### Abstract

It has recently been reported that the performance of hyperelliptic curve cryptosystems (HECC) is competitive to that of elliptic curve cryptosystems (ECC). However, it is expected that HECC still can be improved due to their mathematically rich structure. We consider here the application of degenerate divisors of HECC to scalar multiplication. We investigate the operations of the degenerate divisors in the Harley algorithm and the Cantor algorithm of genus 2. The timings of these operations are reported. We then present a novel efficient scalar multiplication method using the degenerate divisors. This method is applicable to cryptosystems with fixed base point, e.g., ElGamal-type encryption, sender of Diffie-Hellman, and DSA. Using a Xeon processor, we found that the double-and-add-always method using the degenerate base point can achieve about a 20% increase in speed for a 160-bit HECC. However, we mounted an timing attack using the time difference to designate the degenerate divisors. The attack assumes that the secret key is fixed and the base point can be freely chosen by the attacker. Therefore, the attack is applicable to ElGamal-type decryption and single-pass Diffie-Hellman - SSL using a hyperelliptic curve could be vulnerable to the proposed attack. Our experimental results show that one bit of the secret key for a 160-bit HECC can be recovered by calling the decryption oracle 500 times.

Original language | English |
---|---|

Pages (from-to) | 345-359 |

Number of pages | 15 |

Journal | Lecture Notes in Computer Science |

Volume | 3325 |

Publication status | Published - Sep 1 2005 |

Event | 5th International Workshop on Information Security Applications, WISA 2004 - Jeju Island, Korea, Republic of Duration: Aug 23 2004 → Aug 25 2004 |

### All Science Journal Classification (ASJC) codes

- Theoretical Computer Science
- Computer Science(all)

## Fingerprint Dive into the research topics of 'Novel efficient implementations of hyperelliptic curve cryptosystems using degenerate divisors'. Together they form a unique fingerprint.

## Cite this

*Lecture Notes in Computer Science*,

*3325*, 345-359.