a programmable logic controller. The display 118 , the processor 114 , the memory 112 and any associated computer readable recording media are in communication with one another by any suitable type of data bus, as is well known in the art.
Examples of computer-readable recording media include a magnetic recording apparatus, an optical disk, a magneto-optical disk, and/or a semiconductor memory (for example, RAM, ROM, etc.). Examples of magnetic recording apparatus that may be used in addition to memory 112 , or in place of memory 112 , include a hard disk device (HDD), a flexible disk (FD), and a magnetic tape (MT). Examples of the optical disk include a DVD (Digital Versatile Disc), a DVD-RAM, a CD-ROM (Compact Disc-Read Only Memory), and a CD-R (Recordable)/RW.
It is to be understood that the present invention is not limited to the embodiments described above, but encompasses any and all embodiments within the scope of the following claims.
1.PublishNumber: US-2011200187-A1
2.Date Publish: 20110818
3.Inventor: GHOUTI LAHOUARI
IBRAHIM MOHAMMAD K.
GUTUB ADNAN A.
4.Inventor Harmonized: GHOUTI LAHOUARI(SA)
IBRAHIM MOHAMMAD K(GB)
GUTUB ADNAN A(SA)
5.Country: US
6.Claims:
(en)Elliptic polynomial cryptography with secret key embedding is a method that allows for the encryption of messages through elliptic polynomial cryptography and, particularly, with the embedding of secret keys in the message bit string. The method of performing elliptic polynomial cryptography is based on the elliptic polynomial discrete logarithm problem. It is well known that an elliptic polynomial discrete logarithm problem is a computationally “difficult” or “hard” problem.
7.Description:
(en)BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to computerized cryptographic systems and methods for encrypting communications in a computer network or electronic communications system, and particularly to a method of performing elliptic polynomial cryptography with secret key embedding.
2. Description of the Related Art
In recent years, the Internet community has experienced explosive and exponential growth. Given the vast and increasing magnitude of this community, both in terms of the number of individual users and web sites, and the sharply reduced costs associated with electronically communicating information, such as e-mail messages and electronic files, between one user and another, as well as between any individual client computer and a web server, electronic communication, rather than more traditional postal mail, is rapidly becoming a medium of choice for communicating information. The Internet, however, is a publicly accessible network, and is thus not secure. The Internet has been, and increasingly continues to be, a target of a wide variety of attacks from various individuals and organizations intent on eavesdropping, intercepting and/or otherwise compromising or even corrupting message traffic flowing on the Internet, or further illicitly penetrating sites connected to the Internet.
Encryption by itself provides no guarantee that an enciphered message cannot or has not been compromised during transmission or storage by a third party. Encryption does not assure integrity due to the fact that an encrypted message could be intercepted and changed, even though it may be, in any instance, practically impossible, to cryptanalyze. In this regard, the third party could intercept, or otherwise improperly access, a ciphertext message, then substitute a predefined illicit ciphertext block(s), which that party, or someone else acting in concert with that party, has specifically devised for a corresponding block(s) in the message. The intruding party could thereafter transmit the resulting message with the substituted ciphertext block(s) to the destination, all without the knowledge of the eventual recipient of the message.
The field of detecting altered communication is not confined to Internet messages. With the burgeoning use of stand-alone personal computers, individuals or businesses often store confidential information within the computer, with a desire to safeguard that information from illicit access and alteration by third parties. Password controlled access, which is commonly used to restrict access to a given computer and/or a specific file stored thereon, provides a certain, but rather rudimentary, form of file protection. Once password protection is circumvented, a third party can access a stored file and then change it, with the owner of the file then being completely oblivious to any such change.
Methods of adapting discrete-logarithm based algorithms to the setting of elliptic polynomials are known. However, finding discrete logarithms in this kind of group is particularly difficult. Thus, elliptic polynomial-based crypto algorithms can be implemented using much smaller numbers than in a finite-field setting of comparable cryptographic strength. Therefore, the use of elliptic polynomial cryptography is an improvement over finite-field based public-key cryptography.
In practice, an elliptic curve group over a finite field F is formed by choosing a pair of a and b coefficients, which are elements within F. The group consists of a finite set of points P(x,y) that satisfy the elliptic curve equation F(x,y)=y 2 −x 3 −ax−b=0, together with a point at infinity, O. The coordinates of the point, x and y, are elements of F represented in N-bit strings. In the following, a point is either written as a capital letter (e.g., point P) or as a pair in terms of the affine coordinates; i.e. (x,y).
The elliptic curve cryptosystem relies upon the difficulty of the elliptic curve discrete logarithm problem (ECDLP) to provide its effectiveness as a cryptosystem. Using multiplicative notation, the problem can be described as: given points B and Q in the group, find a number k such that B k =Q; where k is the discrete logarithm of Q to the base B. Using additive notation, the problem becomes: given two points B and Q in the group, find a number k such that kB=Q.
In an elliptic curve cryptosystem, the large integer k is kept private and is often referred to as the secret key. The point Q together with the base point B are made public and are referred to as the public key. The security of the system, thus, relies upon the difficulty of deriving the secret k, knowing the public points B and Q. The main factor that determines the security strength of such a system is the size of its underlying finite field. In a real cryptographic application, the underlying field is made so large that it is computationally infeasible to determine k in a straightforward way by computing all the multiples of B until Q is found.
At the heart of elliptic curve geometric arithmetic is scalar multiplication, which computes kB by adding together k copies of the point B. Scalar multiplication is performed through a combination of point-doubling and point-addition operations. The point-addition operations add two distinct points together and the point-doubling operations add two copies of a point together. To compute, for example, B=(2*(2*(2B)))+2B=Q, it would take three point-doublings and two point-additions.
Addition of two points on an elliptic curve is calculated as follows: when a straight line is drawn through the two points, the straight line intersects the elliptic curve at a third point. The point symmetric to this third intersecting point with respect to the x-axis is defined as a point resulting from the addition. Doubling a point on an elliptic curve is calculated as follows: when a tangent line is drawn at a point on an elliptic curve, the tangent line intersects the elliptic curve at another point. The point symmetric to this intersecting point with respect to the x-axis is defined as a point resulting from the doubling.
Table 1 illustrates the addition rules for adding two points (x 1 , y 1 ) and (x 2 , y 2 ); i.e., (x 3 , y 3 )=(x 1 , y 1 )+(x 2 , y 2 ):
TABLE 1
Summary of Addition Rules: (x 3 , y 3 ) = (x 1 , y 1 ) + (x 2 , y 2 )
General Equations
x 3 = m 2 − x 2 − x 1
y 3 = m (x 3 − x 1 ) + y 1
Point Addition
m
=
y
2
-
y
1
x
2
-
x
1
Point Doubling
m
=
3
x
1
2
-
a
2
y
1
(x 3 , y 3 ) = 2(x 1 ,y 1 )
(x 2 , y 2 ) = −(x 1 , y 1 )
(x 2 , y 2 ) = O
(x 3 , y 3 ) = (x 1 , y 1 ) + (−(x 2 , y 2 )) = O
(x 3 , y 3 ) = (x 1 , y 1 ) + O = (x 1 , y 1 )−(x 1 , y 1 ) = (x 1 , −y 1 )
For elliptic curve encryption and decryption, given a message point (x m , y m ), a base point (x B , y B ), and a given key, k, the cipher point (x C , y C ) is obtained using the equation (x C , y C )=(x m , y m )+k(x B , y B ).
There are two basics steps in the computation of the above equations. The first step is to find the scalar multiplication of the base point with the key, k(x B , y B ). The resulting point is then added to the message point, (x m , y m ) to obtain the cipher point. At the receiver, the message point is recovered from the cipher point, which is usually transmitted, along with the shared key and the base point (x m , y m )=(x C , y C )−k(x B , y B ).
As noted above, the x-coordinate, x m , is represented as an N-bit string. However, not all of the N-bits are used to carry information about the data of the secret message. Assuming that the number of bits of the x-coordinate, x m , that do not carry data is L, then the extra bits L are used to ensure that message data, when embedded into the x-coordinate, will lead to an x m value which satisfies the elliptic curve equation (1). Typically, if the first guess of x m is not on a curve, then the second or third try will be.
Thus, the number of bits used to carry the bits of the message data is (N−L). If the secret data is a K-bit string, then the number of elliptic curve points needed to encrypt the K-bit data is
⌈
K
N
-
L
⌉
.
It is important to note that the y-coordinate, y m , of the message point carries no data bits.
An attack method, referred to as power analysis exists, in which the secret information is decrypted on the basis of leaked information. An attack method in which change in voltage is measured in cryptographic processing using secret information, such as DES (Data Encryption Standard) or the like, such that the process of the cryptographic processing is obtained, and the secret information is inferred on the basis of the obtained process is known.
As one of the measures against power analysis attack on elliptic curve cryptosystems, a method using randomized projective coordinates is known. This is a measure against an attack method of observing whether a specific value appears or not in scalar multiplication calculations, and inferring a scalar value from the observed result. By multiplication with a random value, the appearance of such a specific value is prevented from being inferred.
In the above-described elliptic curve cryptosystem, attack by power analysis, such as DPA or the like, was not taken into consideration. Therefore, in order to relieve an attack by power analysis, extra calculation has to be carried out using secret information in order to weaken the dependence of the process of the cryptographic processing and the secret information on each other. Thus, time required for the cryptographic processing increases so that cryptographic processing efficiency is lowered.
With the development of information communication networks, cryptographic techniques have been indispensable elements for the concealment or authentication of electronic information. Efficiency in terms of computation time is a necessary consideration, along with the security of the cryptographic techniques. The elliptic curve discrete logarithm problem is so difficult that elliptic curve cryptosystems can make key lengths shorter than that in Rivest-Shamir-Adleman (RSA) cryptosystems, basing their security on the difficulty of factorization into prime factors. Thus, the elliptic curve cryptosystems offer comparatively high-speed cryptographic processing with optimal security. However, the processing speed is not always high enough to satisfy smart cards, for example, which have restricted throughput or servers that have to carry out large volumes of cryptographic processing.
The pair of equations for m in Table 1 are referred to as “slope equations”. Computation of a slope equation in finite fields requires one finite field division. Alternatively, the slope computation can be computed using one finite field inversion and one finite field multiplication. Finite field division and finite field inversion are costly in terms of computational time because they require extensive CPU cycles for the manipulation of two elements of a finite field with a large order. Presently, it is commonly accepted that a point-doubling and a point-addition operation each require one inversion, two multiplications, a square, and several additions. At present, there are techniques to compute finite field division and finite field inversion, and techniques to trade time-intensive inversions for multiplications through performance of the operations in projective coordinates.
In cases where field inversions are significantly more time intensive than multiplication, it is efficient to utilize projective coordinates. An elliptic curve projective point (X, Y, Z) in conventional projective (or homogeneous) coordinates satisfies the homogeneous Weierstrass equation: {tilde over (F)}(X, Y, Z)=Y 2 Z−aXZ 2 −bZ 3 =0, and, when Z≠0, it corresponds to the affine point
(
x
,
y
)
=
(
X
Z
,
Y
Z
)
.
Other projective representations lead to more efficient implementations of the group operation, such as, for example, the Jacobian representations, where the triplets (X, Y, Z) correspond to the affine coordinates
(
x
,
y
)
=
(
X
Z
2
,
Y
Z
3
)
whenever Z≠0. This is equivalent to using a Jacobian elliptic curve equation that is of the form {tilde over (F)} j (X, Y, Z)=Y 2 −X 3 −aXZ 4 −bZ 6 =0.
Another commonly used projection is the Chudnovsky-Jacobian coordinate projection. In general terms, the relationship between the affine coordinates and the projection coordinates can be written as
(
x
,
y
)
=
(
X
Z
i
,
Y
Z
j
)
where the values of i and j depend on the choice of the projective coordinates. For example, for homogeneous coordinates, i=1 and j=1.
The use of projective coordinates circumvents the need for division in the computation of each point addition and point doubling during the calculation of scalar multiplication. Thus, finite field division can be avoided in the calculation of scalar multiplication,
k
(
X
B
Z
B
i
,
Y
B
Z
B
j
)
,
when using projective coordinates.
The last addition for the computation of the cipher point,
(
X
C
Z
C
i
,
Y
C
Z
C
j
)
;
i.e., the addition of the two points
(
X
m
Z
m
i
,
Y
m
Z
m
j
)
and
k
(
X
B
Z
B
i
,
Y
B
Z
B
j
)
,
can also be carried out in the chosen projection coordinate:
(
X
C
Z
C
i
,
Y
C
Z
C
j
)
=
(
X
m
Z
m
i
,
Y
m
1
Z
m
j
)
+
(
X
B
Z
B
i
,
Y
B
Z
B
j
)
.
It should be noted that Z m =1.
However, one division (or one inversion and one multiplication) must still be carried out in order to calculate
x
C
=
X
C
Z
C
i
,
since only the affine x-coordinate of the cipher point, x C , is sent by the sender.
Thus, the encryption of (N−L) bits of the secret message using elliptic curve encryption requires at least one division when using projective coordinates. Similarly, the decryption of a single message encrypted using elliptic curve cryptography also requires at least one division when using projective coordinates.
Thus, elliptic polynomial cryptography with secret key embedding solving the aforementioned problems is desired.
SUMMARY OF THE INVENTION
The elliptic polynomial cryptography with secret key embedding includes a method that allows for the encryption of messages through elliptic polynomial cryptography and, particularly, with the embedding of secret keys in the message bit string. The method of performing elliptic polynomial cryptography is based on the elliptic polynomial discrete logarithm problem. It is well known that an elliptic polynomial discrete logarithm problem is a computationally “difficult” or “hard” problem.
The method includes the following steps: (a) defining a maximum block size that can be embedded into (nx+1) x-coordinates and ny y-coordinates, wherein n is an integer, and setting the maximum block size to be (nx+ny+1)N bits, wherein N is an integer; (b) a sending correspondent and a receiving correspondent agree upon the values of nx and ny, and further agree on a set of coefficients a 1k , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F, wherein F represents a finite field where the field's elements can be represented in N-bits, the sending and receiving correspondents further agreeing on a set of random numbers k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l , where l=1, . . . N p , the set of random numbers being shared secret keys for communication, wherein N p ≧1, the sending and receiving correspondents further agreeing on a base point on an elliptic polynomial (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ) ∈EC xn+ny+2 , wherein EC nx+ny+2 represents the elliptic polynomial.
The sending correspondent then performs the following steps: (c) establishing an integer index j such that a first block is defined by j=0, and embedding j th block of a secret message bit string into an elliptic curve message point (x 0,m,j , x 1,m,j , . . . , x nx,m,j , y 0,m,j , y 1,m,j , . . . , y ny,m,j ); (d) embedding the j th block shared secret key bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into a set of elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), where l=1, . . . N p , and wherein the set of elliptic curve key points satisfy the elliptic polynomial at the elliptic curve message point; (e) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) + k s,j,l ( xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (f) calculating a cipher point of the j th data block (x c,0,j , x c,1,j , . . . , x c,nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
p
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
y
r
y
,
1
,
j
,
l
,
…
,
y
r
y
,
ny
,
j
,
l
)
;
(g) sending appropriate bits of the x-coordinates and the y-coordinates of the cipher point (x c,0,j , x c,1,j , . . . , x c,nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) to the receiving correspondent; (h) generating a new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p for encryption of a following block of the message data bit string using at least one random number generators and the set of keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l for l=1, . . . N p ; and (i) returning to step (c) to encrypt the following block of the message data bit string.
The receiving correspondent then performs the following steps: (j) embedding the h th block of the shared secret message bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into the elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), wherein the set of elliptic curve key points satisfy the elliptic polynomial; (k) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,1 , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) + k s,j,l ( xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (l) calculating the message point of the j th data block (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
p
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
y
r
y
,
1
,
j
,
l
,
…
,
y
r
y
,
ny
,
j
,
l
)
;
(m) recovering the secret messages bit string for the j th block from the appropriate x-coordinates and y-coordinates of the message point (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ); (n) generating the new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p for the encryption of the following block of the message data bit string, using at least one random number generator and the keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l ; and (o) returning to step (j) to encrypt the following block of the message data bit string.
These and other features of the present invention will become readily apparent upon further review of the following specification and drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
The sole drawing Figure is a block diagram illustrating system components for elliptic polynomial cryptography with secret key embedding according to the present invention.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Elliptic polynomial cryptography with secret key embedding provides elliptic polynomial cryptographic methods based on the elliptic polynomial discrete logarithm problem. It is well known that an elliptic polynomial discrete logarithm problem is a computationally “difficult” or “hard” problem.
The secret key embedding methods to be described below use elliptic polynomials in their generation, where different elliptic polynomials are used for different blocks of the same plaintext. Particularly, the password protocols use an elliptic polynomial with more than one independent x-coordinate. More specifically, a set of elliptic polynomial points are used which satisfy an elliptic polynomial equation with more than one independent x-coordinate which is defined over a finite field F having the following properties: One of the variables (the y-coordinate) has a maximum degree of two, and appears on its own in only one of the monomials; the other variables (the x-coordinates) have a maximum degree of three, and each must appear in at least one of the monomials with a degree of three; and all monomials which contain x-coordinates must have a total degree of three.
The group of points of the elliptic polynomial with the above form is defined over additions in the extended dimensional space, and, as will be described in detail below, the method makes use of elliptic polynomials where different elliptic polynomials are used for different blocks of the same plaintext.
The particular advantage of using elliptic polynomial cryptography with more than one x-coordinate is that additional x-coordinates are used to embed extra message data bits in a single elliptic point that satisfies the elliptic polynomial equation. Given that nx additional x-coordinates are used, with nx being greater than or equal to one, a resulting elliptic point has (nx+1) x-coordinates, where all coordinates are elements of the finite field F. The number of points which satisfy an elliptic polynomial equation with nx additional x-coordinates defined over F and which can be used in the corresponding cryptosystem is increased by a factor of (#F) nx , where # denotes the size of a field.
Through the use of this particular method, security is increased through the usage of different elliptic polynomials for different message blocks during the secret key embedding. Further, each elliptic polynomial used for each message block is selected at random, preferably using an initial value and a random number generator.
Given the form of the elliptic polynomial equation described above, the elliptic polynomial and its twist are isomorphic with respect to one another. The method uses an embedding technique, to be described in greater detail below, which allows the embedding of a bit string into the x-coordinates of an elliptic polynomial point in a deterministic and non-iterative manner when the elliptic polynomial has the above described form. This embedding method overcomes the disadvantage of the time overhead of the iterative embedding methods used in existing elliptic polynomial.
The difficulty of using conventional elliptic polynomial cryptography for secret key embedding typically lies in the iterative and non-deterministic method needed to embed a bit string into an elliptic polynomial point, which has the drawback of the number of iterations needed being different for different bit strings which are being embedded. As a consequence, different calculation times are required for different blocks of bit strings. Such a data-dependant generation time is not suitable for secret key embedding methods, which require data independent encryption time. Further, with regard to iterative and non-deterministic methods in conventional elliptic polynomial cryptography, given an elliptic polynomial defined over a finite field that needs N-bits for the representation of its elements, only ((nx+ny+1)N−L) bits of the message data bits can be embedded in any elliptic polynomial point.
The isomorphic relationship between an elliptic polynomial and its twist, which is obtained as a result of the given form of the elliptic polynomial equation, ensures that any bit string whose equivalent binary value is an element of the underlying finite field has a bijective relationship between the bit string and a point which is either on the elliptic polynomial or its twist. This bijective relationship allows for the development of the secret key embedding to be described below.
In the conventional approach to elliptic polynomial cryptography, the security of the resulting cryptosystem relies on breaking the elliptic polynomial discrete logarithm problem, which can be summarized as: given the points k(x 0,B , x 1,B , . . . , x nx,B , y B ) and (x 0,B , x 1,B , . . . , x nx,B , y B ), find the scalar k.
Further, projective coordinates are used at the sending and receiving entities in order to eliminate inversion or division during each point addition and doubling operation of the scalar multiplication. It should be noted that all of the elliptic polynomial cryptography-based method with secret key embedding disclosed herein are scalable.
In the following, with regard to elliptic polynomials, the “degree” of a variable u i is simply the exponent i. A polynomial is defined as the sum of several terms, which are herein referred to as “monomials”, and the total degree of a monomial u i v j w k is given by (i+j+k). Further, in the following, the symbol ∈ denotes set membership.
One form of the subject elliptic polynomial equation with more than one x-coordinate and one or more y-coordinates is defined as follows: the elliptic polynomial is a polynomial with more than two independent variables such that the maximum total degree of any monomial in the polynomial is three; at least two or more of the variables, termed the x-coordinates, have a maximum degree of three, and each must appear in at least one of the monomials with a degree of three; and at least one or more variables, termed the y-coordinates, have a maximum degree of two, and each must appear in at least one of the monomials with a degree of two.
Letting S nx represent the set of numbers from 0 to nx (i.e., S nx ={0, . . . , nx}), and letting S ny represents the set of numbers from 0 to ny (i.e., S ny ={0, . . . , ny}), and further setting (nx+ny)≧1, then, given a finite field, F, the following equation defined over F is one example of the polynomial described above:
∑
k
∈
S
ny
a
1
k
y
k
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
y
l
+
∑
k
∈
S
ny
a
3
k
y
k
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
c
1
kli
y
k
y
l
x
i
+
∑
k
∈
S
ny
,
l
∈
S
nx
c
2
kl
y
k
x
l
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
c
3
kli
y
k
x
l
x
i
=
∑
l
∈
S
nx
b
1
l
x
l
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
2
x
k
+
∑
l
,
k
∈
S
nx
b
3
lk
x
l
x
k
+
∑
k
∈
S
nx
b
4
k
x
k
+
b
c
,
(
1
)
where a 1l , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F.
Two possible examples of equation (1) are y 0 2 =x 0 3 +x 1 3 +x 0 x 1 and y 0 2 +x 0 x 1 y 0 +y 0 =x 0 3 +x 1 3 +x 0 2 x 1 +x 0 x 1 2 +x 0 x 1 +x 0 +x 1 .
With regard to the use of the elliptic polynomial equation in the addition of points of an elliptic polynomial with more than one x-coordinate and one or more y-coordinates, we may examine specific coefficients a 1k , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F for F, wherein a set of points EC nx+ny+2 is defined as the (nx+ny+2)-tuple (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ), where x i , y k ∈ F, i ∈ S nx and k ∈ S ny . This set of points comprises solutions of F, although excluding the point (0, 0, . . . , 0) and the point at infinity, (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ).
The rules for conventional elliptic polynomial point addition may be adopted to define an additive binary operation, “+”, over EC nx+ny+2 , i.e., for all
(x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and
(x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 ,
the sum:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )+( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )
is also
(x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 ) ∈ EC nx+ny+2 .
As will be described in greater detail below, (EC nx+ny+2 , +) forms a pseudo-group (p-group) over addition that satisfies the following axioms:
(i) There exists a set (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 such that
( x 0 , x 1 , . . . , x nx ,y 0 , y 1 , . . . , y ny )+( x 0,I , x 1,I , . . . , x nx,I , y o,I , y 1,I , . . . , y ny,i ) =( x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny )
for all (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 ;
(ii) for every set (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 , there exists an inverse set, −(x 0 , x 1 , . . . , x nx , y 0 y 1 , . . . , y ny ) ∈ EC nx+ny+2 , such
( x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny )−( x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny )=that ( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I )
(iii) the additive binary operation in (EC nx+ny+2 , +) is commutative, and the p-group (EC nx+ny+2 , +) forms a group over addition when:
(iv) the additive binary operation in (EC nx+ny+2 , +) is associative.
Prior to a more detailed analysis of the above axioms, the concept of point equivalence must be further developed. Mappings can be used to indicate that an elliptic point represented using (nx+1) x-coordinates and (ny+1) y-coordinates, (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ), is equivalent to one or more elliptic points that satisfy the same elliptic polynomial equation, including the equivalence of an elliptic point to itself.
Points that are equivalent to one another can be substituted for each other at random, or according to certain rules during point addition and/or point doubling operations. For example, the addition of two points
(x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) and
(x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) is given by:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) +( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ).
If the point (x″ 0,1 , x″ 1,1 , . . . , x″ nx,1 , y″ 0,1 , y″ 1,1 , . . . , y″ ny,1 ) is equivalent to the point (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ), then the former can be substituted for (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) in the above equation in order to obtain:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x″ 0,1 , x″ 1,1 , . . . , x nx,1 , y″ 0,1 , y″ 1,1 , . . . , y″ ny,1 ) +( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ).
Mappings that are used to define equivalences can be based on certain properties that exist in elliptic polynomial equations, such as symmetry between variables. As an example, we consider the point (x 0 , x 1 , y 0 ) that satisfies the equation y 0 2 =x 0 3 +x 1 3 +x 0 x 1 . The equivalent of this point may be defined as (x 1 , x 0 , −y 0 ).
With regard to the addition rules for (EC nx+ny+2 , +), the addition operation of two points (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and (x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 , otherwise expressed as:
(x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )+( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )
is calculated in the following manner: First, a straight line is drawn which passes through the two points to be added. The straight line intersects EC nx+ny+2 at a third point, which we denote (x′ 0,3 , x′ 1,3 , . . . , x′ nx,3 , y′ 0,3 , y′ 1,3 , . . . , y′ ny,3 ) ∈ EC nx+ny+2 . The sum point is defined as
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=−( x′ 0,3 , x′ 1,3 , . . . , x′ nx,3 , y′ 0,3 , y′ 1,3 , . . . , y′ ny,3 ).
From the above definition of the addition rule, addition over EC nx+ny+2 is commutative, that is:
( x 0,1 , x 1,1 , . . . , x nx,1 , y 0.1 , y 1,1 , . . . , y ny,1 )+( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )=( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )+( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )
for all (x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and for all (x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 . This community satisfies axiom (iii) above.
There are two primary cases that need to be considered for the computation of point addition for (EC nx+ny+2 , +): (A) for at least one j ∈ S nx , x j,1 ≠x j,2 ; and (B) for all j ∈ S nx , x j,1 =x j,2 =x j,o . Case B includes three sub-cases:
i. for all k ∈ S ny y k,1 =y k,2 that is:
( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )=( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )′
which corresponds to point doubling;
ii. for k ∈ S ny & k≠0, y k,1 =y k,2 , and where y 0,1 & y 0,2 are the roots of the following quadratic equation in y 0 :
a
10
y
0
2
+
∑
k
∈
S
ny
,
k
≠
0
a
1
k
y
k
,
1
2
+
y
0
{
∑
k
∈
S
ny
,
k
≠
0
a
2
k
0
y
k
,
1
+
∑
l
∈
S
ny
,
l
≠
0
a
20
l
y
l
,
1
}
+
∑
k
,
l
∈
S
ny
,
l
≠
k
,
l
&
k
≠
0
a
2
kl
y
k
,
1
y
l
,
1
+
a
30
y
0
+
∑
k
∈
S
ny
,
k
≠
0
a
3
k
y
k
,
1
+
y
0
2
∑
i
∈
S
nx
c
100
i
x
i
,
1
+
y
0
{
∑
k
∈
S
ny
,
i
∈
S
nx
c
1
k
0
i
y
k
,
1
x
i
,
1
+
∑
l
∈
S
ny
,
i
∈
S
nx
c
10
li
y
l
,
1
x
i
,
1
}
+
∑
k
,
l
∈
S
ny
,
l
&
k
≠
0
,
i
∈
S
nx
c
1
kli
y
k
,
1
y
l
,
1
x
i
,
1
+
y
0
∑
l
∈
S
nx
c
20
l
x
l
,
1
+
∑
k
∈
S
ny
,
k
≠
0
,
l
∈
S
nx
c
2
kl
y
k
,
1
x
l
,
1
+
y
0
∑
l
,
i
∈
S
nx
c
30
li
x
l
,
1
x
i
,
1
+
∑
k
∈
S
ny
,
k
≠
0
,
l
,
i
∈
S
nx
c
3
kli
y
k
,
1
x
l
,
1
x
i
,
1
=
∑
l
∈
S
nx
b
1
l
x
l
,
1
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
,
1
2
x
k
,
1
+
∑
l
,
k
∈
S
nx
b
3
lk
x
l
,
1
x
k
,
1
+
∑
k
∈
S
nx
b
4
k
x
k
,
1
+
b
c
,
which corresponds to point inverse; and
iii. all other conditions except those in Cases B.i and B.ii. This case occurs only when ny is greater than or equal to one.
For Case A, for at least one j ∈ S nx x j,1 ≠x j,2 , a straight line in (nx+ny+2)-dimensional space is defined by
y
k
-
y
k
,
1
y
k
,
2
-
y
k
,
1
=
x
j
-
x
j
,
1
x
j
,
2
-
x
j
,
1
,
k ∈ S ny and j ∈ S nx and
x
i
-
x
i
,
1
x
i
,
2
-
x
i
,
1
=
x
j
-
x
j
,
1
x
j
,
2
-
x
j
,
1
,
i≠j, i ∈ S nx .
For this case, y k =m yk x j +c yk , where
m
yk
=
y
k
,
2
-
y
k
,
1
x
j
,
2
-
x
j
,
1
and c yk =y k,1 −x j,1 m yk . Further, x i =m xi x j +c xi , where
m
xi
=
x
i
,
2
-
x
i
,
1
x
j
,
2
-
x
j
,
1
and c xi =x i,1 −x j,1 m xi . Equation (1) can then be re-written as
∑
k
∈
S
ny
a
1
k
y
k
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
y
l
+
∑
k
∈
S
ny
a
3
k
y
k
+
x
j
∑
k
,
l
∈
S
ny
c
1
klj
y
k
y
l
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
1
kli
y
k
y
l
x
i
+
x
j
∑
k
∈
S
ny
c
2
kj
y
k
+
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
c
2
kl
y
k
x
l
+
x
j
2
∑
k
∈
S
ny
c
3
kjj
y
k
+
x
j
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
c
3
klj
y
k
x
l
+
x
j
∑
k
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
3
kji
y
k
x
i
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
,
l
&
i
≠
j
c
3
kli
y
k
x
l
x
i
=
b
1
j
x
j
3
+
∑
l
∈
S
nx
,
l
≠
j
b
1
l
x
l
3
+
x
j
2
∑
k
∈
S
nx
,
k
≠
j
b
2
jk
x
k
+
x
j
∑
l
∈
S
nx
,
l
≠
j
b
2
lj
x
l
2
+
∑
l
,
k
∈
S
nx
,
l
,
k
≠
j
,
l
≠
k
b
2
lk
x
l
2
x
k
+
b
3
jj
x
j
2
+
x
j
∑
k
∈
S
nx
,
k
≠
j
b
3
jk
x
k
+
x
j
∑
l
∈
S
nx
,
l
≠
j
b
3
lj
x
l
+
∑
l
,
k
∈
S
nx
,
l
,
k
≠
j
b
3
lk
x
l
x
k
+
b
4
j
x
j
+
∑
k
∈
S
nx
,
k
≠
j
b
4
k
x
k
+
b
c
,
and substitution of the above into the rewritten equation (1) for y k , k ∈ S ny and x i , i ∈ S nx & i≠j, results in:
∑
k
∈
S
ny
a
1
k
(
m
yk
x
j
+
c
yk
)
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
(
m
yk
x
j
+
c
yk
)
(
m
yl
x
j
+
c
yl
)
+
∑
k
∈
S
ny
a
3
k
(
m
yk
x
j
+
c
yk
)
+
x
j
∑
k
,
l
∈
S
ny
c
1
klj
(
m
yk
x
j
+
c
yk
)
(
m
yl
x
j
+
c
yl
)
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
1
kli
(
m
yk
x
j
+
c
yk
)
(
m
yl
x
j
+
c
yl
)
(
m
xi
x
j
+
c
xi
)
+
x
j
∑
k
∈
S
ny
c
2
kj
(
m
yk
x
j
+
c
yk
)
+
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
2
2
kl
(
m
yk
x
j
+
c
yk
)
(
m
xl
x
j
+
c
xl
)
+
x
j
2
∑
k
∈
S
ny
c
3
kjj
(
m
yk
x
j
+
c
yk
)
+
x
j
∑
k
∈
S
ny
,
l
∈
S
nx
,
l
≠
j
c
3
klj
(
m
yk
x
j
+
c
yk
)
(
m
xl
x
j
+
c
xl
)
+
x
j
∑
k
∈
S
ny
,
i
∈
S
nx
,
i
≠
j
c
3
kji
(
m
yk
x
j
+
c
yk
)
(
m
xi
x
j
+
c
xi
)
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
,
l
&
i
≠
j
c
3
kli
(
m
yk
x
j
+
c
yk
)
(
m
xl
x
j
+
c
xl
)
(
m
xi
x
j
+
c
xi
)
=
b
1
j
x
j
3
+
∑
l
∈
S
nx
,
l
≠
j
b
1
l
(
m
xl
x
j
+
c
l
)
3
+
x
j
2
∑
k
∈
S
nx
,
k
≠
j
b
2
jk
(
m
xk
x
j
+
c
xk
)
+
x
j
∑
l
∈
S
nx
,
l
≠
j
b
2
lj
(
m
xl
x
j
+
c
xl
)
2
+
∑
l
,
k
∈
S
nx
,
l
&
k
≠
j
,
l
≠
k
b
2
lk
(
m
xl
x
j
+
c
xl
)
2
(
m
xk
x
j
+
c
xk
)
+
b
3
jj
x
j
2
+
x
j
∑
k
∈
S
nx
,
k
≠
j
b
3
jk
(
m
xk
x
j
+
c
xk
)
+
x
j
∑
l
,
k
∈
S
nx
,
l
≠
j
b
3
lj
(
m
xl
x
j
+
c
xl
)
+
∑
l
,
k
∈
S
nx
,
l
&
k
≠
j
b
3
lk
(
m
xl
x
j
+
c
xl
)
(
m
xk
x
j
+
c
xk
)
+
b
4
j
x
j
+
∑
k
∈
S
nx
,
k
≠
j
a
6
k
(
m
xk
x
j
+
c
xk
)
+
b
c
Expanding the terms in the above equation leads to a cubic equation of the form x j , C 3 x j 3 +C 2 x j 2 +C 1 x j +C 0 =0, where C 3 , C 2 , C 1 & C 0 are obtained from the above equation.
Assuming C 3 ≠0, the above cubic equation in x j has three roots x j,1 x j,2 , & x j,3 and can be written as (x j −x j,1 )(x j −x j,2 )(x j −x j,3 )=0. Normalizing by the coefficient of x 3 and equating the coefficients of x 2 in the resulting equation with that in (x j −x j,1 )(x j −x j,2 )(x j −x j,3 )=0, one obtains a solution for x′ j,3 :
x
j
,
3
′
=
-
C
2
C
3
-
x
j
,
1
-
x
j
,
2
.
(
2
)
The values of y′ k,3 , k ∈ S ny , and x′ i,3 , i ∈ S nx & i≠j, may be similarly obtained from equations for x j =x′ j,3 .
For cases where C 3 =0, C 3 x j 3 +C 2 x j 2 +C 1 x j =C 0 =0 becomes a quadratic equation. Such quadratic equations may be used in the definition of point equivalences.
With regard to Case B for all j ∈ S nx , x j,1 =x j,2 , the three sub-cases are considered below. In all cases, x j,0 is defined as x j,0 =x j,1 =x j,2 , j ∈ S nx .
For Case B.i., all k ∈ S ny , y k,1 =y k,2 , which corresponds to point doubling. In this case,
( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 )=( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , . . . , y ny,2 ).
Letting:
( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o )=( x 0,1 , x 1,1 , . . . , x nx,1 , y 0,1 , y 1,1 , . . . , y ny,1 ) =( x 0,2 , x 1,2 , . . . , x nx,2 , y 0,2 , y 1,2 , . . . , y ny,2 )
the sum is written as
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o )+( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o ) (3)
There are several ways of defining the addition in this case. Three possible rules are described below. Case B.i.1: Letting S nx,Lx denote a subset of S nx with Lx elements, i.e., S nx,Lx ⊂ S nx ; letting S ny,Ly denote a subset of S ny with Ly elements and which does not include the element 0; i.e. S ny,Ly ⊂ S ny and 0∉ S ny,Ly ; setting the value of Lx and Ly as at least one, then the straight line in this case can be defined as a tangent to the point (x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o ) defined in a sub-dimensional space with coordinates y n and x m with n ∈ S ny,Ly and M ∈ S nx,Lx .
In this case, the gradients m yn and m xm of the straight line to be used in equation (2) are essentially the first derivatives of y n and x m , n ∈ S ny,Ly and m ∈ S nx,Lx , for F with respect to x j , j ∈ S nx,Lx , i.e.,
m
yn
=
y
n
x
j
and
m
xn
=
x
m
x
j
.
Using these derivatives for the values of the gradients,
m
yn
=
y
n
x
j
,
where n ∈ S ny,Ly , and
m
yn
=
x
m
x
j
,
where m ∈ S nx,Lx , in equation (2) and noting that it is assumed that
x
m
x
j
=
0
,
for m ∈ (S nx −S nx,Lx ) and
y
n
x
j
=
0
,
for n ∈ (S ny −S ny,Lx ) then a solution for x′ j,3 may be obtained.
The values of y′ n,3 for n ∈ S ny and x′ m,3 , for m ∈ S nx & m≠j, can further be obtained for x j =x′ j,3 . The choice of the x m -coordinates, m ∈ S nx,Lx , and y n -coordinates, n ∈ S ny,Ly , which can be used to compute the tangent of the straight line in this case may be chosen at random or according to a pre-defined rule. Further, a different choice of the x m -coordinates, m ∈ S nx,Lx and y n -coordinates, n ∈ S ny,Ly , may be made when one needs to compute successive point doublings, such as that needed in scalar multiplication.
With regard to the next case, Case B.i.2, the second possible way of defining the addition of a point with itself is to apply a sequence of the point doublings according to the rule defined above in Case B.i.1, where the rule is applied with a different selection of the x-coordinate(s) and y-coordinates(s) in each step of this sequence.
In the third sub-case, Case B.i.3, a point is substituted with one of its equivalents. Letting (x 0,oe , x 1,oe , . . . , x nx,oe , y 0,oe , y 1,oe , . . . , y ny,oe ) represent the equivalent point of (x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o ) then equation (3) may be written as:
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,o , y 1,o , . . . , y ny,o )+( x 0,oe , x 1,oe , . . . , x nx,oe , y 0,oe , y 1,oe , . . . , y ny,oe ).
With regard to Case B.ii, for k ∈ S ny & k≠0, y k,1 =y k,2 , and where y 0,1 & y 0,2 are the roots of the quadratic equation in y 0 , this case corresponds to generation of the point inverse.
Letting y k,1 =y k,2 =y k,o for k ∈ S ny & k≠0, then any two points, such as the point (x 0,o , x 1,o , . . . , x nx,o , y 0,1 y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 and the point (x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 , are in the hyper-plane with x i =x i,o , i ∈ S nx and y k =y k,o , k ∈ S ny & k≠0 . Thus, any straight line joining these two points such that (x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,o , . . . , y ny,o )≠(x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,o , . . . , y ny,o ) is also in this hyper-plane.
Substituting the values of x 0,o , x 1,o , . . . , x nx,o , y 1,o , . . . , & y ny,o in an elliptic polynomial equation with multiple x-coordinates and multiple y-coordinates, a quadratic equation for y 0 is obtained, as given above. Thus, y 0 has only two solutions y 0,1 & y 0,2 .
Thus, a line joining points (x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 and (x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,o , . . . , y ny,o ) ∈ EC nx+ny+2 does not intersect with EC nx+ny+2 at a third point.
A line that joins these two points is assumed to intersect with EC nx+ny+2 at infinity (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 . This point at infinity is used to define both the inverse of a point in EC nx+ny+2 and the identity point. According to the addition rule defined above, one can write:
( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny )+( x 0 , x 1 , . . . , x nx , y 0,2 , y 1 , . . . , y ny ) =( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) (4)
since the third point of intersection of such lines is assumed to be at infinity, (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 . Thus, this equation defines a unique inverse for any point (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 , namely:
−( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny )=( x 0 , x 1 , . . . , x nx , y 0,2 , y 1 , . . . , y ny ).
Thus, equation (4) can be written as:
( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny )−( x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny ) =( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) (5).
Further, a line joining the point at infinity (x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) ∈ EC nx+ny+2 and a point (x 0 , x 1 , . . . , x nx , y 0,1 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 will intersect with EC nx+ny+2 at (x 0 , x 1 , . . . , x nx , y 0,2 , y 1 , . . . , y ny ) ∈ EC nx+ny+2 . Thus, from the addition rule defined above,
( x 0 , x 1 , . . . , x nx , y 0 , y 1 , y 2 , . . . , y ny )+( x 0,I , x 1,I , . . . , x nx,I , y 0,I , y 1,I , . . . , y ny,I ) =( x 0 , x 1 , x nx , y 0 , y 1 , . . . , y ny ) (6)
Equation (5) satisfies axiom (ii) while equation (6) satisfies axiom (i), defined above.
Case B.iii applies for all other conditions except those in cases B.i and B.ii. This case occurs only when ny is greater than or equal to one. Given two points (x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 ) ∈ EC nx+ny+2 and (x 0,o , x 1,o , . . . , x nx,o y 0,2 , y 1,2 , . . . , y ny,2 ) ∈ EC nx+ny+2 that do not satisfy the conditions of cases B.i and B.ii above, the sum point is written as
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y y 1,1 , . . . , y ny,1 )+( x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,2 , . . . , y ny,2 )
There are several possible rules to find the sum point in this case. Three possible methods are given below:
1) Using three point doublings and one point addition,
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=4( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 ); −2( x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,2 , . . . , y ny,2 )
2) using one point doublings and three point additions,
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=(2( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 ) +( x 0,o , x 1,o , . . . , x nx,o , x 0,2 , y 1,2 , . . . , y ny,2 ))−; and ( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 )
3) using point equivalence,
( x 0,3 , x 1,3 , . . . , x nx,3 , y 0,3 , y 1,3 , . . . , y ny,3 )=( x 0,o , x 1,o , . . . , x nx,o , y 0,1 , y 1,1 , . . . , y ny,1 )+( x 0,oe , x 1,oe , . . . , x nx,oe , y 0,2e , y 1,2e , . . . , y ny,2e ),
where (x 0,oe , x 1,oe , . . . , x nx,oe , y 0,2e , . . . , y 1,2e , . . . , y ny,2e ) is assumed to be the equivalent point of (x 0,o , x 1,o , . . . , x nx,o , y 0,2 , y 1,2 , . . . , y ny,2 ).
It should be noted that the above methods for defining the sum point are not the only ones that can be defined and are provided for exemplary purposes only. The choice of method used to obtain the sum point in this case should depend on the computation complexity of point addition and point doubling.
With regard to associativity, one way of proving associativity of (EC nx+ny+2 , +) is as follows: Given particular elliptic polynomial equations (i.e., for particular coefficients a 1l , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k , b c ∈ F) defined over a finite F, if it can be shown by computation that any point Q ∈ EC nx+ny+2 (and any of its equivalent points) can be uniquely written as k Q P ∈ EC nx+ny+2 , where P is the generator point of (EC nx+ny+2 , +), then the corresponding EC nx+ny+2 groups based on such polynomials are associative. This is because any three points Q, R, S ∈ EC nx+ny+2 (or any of their equivalent points) can be written as k Q P, k R P, k S P ∈ EC nx+ny+2 , respectively, thus their sum (Q+R+S)=(k Q P+k R P+k S P)=(k Q +k R +k S )P can be carried out in any order.
The following elliptic polynomial equation with nx=1 and ny=0 is used to show an example of the equations in Case A used in point addition: y 0 2 =x 0 3 +x 1 3 +x 0 x 1 . Choosing x j =x 0 , and substituting y k =m yk x j +c yk from Case A above for y 0 , and the corresponding equation x i =m xi x j +c xi for x 1 , one obtains (m y0 x 0 +c y0 ) 2 =x 0 3 +(m xi x 0 +c x1 ) 3 +x 0 (m x1 x 0 +c x1 ).
Expanding this equation yields the equation
m y0 2 x 0 2 +2 m y0 c y0 x 0 +c y0 2 =x 0 3 +m x1 3 x 0 3 +3 m m1 2 c x1 x 0 2 +3 m m1 c x1 2 x 0 +c x1 3 +m x1 x 0 2 +c x1 x 0
or
(1 +m x1 3 ) x 0 3 +(3 m x1 2 c x1 +m x1 −m y0 2 ) x 0 2 +(3 m x1 c x1 2 +c x1 −2 m y0 c y0 ) x 0 +c x1 3 −c y0 2 =0.
From equation (2), the solution for x′ 0,3 in this case is obtained:
x
0
,
3
′
=
-
(
3
m
xl
2
c
x
1
+
m
x
1
-
m
y
0
2
)
(
1
+
m
x
1
3
)
-
x
j
,
1
-
x
j
,
2
.
Similarly, one can obtain the values of y 0,3 and x′ 1,3 for x 0 =x′ 0,3 .
It should be noted that when m x1 =−1, the coefficient of the cubic term in the above is zero; i.e. C 3 =0. In this case, the resulting quadratic equation can be used in the definition of point equivalences for the points that satisfy the elliptic polynomial equation.
Each of the equations for point addition and point doublings derived for cases A and B above require modular inversion or division. In cases where field inversions or divisions are significantly more expensive (in terms of computational time and energy) than multiplication, projective coordinates are used to remove the requirement for field inversion or division from these equations.
Several projective coordinates can be utilized. In the preferred embodiment, the Jacobean projective coordinate system is used. As an example, we examine:
x
i
=
X
i
V
2
for
i
∈
S
nx
;
and
;
and
(
7
)
y
k
=
Y
k
V
3
for
k
∈
S
ny
.
(
8
)
Using Jacobian projection yields:
∑
k
∈
S
ny
a
1
k
Y
k
2
V
6
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
Y
k
V
3
Y
l
V
3
+
∑
k
∈
S
ny
a
3
k
Y
k
V
3
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
c
1
kli
Y
k
V
3
Y
l
V
3
X
i
V
2
+
∑
k
∈
S
ny
,
l
∈
S
nx
c
2
kl
Y
k
V
3
X
l
V
2
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
c
3
kli
Y
k
V
3
X
l
V
2
X
i
V
2
=
∑
l
∈
S
nx
b
1
l
X
l
3
V
6
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
X
l
2
V
4
X
k
V
2
+
∑
l
,
k
∈
S
nx
b
3
lk
X
l
V
2
X
k
V
2
+
∑
k
∈
S
nx
b
4
k
X
k
V
2
+
b
c
(
9
)
which can be rewritten as:
∑
k
∈
S
ny
a
1
k
Y
k
2
V
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
Y
k
Y
l
V
2
+
∑
k
∈
S
ny
a
3
k
Y
k
v
5
+
∑
k
,
l
∈
S
ny
,
i
∈
S
nx
c
1
kli
Y
k
Y
l
X
i
+
∑
k
∈
S
ny
,
l
∈
S
nx
c
2
kl
Y
k
X
l
V
3
+
∑
k
∈
S
ny
,
l
,
i
∈
S
nx
c
3
kli
Y
k
X
l
X
i
V
=
∑
l
∈
S
nx
b
1
l
X
l
3
V
2
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
X
l
2
X
k
V
2
+
∑
l
,
k
∈
S
nx
b
3
lk
X
l
X
k
V
4
+
∑
k
∈
S
nx
b
4
k
X
k
V
6
+
b
c
V
8
.
(
10
)
In the following, the points (X 0 , X 1 , . . . , X nx , Y 0 , Y 1 , . . . , Y ny , V) are assumed to satisfy equation (10). When V≠0, the projected point (X 0 , X 1 , . . . , X nx , Y 0 , Y 1 , . . . , Y ny , V) corresponds to the point:
(
x
0
,
x
1
,
…
,
x
nx
,
y
0
,
y
1
,
…
,
y
ny
)
=
(
X
0
V
2
,
X
1
V
2
,
…
,
X
nx
V
2
,
Y
0
V
3
,
Y
1
V
3
,
…
,
Y
ny
V
3
)
,
which satisfies equation (1).
Using Jacobean projective coordinates, equation (10) can be written as:
(
X
0
,
3
V
3
2
,
X
1
,
3
V
3
2
,
…
,
X
nx
,
3
V
3
2
,
Y
0
,
3
V
3
3
,
Y
1
,
3
V
3
3
,
…
,
Y
ny
,
3
V
3
3
)
=
(
X
0
,
1
V
1
2
,
X
1
,
1
V
1
2
,
…
,
X
nx
,
1
V
1
2
,
Y
0
,
1
V
1
3
,
Y
1
,
1
V
1
3
,
…
,
Y
ny
,
1
V
1
3
)
+
(
X
0
,
2
V
2
2
,
X
1
,
2
V
2
2
,
…
,
X
nx
,
2
V
2
2
,
Y
0
,
2
V
2
3
,
Y
1
,
2
V
2
3
,
…
,
Y
ny
,
2
V
2
3
)
.
(
11
)
By using Jacobian projective coordinates in the equations of Cases A and B above, and by an appropriate choice of the value of V 3 , it can be shown that point doubling and point addition can be computed without the need for field inversion or division.
In order to examine the embedding method, the twist of an elliptic polynomial equation needs to be defined. Given an elliptic polynomial with (nx+1) x-coordinates and (ny+1) y-coordinates of the form described above:
y
0
2
+
∑
k
∈
S
ny
a
1
k
y
k
2
+
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
y
l
=
∑
l
∈
S
nx
b
1
l
x
l
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
2
x
k
,
(
12
)
where a 1l , a 2kl , b 1l , b 2lk ∈ F.
Given certain values for the x-coordinates x 0,o , x 1,o , . . . , x nx,o and y-coordinates y 1,o , . . . , y ny,o , respectively, that are elements of the finite field, F, these values are substituted into the elliptic polynomial equation (1) in order to obtain a quadratic equation in y 0 :
y
0
2
=
-
∑
k
∈
S
ny
a
1
k
y
k
,
o
2
-
∑
k
,
l
∈
S
ny
,
l
≠
k
a
2
kl
y
k
,
o
y
l
,
o
+
∑
l
∈
S
nx
b
1
l
x
l
,
o
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
,
o
2
x
k
,
o
=
T
.
If a solution of the above quadratic equation (i.e., y 0 2 =T) is an element of the finite field F, the point (x 0,o , x 1,o , . . . , x nx,o , y 0 , y 1,o , . . . , y ny,o ) is said to satisfy the given elliptic polynomial equation. If a solution of the above quadratic equation is not an element of the finite field F, the point (x 0,o , x 1,o , . . . , x nx,o , y 0 , y 1,o , . . . , y ny,o ) is said to satisfy the twist of the given elliptic curve equation. The inventive embedding method is based on the isomorphic relationship between a curve and its twist as described in the following theorem:
An elliptic polynomial equation of the form given above is isomorphic to its twist if:
1) there are mathematical mappings that can be defined on the values of x 0 , x 1 , . . . , x nx , y 1 , . . . , y ny (i.e., φ x (x i ) and φ y (y i )) such that any point (x 0 , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) that satisfies such an elliptic polynomial equation can be mapped into another point (φ x (x 0 ), φ x (x 1 ), . . . , φ x (x xn ), φ y (y 0 ), φ y (y 1 ), . . . , φ y (y ny )) that satisfies the twist of the same elliptic polynomial equation; and 2) the mapping between the points (x o , x 1 , . . . , x nx , y 0 , y 1 , . . . , y ny ) and (φ x (x 0 ), φ x (x 1 ), . . . , φ x (x xn ), φ y (y 0 ), φ y (y 1 ), . . . , φ y (y ny )), is unique, i.e., a one-to-one correspondence.
The proof of this theorem is as follows. Re-writing equation (12) as:
y
0
2
=
-
∑
k
∈
S
n
y
a
1
k
y
k
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
y
l
+
∑
l
∈
S
n
x
b
1
l
x
l
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
2
x
k
,
(
13
)
,
and letting the right-hand side of equation (13) be denoted as T, then:
T
=
-
∑
k
∈
S
n
y
a
1
k
y
k
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
y
l
+
∑
l
∈
S
n
x
b
1
l
x
l
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
2
x
k
.
(
14
)
Thus, any value of x 0 , x 1 , . . . , x nx , y 1 , . . . , y ny will lead to a value of T ∈ F(p). T could be quadratic residue or non-quadratic residue. If T is quadratic residue, then equation (14) is written as:
T
q
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
y
l
,
q
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
2
x
k
,
q
(
15
)
where x 0,q , x 1,q , . . . , x nx,q , y 1,q , . . . , y ny,q ∈ F denotes the values of x 0 , x 1 , . . . , x nx , y 1 , . . . , y ny that result in a quadratic residue value of T, which is hereafter denoted as T q .
If T is non-quadratic residue, then equation (14) is written as:
T
q
_
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
_
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
_
y
l
,
q
_
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
_
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
_
2
x
k
,
q
_
(
16
)
where x 0, q , x 1, q , . . . , x nx, q , y 1, q , . . . , y ny, q ∈ F denotes the values of x 0 , x 1 , . . . , y 1 , . . . , y ny that result in a non-quadratic residue value of T, denoted as T q .
Letting g be any non-quadratic residue number in F (i.e., g ∈ F(p) & √{square root over (g)}∉ F(p)), then multiplying equation (15) with g 3 yields:
g
3
T
q
=
-
g
3
∑
k
∈
S
n
y
a
1
k
y
k
,
q
2
-
g
3
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
y
l
,
q
+
g
3
∑
l
∈
S
n
x
b
1
l
x
l
,
q
3
+
g
3
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
2
x
k
,
q
,
which can be re-written as:
g
3
T
q
=
-
∑
k
∈
S
n
y
a
1
k
(
g
3
y
k
,
q
)
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
g
3
(
g
3
y
k
,
q
)
(
g
3
y
l
,
q
)
+
∑
l
∈
S
n
x
b
1
l
(
g
x
l
,
q
)
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
(
g
x
l
,
q
)
2
(
g
x
k
,
q
)
.
(
17
)
It should be noted that if g is non-quadratic residue, then g 3 is also non-quadratic residue. Further, the result of multiplying a quadratic residue number by a non-quadratic residue number is a non-quadratic residue number. Thus, g 3 T q is non-quadratic residue.
By comparing the terms of equations (16) and (17), we obtain the following mappings:
x i, q =gx i,q ; (18);
y i , q =√{square root over (g 3 )}y i,q ; (19); and
T q =g 3 T q (20).
The mappings between the variables x i,q and x i, q in equation (18), y i,q and y i, q in equation (19), and T q and T q in equation (20) are all bijective, i.e., there is a one-to-one correspondence from basic finite field arithmetic. As a consequence, the mappings between the (nx+ny+2)-tuple (x 0,q , x 1,q , . . . , x nx,q , T q , y 1,q , . . . , y ny,q ) and the (nx+ny+2)-tuple (x 0, q , x 1, q , . . . , x nx, q , T q , y 1, q , . . . , y ny, q ) are also bijective.
Therefore, for every solution of equation (15), there is an isomorphic solution that satisfies equation (16), and since the mappings of the coordinates of one to the other are given in equations (18)-(20), these two solutions are isomorphic with respect to each other.
Since T q is quadratic residue, this expression can be written as:
T q =y 0 2 . (21)
Thus, from equation (20), T q can be written as:
T q =g 3 y 0 2 (22).
Using equations (21) and (22), equations (15) and (16) can be written as:
y
0
2
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
y
l
,
q
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
2
x
k
,
q
;
(
23
)
and
g
3
y
0
2
=
-
∑
k
∈
S
n
y
a
1
k
y
k
,
q
_
2
-
∑
k
,
l
∈
S
n
y
,
l
≠
k
a
2
k
l
y
k
,
q
_
y
l
,
q
_
+
∑
l
∈
S
n
x
b
1
l
x
l
,
q
_
3
+
∑
l
,
k
∈
S
n
x
,
l
≠
k
b
2
l
k
x
l
,
q
_
2
x
k
,
q
_
.
(
24
)
Since any solution of equation (15) has an isomorphic solution that satisfies equation (16), it follows that the solution of equation (23), denoted as (x 0,q , x 1,q , . . . , x nx,q , y 0 , y 1,q , . . . , y ny,q ), has an isomorphic solution that satisfies equation (24), denoted as
(
g
x
0
,
q
,
g
x
1
,
q
,
…
,
g
x
n
x
,
q
,
g
3
2
y
0
,
g
3
2
y
1
,
q
,
…
,
g
3
2
y
n
y
,
q
)
.
The solutions of equation (23) form the points (s 0,q , x 1,q , . . . , x nx,q , y 0 , y 1,q , . . . , y ny,q ) that satisfy an elliptic polynomial. Similarly, the solutions of equation (24) form the points
(
g
x
0
,
q
,
g
x
1
,
q
,
…
,
g
x
n
x
,
q
,
g
3
2
y
0
,
g
3
2
y
1
,
q
,
…
,
g
3
2
y
n
y
,
q
)
that satisfy its twist. This proves the above theorem.
An example of a mapping of the solutions of equation (23) defined over F(p), where p=3 mod 4, to the solutions of its twist is implemented by using −x i for the x-coordinates and −y i 2 for the y-coordinates.
The isomorphism between an elliptic polynomial and its twist, discussed above, is exploited for the embedding of the bit sting of a shared secret key into the appropriate x and y coordinates of an elliptic polynomial point without the need for an iterative search for a quadratic residue value of the corresponding y 0 -coordinate, which usually requires several iterations, where the number of iterations needed is different for different bit strings which are being embedded.
Assuming F=F(p) and that the secret key is an M-bit string such that (nx+ny+1)N>M>N−1, where N is the number of bits needed to represent the elements of F(p), then the secret key bit string is divided into (nx+ny+1) bit-strings k x,0 , k x,1 , . . . , k x,nx , k y,1 , . . . , k k,ny . The value of the bit-strings k x,0 , k x,1 , . . . , k x,nx , k y,1 , . . . , k k,ny must be less than p. In the preferred embodiment of embedding the (nx+ny+1) bit-strings k x,0 , k x,1 , . . . , k x,nx , k y,1 , . . . , k k,ny , the embedding is as follows.
First, assign the value of the bit string of k x,0 , k x,1 , . . . , k x,nx to x 0,k , x 1,k , . . . , x nx,k . Next, assign the value of the bit string of k y,1 , . . . , k k,ny to y 1,k , . . . , y ny,k . Then, compute:
T
=
-
∑
i
∈
S
n
y
a
1
i
y
i
,
k
2
-
∑
i
,
l
∈
S
n
y
,
l
≠
i
a
2
i
l
y
i
,
k
y
l
,
k
+
∑
l
∈
S
n
x
b
1
l
x
l
,
k
3
+
∑
l
,
i
∈
S
n
x
,
l
≠
i
b
2
l
i
x
l
,
k
2
x
i
,
k
.
Finally, use the Legendre test to see if T has a square root. If T has a square root, assign one of the roots to y 0 ; otherwise, the x-coordinates and y-coordinates of the elliptic polynomial point with the embedded shared secret key bit string are given by gx i,k and
g
3
2
y
i
,
k
,
respectively, where g is non-quadratic residue in F.
It should be noted that p is usually predetermined prior to encryption, so that the value of g can also be predetermined. Further, the receiver can identify whether the point (x, 0,k , x 1,k , . . . , x nx,k , y 0,k , . . . , y ny,k ) or the point
(
g
x
0
,
k
,
g
x
1
,
k
,
…
,
g
x
n
x
,
k
,
g
3
2
y
0
,
k
,
g
3
2
y
1
,
k
,
…
,
g
3
2
y
n
y
,
k
)
is the elliptic polynomial point with the embedded secret key bit strings without any additional information. Additionally, any non-quadratic value in F(p) can be used for g. For efficiency, g is chosen to be −1 for p≡3 mod 4 and g is chosen to be 2 for p≡1 mod 4 .
The same deterministic and non-iterative method described above can be used to embed a secret message bit string into an elliptic polynomial point in a deterministic and non-iterative manner. Assuming F=F(p) and that the message is an M-bit string such that (nx+ny+1)N>M>N−1, where N is the number of bits needed to represent the elements of F(p), then the message bit string is divided into (nx+ny+1) bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 . . . , m k,ny . The value of the bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 , . . . , m k,ny must be less than p. As the previous embodiment, the embedding of the (nx+ny+1) bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 , . . . , m k,ny can be accomplished out as follows.
First, assign the value of the bit string of m x,0 , m x,1 , . . . , m x,nx to x 0,m , x 1,m , . . . , x nx,m . Next, assign the value of the bit string of m y,1 , . . . , m k,ny to y 1,m , . . . , y ny,m . Then compute:
T
=
-
∑
i
∈
S
n
y
a
1
i
y
i
,
m
2
-
∑
i
,
l
∈
S
n
y
,
l
≠
i
a
2
i
l
y
i
,
m
y
l
,
m
+
∑
l
∈
S
n
x
b
1
l
x
l
,
m
3
+
∑
l
,
i
∈
S
n
x
,
l
≠
i
b
2
l
i
x
l
,
m
2
x
i
,
m
.
Finally, use the Legendre test to see if T has a square root. If T has a square root, then assign one of the roots to y 0 , otherwise the x-coordinates and y-coordinates of the elliptic polynomial point with the embedded shared secret key bit string are given by gx i,m and
g
3
2
y
i
,
m
,
respectively.
It should be noted that p is usually predetermined prior to encryption; thus, the value of g can also be predetermined. Further, when using the above method, the strings m x,0 , m x,1 , . . . , m x,nx and m y,1 , . . . , m k,ny can be recovered directly from x 0,m , x 1,m , . . . , x nx,m and y 1,m , . . . , y ny,m , respectively. An extra bit is needed to identify whether (x 0,m , x 1,m , . . . , x nx,m , y 0,m , y 1,m , . . . , y 1,m , . . . , y ny,m ) or
(
g
x
0
,
m
,
g
x
1
,
m
,
…
,
g
x
n
x
,
m
,
g
3
2
y
0
,
m
,
g
3
2
y
1
,
m
,
…
,
g
3
2
y
n
y
,
m
)
is used at the sending correspondent. Additionally, any non-quadratic value in F(p) can be used for g. For efficiency, g is chosen to be −1 for p≡3 mod 4 and is chosen to be 2 for p≡1 mod 4. Further, at the receiver, the process is reversed. In the case of g=2, a division by two is carried out. It should noted that dividing x i,m by two is computed using one modulo addition, because:
x i,m /2=(( x i,m −( x i,m )mod 2)/2)+( x i,m )mod 2*(1/2)mod p; (i)
(x i,m )mod 2 is the least significant bit of x i,m ; (ii) and
(1/2)mod p =( p+ 1)/2. (iii)
In the following, it is assumed that F=F(p). In conventional elliptic polynomial cryptography, the cipher point is computed as:
( x 0,c , x 1,c , . . . , x nx,c , y 0,c , y 1,c , . . . , y ny,c )=( x 0,m , x 1,m , . . . , x nx,m , y 0,m , y 1,m , . . . , y ny,m )+(x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )
where (x 0,m , x 1,m , . . . , x nx,m , y 0,m , y 1,m , . . . , y ny,m ) is the message point and (x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk ) is referred to as “the encrypting point”, which is typically computed as the scalar multiplication of the shared secret key k and a base point (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ), which is known publicly, such that:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )= k ( x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B )
where all of the coordinates are elements of the underlying field F(p), and which are presented in N-bits.
In the present method, the underlying elliptic polynomial cryptography is based on the principle of blinding the base point by making the base point dependant on the shared secret key. In a first embodiment, the shared secret key is represented using nN-bits, where n is greater than or equal to (nx+ny+2). The nN bits of the shared secret key could be given as one block, or may alternatively be generated using a smaller bit-string and a random number generator.
When n=2(nx+ny+2), various techniques may be used to generate the encrypting point (x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk ). In the preferred embodiment, which utilizes one blinded base point, (nx+ny+1)N-bits of the shared secret key bit-string k1 are embedded into the x-coordinates and the y-coordinates of an elliptic polynomial point to obtain a blinded base point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ). The other (nx+ny+2)N-bits of the key are used as a scalar value k2, which is then multiplied by the blinded base point to obtain the encrypting point:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )= k 2( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ).
In an alternative embodiment, one blinded base point and one public base point are utilized. In this embodiment, (nx+ny+1)N-bits of the shared secret key bit-string k1 are embedded into the x-coordinates and the y-coordinates of an elliptic curve point to obtain a blinded base point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ). The other N-bits of the key are used as a scalar value k2 that is multiplied by the public base point (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ), with the encrypting point being obtained by:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )=( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+ k 2( x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B )
In another alternative embodiment, two blinded base points are utilized. In this embodiment, the two blinded points are generated from the 2(nx+ny+2)N-bits of the shared secret key bit-string. One blinded point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ) is generated from (nx+ny+1)N-bits out of the (nx+ny+2)N-bits of the shared secret key bit-string, while the other blinded base point (x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ) is obtained from a different (nx+ny+1)N-bits combination of the 2(nx+ny+2)N-bits of the shared secret key bit-string. Part of the key used to generate the point (x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ), k 1 , is used as a scalar and multiplied by the point (x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ) to obtain the scalar multiplication k1(x 0,k2 , x 1,k2 , . . . , x nx,k2 y 0,k2 , y 1,k2 , . . . , y ny,k2 ). The other part of the key used to generate the point (x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ), k 2 is used as a scalar and multiplied by the point (x 0,k1 , x 1,k1 , . . . , y 0,k1 , y 1,k1 , . . . , y ny,k1 ) to obtain the scalar multiplication k2(x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 ). The encrypting point can be selected using a first scalar multiplication and addition:
( x 0,bk , x 1,bk , . . . , x nx,bk y 0,bk , y 1,bk , . . . , y ny,bk )= k 2( x 0,k1 , x 1,k1 , . . . , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+ k 1( x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 )
or, alternatively, a scalar multiplication and addition as:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )=( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+ k 1( x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 )
or, in a further alternative, a scalar multiplication and addition given by:
( x 0,bk , x 1,bk , . . . , x nx,bk , y 0,bk , y 1,bk , . . . , y ny,bk )= k 2( x 0,k1 , x 1,k1 , . . . , x nx,k1 , y 0,k1 , y 1,k1 , . . . , y ny,k1 )+( x 0,k2 , x 1,k2 , . . . , x nx,k2 , y 0,k2 , y 1,k2 , . . . , y ny,k2 ).
In a first embodiment, the elliptic polynomial encryption method includes the following steps (where it is assumed that the data bit stream is encrypted in blocks of M-bits where (nx+ny+1)N>M>N−1, where N is the number of bits needed to represent the elements of f(p): (a) defining a maximum block size that can be embedded into (nx+1) x-coordinates and ny y-coordinates, wherein n is an integer, and setting the maximum block size to be (nx+ny+1)N bits, wherein N is an integer; (b) a sending correspondent and a receiving correspondent agree upon the values of nx and ny, and further agree on a set of coefficients a 1k , a 2kl , a 3k , c 1lki , c 2kl , c 3kli , b 1l , b 2lk , b 3lk , b 4k & b c ∈ F, wherein F represents a finite field where the field's elements can be represented in N-bits, the sending and receiving correspondents further agreeing on a set of random numbers k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l , where l=1, . . . N p , the set of random numbers being shared secret keys for communication, wherein N p ≧1, the sending and receiving correspondents further agreeing on a base point on an elliptic polynomial (x 0,B , x 1,B , . . . , x nx,B , y 0,B , y 1,B , . . . , y ny,B ) ∈ EC xn+ny+2 , wherein EC nx+ny+2 represents the elliptic polynomial.
The sending correspondent then performs the following steps: (c) establishing an integer index /such that a first block is defined by j=0, and embedding a j th block of a secret message bit string into an elliptic curve message point (x 0,m,j , x 1,m,j , . . . , x nx,m,j , y 0,m,j , y 1,m,j , . . . , y ny,m,j ); (d) embedding the, j th block shared secret key bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into a set of elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), where l=1, . . . N p , and wherein the set of elliptic curve key points satisfy the elliptic polynomial at the elliptic curve message point; (e) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr ny,1,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l )+ k s,j,l (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (f) calculating a cipher point of the j th data block (x c,0,j , x c,1,j , . . . , x nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
P
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
yr
y
,
1
,
j
,
l
,
…
,
yr
y
,
ny
,
j
,
l
)
;
(g) sending appropriate bits of the x-coordinates and the y-coordinates of the cipher point ( c,0,j , x c,1,j , . . . , x c,nx,j , y c,0,j , y c,1,j , . . . , y c,ny,j ) to the receiving correspondent, together with any other information needed to recover the message point without sacrificing security; (h) generating a new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p for encryption of a following block of the message data bit string using at least one random number generators and the set of keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l for l=1, . . . N p ; and (i) returning to step c) to encrypt the following block of the message data bit string.
The receiving correspondent then performs the following steps: (j) embedding the j th block of the shared secret message bit strings k x,0,j,l , k x,1,j,l , . . . , k x,nx,j,l and k y,1,j,l , . . . , k k,ny,j,l into the elliptic curve key points (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ), wherein the set of elliptic curve key points satisfy the elliptic polynomial, which may be received from the sending correspondent; (k) calculating the set of elliptic curve key points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ) as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+k r,j,l (xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l )+ k r,j,l ( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l )+ k s,j,l ( xk x,0,j,l , xr x,1,j,l , . . . , xk x,nx,h,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
based upon the shared secret keys k r,j,i and k s,j,i ; (l) calculating the message point of the j th data block (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ) as:
(
x
c
,
0
,
j
,
x
c
,
1
,
j
,
…
,
x
c
,
nx
,
j
,
y
c
,
0
,
j
,
y
c
,
1
,
j
,
…
,
y
c
,
ny
,
j
)
=
+
(
x
m
,
0
,
j
,
x
m
,
1
,
j
,
…
,
x
m
,
nx
,
j
,
y
m
,
0
,
j
,
y
m
,
1
,
j
,
…
,
y
m
,
ny
,
j
)
+
∑
l
=
1
N
P
(
xr
x
,
0
,
j
,
l
,
xr
x
,
1
,
j
,
l
,
…
,
xr
x
,
nx
,
j
,
l
,
yr
y
,
0
,
j
,
l
,
yr
y
,
1
,
j
,
l
,
…
,
yr
y
,
ny
,
j
,
l
)
together with any other information sent by the sending correspondent needed to recover the message point without sacrificing security; (m) recovering the secret messages bit string for the j th block from the appropriate x-coordinates and y-coordinates of the message point (x m,0,j , x m,1,j , . . . , x m,nx,j , y m,0,j , y m,1,j , . . . , y m,ny,j ), and any information received from the sending correspondent; (n) generating the new set of shared secret keys k r,j+1,i , k s,j+1,i , k x,i,j+1,l , k y,i,j+1,l for l=1, . . . N p which are needed for the encryption of the next block of the message data bit string, using at least one random number generator and the keys k r,j,i , k s,j,i , k x,i,j,l , k y,i,j,l ; and (o) returning to step (j) to encrypt the following block of the message data bit string.
A special case of the above method is when k r,j,l =k s,j,l =1 for l=1, . . . , N P . In this case, no scalar multiplication is needed to calculate the points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P , which can be computed using one point addition:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+( xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) (xk x,0,j,l , xk x,1,j,l , . . . , xk x,na,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ).
An alternative embodiment of the method includes using a non-linear process in generating the points: (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P , with the keys k r,j,i =k s,j,l =1, and then using the points: (xr x,0,j−1,l , xr x,1,j−1,l , . . . , xr x,nx,j−1,l , yr y,0,j−1,l , yr y,1,j−1,l , . . . , yr y,ny,j−1,l ) and (xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l ) for l=1, . . . , N P . (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P , are computed as:
( xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l )=+ k r,j,l ( xr′ x,0,j−1,l , xr′ x,1,j−1,l , . . . , xr′ x,nx,j−1,l , yr′ y,0,j−1,l , yr′ y,1,j−1,l , . . . , yr′ y,ny,j−1,l )+ k s,j,l ( xk x,0,j,l , xk x,1,j,l , . . . , xk x,nx,j,l , yk y,0,j,l , yk y,1,j,l , . . . , yk y,ny,j,l )
where,
xr″ x,i,j,l =ƒ x ( xr x,i,j−1,l , yr z,i,j−1,l |l= 1, . . . , N P ) and
yr″ x,i,j,l =ƒ y ( xr x,i,j−1,l , yr z,i,j−1,l |l= 1, . . . , N P ),
and if yr″ y,0,j−1,l is quadratic residue, then:
( xr′ x,0,j−1,l , xr′ x,1,j−1,l , . . . , xr′ x,nx,j−1,l , yr′ y,0,j−1,l , yr′ y,1,j−1,l , . . . , yr′ y,ny,j−1,l )=( xr″ x,0,j−1,l , xr″ x,1,j−1,l , . . . , xr″ x,nx,j−1,l , yr″ y,0,j−1,l , yr″ y,1,j−1,l , . . . , yr″ y,ny,j−1,l )
otherwise:
(
xr
x
,
0
,
j
-
1
,
l
′
,
xr
x
,
1
,
j
-
1
,
l
′
,
…
,
xr
x
,
nx
,
j
-
1
,
l
′
,
yr
y
,
0
,
j
-
1
,
l
′
,
yr
y
,
1
,
j
-
1
,
l
′
,
…
,
yr
y
,
ny
,
j
-
1
,
l
′
)
=
(
gxr
x
,
0
,
j
-
1
,
l
″
,
gxr
x
,
1
,
j
-
1
,
l
″
,
…
,
gxr
x
,
nx
,
j
-
1
,
l
″
,
g
3
2
yr
y
,
0
,
j
-
1
,
l
″
,
g
3
2
yr
y
,
1
,
j
-
1
,
l
″
,
…
,
g
3
2
yr
y
,
ny
,
j
-
1
,
l
″
)
where ƒ x (.) and ƒ y (.) are any non-linear functions.
Both the transmitting correspondent and the receiving correspondent perform the same non-linear function in generating the points (xr x,0,j,l , xr x,1,j,l , . . . , xr x,nx,j,l , yr y,0,j,l , yr y,1,j,l , . . . , yr y,ny,j,l ), for l=1, . . . , N P .
As noted above, the methods include data embedding. In order to embed a message bit string into a point (x, √{square root over (α)}y) which satisfies either an elliptic curve equation y 2 =x 3 ax+b or its twist, α y 2 =x 3 +ax+b, the message bit string is first divided into N-bit strings and the i th block is denoted as m i . Following this, the value of the bit string of m i is assigned to x m i , and the values of x m i are substituted and the value of t m i is computed using t m i =x m i 3 +ax m i +b.
If t m i is quadratic residue, then y m i =√{square root over (t m i )} and the point is given as (x m i , y m i ). However, if t m i is non-quadratic residue, then
y
m
i
=
t
m
i
α
_
and the point is given as (x m i , √{square root over ({tilde over (α)}y m i ). The message point is then denoted as (x m i , √{square root over (α m i )}y m i ), where the point is on the elliptic curve if α m i =1, and the point is on the twist if α m i = α .
Alternatively, data embedding may be performed as follows: (a) dividing the message bit string into (nx+ny+1) bit-strings m x,0 , m x,1 , . . . , m x,nx , m y,1 , . . . , m k,ny ; (b) assigning the value of the bit string of m x,0 , m x,1 , . . . , m x,nx to x 0,m , x 1,m , . . . , x nx,m ; (c) assigning the value of the bit string of to m y,1 , . . . , m k,ny to y 1,m , . . . , y ny,m ; (d) computing
T
(
s
-
1
)
=
∑
l
∈
S
nx
b
1
l
x
l
3
+
∑
l
,
k
∈
S
nx
,
l
≠
k
b
2
lk
x
l
2
x
k
;
and (e) performing a Legendre test to determine if T has a square root, wherein if T has a square root, then assigning the square root to y 0 , and if T does not have a square root, then the x-coordinates and y-coordinates of the elliptic polynomial point with the embedded shared secret key bit string are selected as gx i,m and
g
2
3
y
i
,
m
,
respectively, where g is non-quadratic residue in F.
The Legendre Symbol is used to test whether an element of F(p) has a square root or not, i.e., whether an element is quadratic residue or not. The Legendre Symbol and test are as follows. Given an element of a finite field F(p), such as d, the Legendre symbol is defined as (d/p). In order to test whether d is quadratic residue or not, the Legendre symbol, (d/p), is computed such that:
(
d
p
)
=
{
+
1
if
x
is
quadratic
residue
0
if
x
≡
0
mod
F
(
p
)
-
1
otherwise
.
In the above, the password protocols use the scalar multiplication k m (x Pu , y Pu ). It should be noted that, in order to find a collision means, that there are two message bits strings m and m′ such that their integer values k m and k m′ will lead to k m k(x B , y B )≡k m′ k(x B , y B ). This collision implies that integers can be found such that k m k−k m ,k=l*#EC, where #EC is the order of the group (EC,+). This is equivalent to solving the elliptic curve discrete logarithm problem. This also applies to finding a collision for the points on the twist of an elliptic curve, k m k(x TB , √{square root over ( α y TB )=k m′ k(x TB , √{square root over ( α y TB ).
Thus, security of the password protocols depends on the security of the underlying elliptic polynomial cryptography. The security of elliptic polynomial cryptosystems is assessed by both the effect on the solution of the elliptic curve discrete logarithmic problem (ECDLP) and power analysis attacks.
It is well known that the elliptic curve discrete logarithm problem (ECDLP) is apparently intractable for non-singular elliptic curves. The ECDLP problem can be stated as follows: given an elliptic curve defined over F that needs N-bits for the representation of its elements, an elliptic curve point (x P , y P ) ∈ EC, defined in affine coordinates, and a point (x Q , y Q ) 531 EC, defined in affine coordinates, determine the integer k,0≦k≦#F, such that (x Q , y Q )=k(x P , y P ), provided that such an integer exists. In the below, it is assumed that such an integer exists.
The most well known attack used against the ECDLP is the Pollard ρ-method, which has a complexity of O(√{square root over (πK)}/2), where K is the order of the underlying group, and the complexity is measured in terms of an elliptic curve point addition.
Since the underlying cryptographic problems used in the above block cipher chaining methods is the discrete logarithm problem, which is a known difficult mathematical problem, it is expected that the security of the above methods are more secure than prior art ciphers which are not based on such a mathematically difficult problem.
Projective coordinate can also be used by the sending correspondent and the receiving correspondent to embed extra message data bits in the projective coordinate wherein the addition of the corresponding elliptic points is defined in (nx+ny+3) dimensional space where there are (nx+1) x-coordinates, (ny+1) y-coordinates and one projective coordinate.
The equations for the addition rule can be obtained by using the elliptic polynomial equation with (nx+1) x-coordinates and (nx+1) y-coordinates in projective coordinates and substituting a straight line equation to obtain a cubic equation in terms of one of the x-coordinates. This cubic equation can be used to identify the third point of intersection between a straight line and the elliptic polynomial in (nx+ny+3) dimensions given two other intersection points. This third point of intersection is used to identify the sum of the given two points.
It should be understood that the calculations may be performed by any suitable computer system, such as that diagrammatically shown in the sole drawing Figure. Data is entered into system 100 via any suitable type of user interface 116 , and may be stored in memory 112 , which may be any suitable type of computer readable and programmable memory. Calculations are performed by processor 114 , which may be any suitable type of computer processor and may be displayed to the user on display 118 , which may be any suitable type of computer display.
Processor 114 may be associated with, or incorporated into, any suitable type of computing device, for example, a personal computer or a programmable logic controller. The display 118 , the processor 114 , the memory 112 and any associated computer readable recording media are in communication with one another by any suitable type of data bus, as is well known in the art.
Examples of computer-readable recording media include a magnetic recording apparatus, an optical disk, a magneto-optical disk, and/or a semiconductor memory (for example, RAM, ROM, etc.). Examples of magnetic recording apparatus that may be used in addition to memory 112 , or in place of memory 112 , include a hard disk device (HDD), a flexible disk (FD), and a magnetic tape (MT). Examples of the optical disk include a DVD (Digital Versatile Disc), a DVD-RAM, a CD-ROM (Compact Disc-Read Only Memory), and a CD-R (Recordable)/RW.
It is to be understood that the present invention is not limited to the embodiments described above, but encompasses any and all embodiments within the scope of the following claims.
You are contracting for Elliptic polynomial cryptography with secret key embedding
Expert Elliptic polynomial cryptography with secret key embedding
You are commenting for Elliptic polynomial cryptography with secret key embedding