21.4 Elliptic Curves in Characteristic 2

Many applications use elliptic curves mod 2, or elliptic curves defined over the finite fields GF(2n) (these are described in Section 3.11). This is often because mod 2 adapts well to computers. In 1999, NIST recommended 15 elliptic curves for cryptographic uses (see [FIPS 186-2]). Of these, 10 are over finite fields GF(2n).

If we’re working mod 2, the equations for elliptic curves need to be modified slightly. There are many reasons for this. For example, the derivative of y2 is 2yy=0,  since 2 is the same as 0. This means that the tangent lines we compute are vertical, so 2P= for all points P. A more sophisticated explanation is that the curve y2x3+bx+c(mod2) has singularities (points where the partial derivatives with respect to x and y simultaneously vanish).

The equations we need are of the form

E:y2+a1xy+a3y=x3+a2x2+a4x+a6, 

where a1, , a6 are constants. The addition law is slightly more complicated. We still have three points adding to infinity if and only if they lie on a line. Also, the lines through are vertical. But, as we’ll see in the following example, finding P from P is not the same as before.

Example

Let E:y2+yx3+x(mod2). As before, we can list the points on E:

(0, 0), (0, 1), (1, 0), (1, 1), .

Let’s compute (0, 0)+(1, 1). The line through these two points is y=x. Substituting into the equation for E yields x2+xx3+x,  which can rewritten as x2(x+1)0. The roots are x=0, 0, 1(mod2). Therefore, the third point of intersection also has x=0. Since it lies on the line y=x,  it must be (0, 0). (This might be puzzling. What is happening is that the line is tangent to E at (0, 0) and also intersects E in the point (1, 1).) As before, we now have

(0, 0)+(0, 0)+(1, 1)=.

To get (0, 0)+(1, 1) we need to compute (0, 0). This means we need to find P such that P+(0, 0)=. A line through is still a vertical line. In this case, we need one through (0, 0),  so we take x=0. This intersects E in the point P=(0, 1). We conclude that (0, 0)+(0, 1)=. Putting everything together, we see that

(0, 0)+(1, 1)=(0, 1).

In most applications, elliptic curves mod 2 are not large enough. Therefore, elliptic curves over finite fields are used. For an introduction to finite fields, see Section 3.11. However, in the present section, we only need the field GF(4),  which we now describe.

Let

GF(4)={0, 1, ω, ω2}, 

with the following laws:

  1. 0+x=x for all x.

  2. x+x=0 for all x.

  3. 1x=x for all x.

  4. 1+ω=ω2.

  5. Addition and multiplication are commutative and associative, and the distributive law holds: x(y+z)=xy+xz for all x, y, z.

Since

ω3=ωω2=ω(1+ω)=ω+ω2=ω+(1+ω)=1, 

we see that ω2 is the multiplicative inverse of ω. Therefore, every nonzero element of GF(4) has a multiplicative inverse.

Elliptic curves with coefficients in finite fields are treated just like elliptic curves with integer coefficients.

Example

Consider

E:y2+xy=x3+ω, 

where ωGF(4) is as before. Let’s list the points of E with coordinates in GF(4):

x=0y2=ωy=ω2x=1y2+y=1+ω=ω2 no solutionsx=ωy2+ωy=ω2y=1, ω2x=ω2y2+ω2y=1+ω=ω2 no solutionsx=y=.

The points on E are therefore

(0, ω2), (ω, 1), (ω, ω2), .

Let’s compute (0, ω2)+(ω, ω2). The line through these two points is y=ω2. Substitute this into the equation for E:

ω4+ω2x=x3+ω, 

which becomes x3+ω2x=0. This has the roots x=0, ω, ω. The third point of intersection of the line and E is therefore (ω, ω2),  so

(0, ω2)+(ω, ω2)+(ω, ω2)=.

We need (ω, ω2),  namely the point P with P+(ω, ω2)=. The vertical line x=ω intersects E in P=(ω, 1),  so

(0, ω2)+(ω, ω2)=(ω, 1).

For cryptographic purposes, elliptic curves are used over fields GF(2n) with n large, say at least 150.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset