In the IETF RFC draft named "SM2 Digital Signature Algorithm" a signature algorithm is specified. The RFC does however not mention why this signature algorithm has been defined. Nor does it specify what the advantages of this scheme are over ECDSA. It seems that SM2 is the preferred EC signature format in China.

So the two related questions are:

  1. Why was SM2 defined?
  2. What are the (operational) differences compared to ECDSA?

Note that there seems to be a set of SM2 Elliptic Curve domain parameters defined as well.

According to [SM2 Algorithms Parameters] linked in the draft (and found on Wayback machine), here are the domain parameters:

y^2 = x^3 + ax + b

b = 28E9FA9E 9D9F5E34 4D5A9E4B CF6509A7 F39789F5 15AB8F92 DDBCBD41 4D940E93
Gx = 32C4AE2C 1F198119 5F990446 6A39C994 8FE30BBF F2660BE1 715A4589 334C74C7
Gy = BC3736A2 F4F6779C 59BDCEE3 6B692153 D0A9877C C62A4740 02DF32E5 2139F0A0
  • $\begingroup$ The IETF draft is a translation from a Chinese standard document, it seems. It also defines a key exchange protocol and an encryption algorithm. (I have no idea why it was defined, though.) $\endgroup$ Aug 25, 2013 at 12:32
  • 5
    $\begingroup$ Why? I guess because the chinese guys simply want to have their own standard (and I think being superior to existing schemes is not an issue ;) $\endgroup$
    – DrLecter
    Jul 31, 2014 at 17:30
  • 1
    $\begingroup$ Part 1 (out of 4) of the 2010 Chinese standard defining SM2 algorithms is published here, and some of it is understandable when automatically translated. $\;$ Could no find part 2 on SM2 signature. $\;$ Jing Xu & Dengguo Feng's Comments on the SM2 Key Exchange Protocol is a claimed attack on part 3. $\endgroup$
    – fgrieu
    Aug 1, 2014 at 3:51
  • 1
    $\begingroup$ Part 2 (out of 4) of the 2010 Chinese standard defining SM2 algorithms is published, in IETF $\endgroup$
    – QIngluan
    Dec 7, 2015 at 14:39

1 Answer 1


The SM2 digital signature algorithm. Fix a curve $E/k$ over a field $\mathbb F_p$. Fix a standard base point $G$ of prime order $n$ on $E/\mathbb F_p$. Fix a hash function $H\colon \{0,1\}^* \to \mathbb Z/n\mathbb Z$.

  • A public key is the encoding of a point $P \in E/\mathbb F_p$ together with an string $Z$ identifying a principal.

  • A signature on a message $m$ under the public key $P$ with identifier $Z$ is is a pair of integers $(r, s)$ in $[1, n - 1]$ such that $r + s \ne 0$ and $$r \equiv (H(Z \mathbin\| m) + x([s]G + [r + s]P)) \pmod n.$$

  • The signer knows the secret scalar $d$, the private key, such that $P = [d]G$. To make a signature, the signer picks a scalar $k$ uniformly at random, and computes

    \begin{align} r &\equiv (H(Z \mathbin\| m) + x([k]G)) \pmod n, \\ s &\equiv (k - r d)/(1 + d) \pmod n, \end{align}

    rejecting and starting over if $r \equiv 0$, $r + k \equiv 0$, or $s \equiv 0 \pmod n$.

    This is a valid signature because

    \begin{align} [s]G + [r + s]P &= [s]G + [(r + s) d]G \\ &= [s + s d + r d] G \\ &= [(1 + d) s + r d] G \\ &= [(1 + d)(k - r d)/(1 + d) + r d]G \\ &= [k - r d + r d]G \\ &= [k]G. \end{align}

Why? There has been relatively little analysis of the SM2 signature scheme that I can find in the anglophone cryptography literature beyond some side channel attacks. The most salient aspect of it to my eyes is that it is just different from various other national pride digital signature schemes.

The SM2 signature scheme has a small advantage over ECDSA (US) and ECGDSA (Germany) for verifiers: the verifier need not compute any inverses modulo $n$. Otherwise the signer's costs are similar to the others. Security analysis of ECDSA is extraordinarily complicated and unsatisfying, in contrast to Schnorr; I expect the same is true of the others.

My guess about why SM2 was chosen is precisely to be different from all the others, for the sake of nationalism—they are all worse in performance and confidence than the much simpler inversion-free Schnorr-based EdDSA.


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.