3
$\begingroup$

The encryption in Paillier cryptosystem is like this according to Wikipedia:

  1. Let $m$ be a message to be encrypted where $m \in \mathbb{Z}_n$
  2. Select random $r$ where $r \in \mathbb{Z}_n^*$
  3. Compute ciphertext as: $c = g^m \cdot r^n \bmod n^2 $

So just calculating the inverse of ciphertext $c$ (i.e enc($m$)) will not work, because enc($m$)*enc($-m$) is equal to $r^{2n} \bmod n^2$, not $1 \bmod n^2$.

$\endgroup$
  • $\begingroup$ I tried to convert your question to proper formatting. Please edit if I mangled something. $\endgroup$ – otus Oct 25 '15 at 20:17
4
$\begingroup$

Actually, computing the inverse modulo $n^2$ using (say) the Extended Euclidean method is exactly what you do. If $c = g^m r^n \bmod n^2$ is an encrypted version of $m$, then $c^{-1} = (g^m r^n)^{-1} = g^{-m} (r^{-1})^n$ is a representation of $-m$ (because if $r \in \mathbb{Z}^*_n$, so is $r^{-1}$)

That $enc(m) * enc(-m)$ isn't precisely 1 isn't relevant; however if we use different randomness $r$ and $s$ during the separate encryptions (as we're supposed to), it'll actually be $r^ns^n = (rs)^n$. However, there is lots of encrypted representations of 0; while 1 is such a valid representation, $(rs)^n$ is as well.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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