Cryptosystems That Resist Quantum Fourier Sampling Attacks Hang - - PowerPoint PPT Presentation

cryptosystems that resist
SMART_READER_LITE
LIVE PREVIEW

Cryptosystems That Resist Quantum Fourier Sampling Attacks Hang - - PowerPoint PPT Presentation

McEliece and Niederreiter Cryptosystems That Resist Quantum Fourier Sampling Attacks Hang Dinh Indiana University South Bend joint work with Cristopher Moore Alexander Russell University of New Mexico University of Connecticut Post-quantum


slide-1
SLIDE 1

McEliece and Niederreiter Cryptosystems That Resist Quantum Fourier Sampling Attacks

Hang Dinh

Indiana University South Bend joint work with

Alexander Russell

University of Connecticut

Cristopher Moore

University of New Mexico

slide-2
SLIDE 2

Post-quantum cryptography

  • Shor’s quantum algorithms for Factoring and

Discrete Logarithm break RSA, ElGamal, elliptic curve cryptography...

  • Are there “post-quantum” cryptosystems?

 cryptosystems we can carry out with classical computers

  • [unlike quantum cryptosystems, which require quantum facility]

 which will remain secure even if and when quantum computers are built.

Hang Dinh - IU South Bend

slide-3
SLIDE 3

Post-quantum cryptography

  • Candidates for post-quantum cryptosystems:

 lattice-based  code-based (the McEliece system and its relatives)  hash-based  multivariate  secret-key cryptography

  • Bernstein, 2009:

 These systems are believed to resist quantum computers.

 “Nobody has figured out a way to apply Shor’s algorithm

to any of these systems.”

Hang Dinh - IU South Bend

slide-4
SLIDE 4

We show that

some McEliece and Niederreiter cryptosystems resist the natural analog of Shor’s quantum attack.

Hang Dinh - IU South Bend

slide-5
SLIDE 5

How Shor’s algorithm works

Breaking RSA private key Integer Factorization Hidden Subgroup Problem

  • ver a cyclic group ZN

Quantum Fourier Sampling

  • ver ZN

Discrete Logarithm Breaking ElGamal, elliptic curve cryptography Hidden Subgroup Problem

  • ver an abelian group ZN×ZN

Quantum Fourier Sampling

  • ver ZN×ZN

Hang Dinh - IU South Bend

slide-6
SLIDE 6

Hidden Subgroup Problem (HSP)

  • HSP over a finite group G:

 Input: function f : G {,, …} that distinguishes the left cosets of an unknown subgroup H <G  Output: H

  • Notable reductions to nonabelian HSP:

 Unique Shortest Vector Problem  HSP over Dn [Regev’04]  Graph Isomorphism  HSP over Sn with |H|≤2 H g2H g3H … gkH

Hang Dinh - IU South Bend

slide-7
SLIDE 7

Quantum Fourier Sampling (QFS)

QFS over G to find hidden subgroup H:

Uniform superposition over G

฀  gH

 ij ,i, j

,i, j

Use input function f Quantum Fourier transform Measure ρ ρ column j weak strong ρ block matrix corresponding to irreducible representation ρ of G uniform superposition

  • ver coset gH

random coset state gH

slide-8
SLIDE 8

McEliece/Niederreiter Cryptosystems

  • Scramble M’s rows

Permute M’s columns

Hang Dinh - IU South Bend

slide-9
SLIDE 9

McEliece/Niederreiter Cryptosystems

McEliece system Niederreiter system

Hang Dinh - IU South Bend

  • F𝑟 = F𝑟𝑚 𝑚 = 1
  • M is a generator matrix of

an 𝑜, 𝑙 -code over Fq. Equivalent to the McEliece system using C, if

dim 𝐷 = 𝑜 − 𝑚𝑙 .

  • Originally used classical

binary Goppa codes (q=2)

  • F𝑟 F𝑟𝑚 𝑚 ≥ 1
  • M is a parity check matrix of

an 𝑜, 𝑙′ -code C over Fq.

  • Equivalent to the McEliece

system using C, if 𝑙′ = 𝑜 − 𝑚𝑙.

  • Originally used rational

Goppa codes (GRS codes)

slide-10
SLIDE 10

Security of McEliece and Niederreiter Systems

  • Two basic types of attacks

 Decoding attacks [previous talk]  Attacks on private key [this talk]

  • Recover S, M, P from M*
  • Security against known classical attacks

 Still secure if using classical Goppa codes [EOS’07]  Broken if using rational Goppa codes (Ouch!)

  • Sidelnokov & Shestakov’s attack factors SMP into S and MP.

Hang Dinh - IU South Bend

slide-11
SLIDE 11

McEliece/Niederreiter’s security reduces to HSP

Scrambler-Permutation Problem

  • Given: M and M* = SMP for some (S, P)  GLk(Fq) ×Sn
  • Find: S and P

~

Can this HSP be solved by strong QFS?

Hang Dinh - IU South Bend

slide-12
SLIDE 12
  • Strong QFS yields negligible information about

hidden (S, P) if M is good, meaning

 M has column rank 𝑠 ≥ 𝑙 − 𝑝 𝑜 /𝑚,  𝐵𝑣𝑢 𝑁 ≤ 𝑓𝑝 𝑜 , and  Minimal degree of Aut(M) is (𝑜).

  • Next question:

 Are there matrices M satisfying the conditions above?

Our Answer (1)

the minimal number of points moved by a non-identity permutation in Aut(M)

Hang Dinh - IU South Bend

slide-13
SLIDE 13

Our Answer (2)

  • 1

1 2 2 1 1 1 2 2 1 1 2 1

            

   k n n k k n n n

v v v v v v v v v S M             

 

distinct. are s ' }, { F }, { F , F GL

i q i q i q k

l l l

v S        

Hang Dinh - IU South Bend

slide-14
SLIDE 14

Conclusion

  • The following cryptosystems resist the natural

analog of Shor’s QFS attack:

 McEliece systems using rational Goppa codes  Niederreiter systems using classical Goppa codes.  In general, any McEliece/Niederreiter system using linear codes with good generator/parity check matrices. Warning: This neither rules out other quantum (or classical) attacks nor violates a natural hardness assumption.

Hang Dinh - IU South Bend

slide-15
SLIDE 15

Conclusion (Moral)

McEliece RSA Quantum Fourier Sampling ElGamal Niederreiter

need new ideas

Hang Dinh - IU South Bend

slide-16
SLIDE 16

Open Questions

  • What are other linear codes that possess good

generator/parity check matrices?

  • Can these cryptosystems resist stronger quantum

attacks, e.g., multiple-register QFS attacks?

 Hallgren et al., 2006: subgroups of order 2 require highly-entangled measurements of many coset states.  Does this hold for subgroups of order > 2?

Hang Dinh - IU South Bend

slide-17
SLIDE 17

Questions?

Hang Dinh - IU South Bend

  • Thank you all for staying till the last minute!
slide-18
SLIDE 18
  • In case of Niederreiter systems using a classical

q-ary Goppa code C, we need

  • Typically, 𝑜 = 𝑟𝑚, then we only need 𝑙2 ≤ 0.2𝑜𝑚,

 which implies C must have large dimension:

Parameters

 

n

  • l

n k

q n q e and

3 2 .

2

 

2 / 3

2 . dim l n n kl n C    

Hang Dinh - IU South Bend