6
$\begingroup$

I have coded an implementation of elliptic curves in order to apply some of the ECC algorithms. However, in most of them, Alice needs to choose a point P on a given curve. What is the general procedure for selecting such a point?

Given a small example such as $y^2 = x^3 + x + 1$ over $F_{25}$, is there an algorithm to generate a random point on the curve? In my implementation points on this field are represented by polynomials, if that is relevant.

$\endgroup$
0

1 Answer 1

12
$\begingroup$

is there an algorithm to generate a random point on the curve?

By having a good random source like /dev/urandom you can generate a random point $R$ on a curve at least by two methods;

  1. By using random scalar and scalar multiplication

    1. Choose a generator point $P$ on the curve
    2. Get random integer between $0 < k < \text{Order of the Group}$
    3. Calculate $R =[k]P$ by scalar multiplication, prefably executing by double-and-add algorithm.
  2. Using random element with sample rejection on the curve equation.

    1. Choose a random element $x \in \mathbb{Z_{25}}$.

    2. Check the equation $y^2 = x^3 + x + 1$ has a solution, i.e. it is a quadratic residue.

      If there is no solution return to 1. step ( reject the sample)

      Else, we have two solutions, $y$ and $-y$ (or one $y=-y$). These values can be found using the Tonelli-Shank algorithm and its generalizations.

    3. Now use the random source to select $\bar{y}$ as $y$ or $-y$, or toss a coin.

    4. Form the random point on the curve as $R=(x,\bar{y})$

    Note that, the point at infinity cannot be expressed in affine-coordinates without some tricks. Therefore, this method cannot return the point at infinity. If the point at infinity is also required on the random selection, one can select it with an initial random. Select it within $\frac{1}{\text{Order of The Group}}$ probability. Then use the above.

  3. If you are using SageMath then it has a function

    random_element() and the explanation is given as

    Return a random point on this elliptic curve, uniformly chosen among all rational points.

    SageMath random_element() function uses the 2. method with an addition that it can also return the point at infinity, $\mathcal{O}$

In my implementation points on this field are represented by polynomials, if that is relevant.

The $\mathbb{F}_{25}$ is an Extension Field and it is usual to represent the elements of the field by the polynomial representation which is very advantageous on operations. As a result, the coordinates of points have polynomial representation.

From comments of OP

I used Sage to calculate the order of each point: (sagecell.sagemath.org/…) and it can be seen that even though the cardinality of the group is 32 the biggest order is 16. Is it just a matter of choosing an EC with prime cardinality?

The SageMath construction is not correct;

E = EllipticCurve(GF(25, 'x'),1, 1) provides this curve $y^2 = x^3 + x + 2$ not $y^2 = x^3 + x + 1$

One can construct as below;

    E = EllipticCurve(GF(25, 'x'), [1, 1])
    print(E)
    print(E.abelian_group())
    print ( E.cardinality())
    for i in E.points():
        print (i.order())

The construction EllipticCurve(Finite Field,[a,b]) is for the standard Weierstrass from $y^2 = x^3+ a x + b$ where $a,b \in F$ in which the curve defined. You only need to provide $a,b$, and $F$. The set of points commonly represented as $E_{a,b}(F)$

The abelian_group() functionality prints the abelian group structure of the group of points on this elliptic curve. The result is

$$\mathbb{Z_3} \oplus \mathbb{Z_9}$$

One should not call this function for a large group. There is a theory behind this and this can be found in a book such as Washington's book, however, the reference can go back to J. W. S. Cassels, 1966, Diophantine Equations with Special Reference To Elliptic Curves

Theorem: Let $E$ ben elliptic curve group over the finite field $\mathbb{F_q}$. Then $$E(\mathbb{F_q}) \simeq \mathbb{Z_n} \text{ or } \mathbb{Z_{n_1}} \oplus \mathbb{Z_{n_2}}$$ for some integer $n \geq 1$, or for some integers $n_1,n_2 \geq 1$ with $n_1|n_2$.

$\endgroup$
2
  • 1
    $\begingroup$ nice answer but @srb is not working in $\mathbb{Z}_{25}$ but in $\mathbb{F}_{5^2}$ and x-coordinates are not integers $\endgroup$
    – Ruggero
    Commented Oct 23, 2020 at 12:39
  • $\begingroup$ @Ruggero yes. Thanks. I've corrected those parts with the element word. $\endgroup$
    – kelalaka
    Commented Oct 23, 2020 at 12:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.