Group Action Systems Group Action Systems : : a Mathematical tool - - PowerPoint PPT Presentation

group action systems
SMART_READER_LITE
LIVE PREVIEW

Group Action Systems Group Action Systems : : a Mathematical tool - - PowerPoint PPT Presentation

Group Action Systems Group Action Systems : : a Mathematical tool for deriving a Mathematical tool for deriving Provable Secure Cryptographic Schemes Provable Secure Cryptographic Schemes Mara Isabel Gonzlez Vasco Universidad Rey Juan


slide-1
SLIDE 1

Group Action Systems Group Action Systems:

: a Mathematical tool for deriving a Mathematical tool for deriving Provable Secure Cryptographic Schemes Provable Secure Cryptographic Schemes

María Isabel González Vasco

Universidad Rey Juan Carlos

slide-2
SLIDE 2

Group Action Systems Group Action Systems:

: a Mathematical tool for deriving a Mathematical tool for deriving Provable Secure Cryptographic Schemes Provable Secure Cryptographic Schemes

Joint Joint works

  • rks with

with J. L. Villar (UPC) and

  • J. L. Villar (UPC) and R. Steinwandt

. Steinwandt (FAU) (FAU)

slide-3
SLIDE 3

Overview Overview

Introduction

M.I. González-Vasco, Bochum 05

slide-4
SLIDE 4

Overview Overview

Introduction Some basics about PHFs

Definitions Basic Results Cryptographic Applications

M.I. González-Vasco, Bochum 05

slide-5
SLIDE 5

Overview Overview

Introduction Some basics about PHFs

Definitions Basic Results Cryptographic Applications

Group Action Based PHFs

Group Action Systems Useful AcPHFs. Diversity.

M.I. González-Vasco, Bochum 05

slide-6
SLIDE 6

Overview Overview

Introduction Some basics about PHFs

Definitions Basic Results Cryptographic Applications

Group Action Based PHFs

Group Action Systems Useful AcPHFs. Diversity.

Examples

M.I. González-Vasco, Bochum 05

slide-7
SLIDE 7

Overview Overview

Introduction Some basics about PHFs

Definitions Basic Results Cryptographic Applications

Group Action Based PHFs

Group Action Systems Useful AcPHFs. Diversity

Examples Final Remarks

M.I. González-Vasco, Bochum 05

slide-8
SLIDE 8

Introduction Introduction

Motivation: finding new suitable mathematical

primitives for cryptographic designs.

M.I. González-Vasco, Bochum 05

slide-9
SLIDE 9

Introduction Introduction

Motivation: finding new suitable mathematical

primitives for cryptographic designs.

Fact: work in that direction hardly exploits the

constructions and theoretical frameworks available from number-theoretical cryptography.

M.I. González-Vasco, Bochum 05

slide-10
SLIDE 10

Introduction Introduction

Motivation: finding new suitable mathematical

primitives for cryptographic designs.

Fact: work in that direction hardly exploits the

constructions and theoretical frameworks available from number-theoretical cryptography.

Our Goal: adapt the existing theory of Universal

Projective Hash Functions to allow constructions arising in different areas of mathematics .

M.I. González-Vasco, Bochum 05

slide-11
SLIDE 11

Some Some basics basics about bout PHFs PHFs

slide-12
SLIDE 12

Definitions Definitions

Let X, Π, S be non-empty sets, L⊆ X, and K a finite index set. Consider H:={ Hk : X a Π }k∈ K and α : K a S.

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

slide-13
SLIDE 13

Definitions Definitions

Let X, Π, S be non-empty sets, L⊆ X, and K a finite index set. Consider H:={ Hk : X a Π }k∈ K and α : K a S. Then the tuple H = (H, K, X, L, Π, S, α) is a projective hash family

  • PHF - for (X, L) provided that

α (k) ≈ Hk|L() (i.e., ∀ x∈ L, k1, k2 ∈ K, α(k1) = α(k2) ⇒ Hk1(x) = Hk2(x) ).

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

slide-14
SLIDE 14

L X Π

Hk(x) x

Given Given only nly the he projection projection α(k)… (k)…

X*

Hk(x*)

α(k)

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

slide-15
SLIDE 15

L X Π

Hk(x) x

…it …it could

  • uld be hard

be hard to to compute H compute Hk outside

  • utside L

X*

Hk(x*)

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

slide-16
SLIDE 16

Definitions Definitions

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

Moreover, we say that H = (H, K, X, L , Π, S, α) is

ε-universal :⇔ ∀s ∈ S, x ∈ X\L, π ∈ Π

P[Hk(x) = π / α (k)=s ] ≤ ε ;

slide-17
SLIDE 17

Definitions Definitions

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

Moreover, we say that H = (H, K, X, L , Π, S, α) is

ε-universal :⇔ ∀s ∈ S, x ∈ X\L, π ∈ Π

P[Hk(x) = π / α (k)=s ] ≤ ε;

ε-universal2:⇔ ∀ s ∈ S, x ∈ X\L, x* ∈ X\(LU{x}), π, π* ∈ Π

P[Hk(x) = π / Hk(x*) = π*, α (k)=s ] ≤ ε;

slide-18
SLIDE 18

Definitions Definitions

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

Moreover, we say that H = (H, K, X, L , Π, S, α) is

ε-universal :⇔ ∀s ∈ S, x ∈ X\L, π ∈ Π

P[Hk(x) = π / α (k)=s ] ≤ ε ;

  • ε-universal2:⇔ ∀ s ∈ S, x ∈ X\L, x* ∈ X\(LU{x}), π, π* ∈ Π

P[Hk(x) = π / Hk(x*) = π*, α (k)=s ] ≤ ε ;

ε- smooth : ⇔ (x, α(k), Hk(x)) and (x, α(k), π) are

ε-close for k ∈ K, x ∈ X\L and π ∈ Π chosen uniformly at random ;

slide-19
SLIDE 19

Definitions Definitions

M.I. González-Vasco, Bochum 05 Some Basics About PHFs

Moreover, we say that H = (H, K, X, L , Π, S, α) is

ε-universal :⇔ ∀s ∈ S, x ∈ X\L, π ∈ Π

P[Hk(x) = π / α (k)=s ] ≤ ε;

  • ε-universal2:⇔ ∀ s ∈ S, x ∈ X\L, x* ∈ X\(LU{x}), π, π* ∈ Π

P[Hk(x) = π / Hk(x*) = π*, α (k)=s ] ≤ ε;

ε- smooth : ⇔ (x, α(k), Hk(x)) and (x, α(k), π) are

ε-close for k ∈ K, x ∈ X\L and π ∈ Π chosen uniformly at random;

Strongly universal2≈ worst case smoothness.

slide-20
SLIDE 20

Basic Results Basic Results

Ways of “upgrading” the weaker types of PHFs to achieve

more robust types:

Universal to universal2 - Cramer and Shoup, [EUROCRYPT 2002] Universal to smooth - Cramer and Shoup, [EUROCRYPT 2002] Universal2 to strongly universal2

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-21
SLIDE 21

Basic Results Basic Results

Ways of “upgrading” the weaker types of PHFs to achieve

more robust types:

Universal to universal2 - Cramer and Shoup, [EUROCRYPT 2002] Universal to smooth - Cramer and Shoup, [EUROCRYPT 2002] Universal2 to strongly universal2

Methods for constructing cryptographically useful PHFs

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-22
SLIDE 22

Cryptographic Cryptographic Applications pplications

Cramer and Shoup [EUROCRYPT 2002]

  • IND-CCA Encryption Scheme in the standard model

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-23
SLIDE 23

Cryptographic Cryptographic Applications pplications

Cramer and Shoup [EUROCRYPT 2002]

  • IND-CCA Encryption Scheme in the standard model

Kurosawa and Desmedt [CRYPO 2004]

Hybrid encryption scheme

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-24
SLIDE 24

Cryptographic Cryptographic Applications pplications

Cramer and Shoup [EUROCRYPT 2002]

  • IND-CCA Encryption Scheme in the standard model

Kurosawa and Desmedt [CRYPO 2004]

Hybrid encryption scheme

Genaro and Lindell [EUROCRYPT 2003]

Password based authenticated key exchange

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-25
SLIDE 25

Cryptographic Cryptographic Applications pplications

Cramer and Shoup [EUROCRYPT 2002]

  • IND-CCA Encryption Scheme in the standard model

Kurosawa and Desmedt [CRYPO 2004]

Hybrid encryption scheme

Genaro and Lindell [EUROCRYPT 2003]

Password based authenticated key exchange

Kalai [EUROCRYPT 2005]

2-out-of-1 oblivious transfer protocol.

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-26
SLIDE 26

Cryptographic Applications Cryptographic Applications

Cramer and Shoup [EUROCRYPT 2002]

Π is the message space

  • k is kept secret, α(k) and x are public
  • m ∈ Π is encrypted using Hk(x) as a one time pad, for x ∈ L, i.e.,

E(α(k)) (m) = (x, Hk(x)⊕ m)

IND-CCA security is achieved by appending a proof of integrity

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-27
SLIDE 27

Cryptographic Cryptographic Applications pplications

Kalai [EUROCRYPT 2005]

Sender’s (B) input: two strings γ0, γ1, Receiver’s (A) input: choice bit b. Goal: A learns γ b, but nothing about γb-1 . B learns nothing about b.

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-28
SLIDE 28

Cryptographic Cryptographic Applications pplications

Kalai [EUROCRYPT 2005]

Sender’s (B) input: two strings γ0, γ1. Receiver’s (A) input: choice bit b. Goal: A learns γ b, but nothing about γ1-b . B learns nothing about b.

A chooses xb ∈ L and x1-b ∈ X\L and sends (X, x0, x1) to B;

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-29
SLIDE 29

Cryptographic Cryptographic Applications pplications

Kalai [EUROCRYPT 2005]

Sender’s (B) input: two strings γ0, γ1. Receiver’s (A) input: choice bit b. Goal: A learns γ b, but nothing about γ1-b . B learns nothing about b.

A chooses xb ∈ L and x1-b ∈ X\L and sends (X, x0, x1) to B;

  • B chooses independently two random keys k0, k1 and sends

α(k0), α(k1), y0 = γ0 ⊕ Hk0(x0) and y1 = γ1 ⊕ Hk1(x1);

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-30
SLIDE 30

Cryptographic Cryptographic Applications pplications

Kalai [EUROCRYPT 2005]

Sender’s (B) input: two strings γ0, γ1. Receiver’s (A) input: choice bit b. Goal: A learns γ b, but nothing about γ1-b . B learns nothing about b.

A chooses xb ∈ L and x1-b ∈ X\L and sends (X, x0, x1) to B;

  • B chooses independently two random keys k0, k1 and sends

α(k0), α(k1), y0 = γ0 ⊕ Hk0(x0) and y1 = γ1 ⊕ Hk1(x1);

A retrieves γb by computing yb⊕ Hkb(xb) using the projection key

α(kb). Note that as x1-b ∈ X\L, α(k1-b) does not give enough information for computing Hk1-b outside L.

Some Basics About PHFs M.I. González-Vasco, Bochum 05

slide-31
SLIDE 31

Group Group Action ction Based ased Projective Projective Hash Hash Families Families

slide-32
SLIDE 32

Group Group Systems ystems

“Atoms” from which PHFs are derived for Cramer-Shoup

Encryption Scheme [EUROCRYPT 2002].

Group Action Based PHFs M.I. González-Vasco, Bochum 05

slide-33
SLIDE 33

Group Group Systems ystems

“Atoms” from which PHFs are derived for Cramer-Shoup

Encryption Scheme [EUROCRYPT 2002].

A group system is a tuple (H, X, L, Π), where X and Π are finite

abelian groups, L ≤ X, H ≤ Hom(X, Π).

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-34
SLIDE 34

Group Group Systems ystems

“Atoms” from which PHFs are derived for Cramer-Shoup

Encryption Scheme [EUROCRYPT 2002].

A group system is a tuple (H, X, L, Π), where X and Π are finite

abelian groups, L ≤ X, H ≤ Hom(X, Π).

To derive a PHF, one must specify the action of H on L in

terms of a set {g1,…,gd} of generators for L, i.e. α(k) = (Hk(g1), …, Hk(gd)).

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-35
SLIDE 35

Group Group Systems ystems

“Atoms” from which PHFs are derived for Cramer and

Shoup’s Encryption Scheme [EUROCRYPT 2002].

A group system is a tuple (H, X, L, Π), where X and Π are finite

abelian groups, L ≤ X, H ≤ Hom(X, Π).

To derive a PHF, one must specify the action of H on L in

terms of a set {g1,…,gl} of generators for L, i.e. α(k) = (Hk(g1), …, Hk(gl)).

Using group systems, they derived instances of their

encryption scheme based on the DDH problem and the Decision Composite Residuosity assumption.

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-36
SLIDE 36

Group Group Action ction Systems ystems (I) (I)

Let X be a finite set and H a finite group left-acting on X. Denote by φ(h) the permutation induced by h∈ H on X .

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-37
SLIDE 37

Group Group Action ction Systems ystems (I) (I)

Let X be a finite set and H a finite group left-acting on X. Denote by φ(h) the permutation induced by h∈ H on X . Let S be a finite group and χ: H a S a group homorphism. Then, the tuple (X, H, χ,S) is called a group action system.

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-38
SLIDE 38

Group Group Action ction Systems ystems (II) (II)

Given a group action system (X, H, χ,S), a PHF can be constructed via a suitable indexing of H, i.e., given a finite set K, ~ : K a H the tuple (X, H, K, S, χ, ~) defines a PHF (AcPHF) H = (H, K, X, L, X, S, χ °~ ), where L:= { x ∈ X | |(Kerχ)(x)| = 1 }.

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-39
SLIDE 39

Group Group Action ction Systems ystems (III) (III)

Note that:

L:= { x ∈ X | (Kerχ)(x) = x };

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-40
SLIDE 40

Group Group Action ction Systems ystems (III) (III)

Note that:

L:= { x ∈ X | (Kerχ)(x) = x }; Kerχ ⊆ Stab(L);

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-41
SLIDE 41

Group Group Action ction Systems ystems (III) (III)

Note that:

L:= { x ∈ X | (Kerχ)(x) = x }; Kerχ ⊆ Stab(L); H leaves L invariant;

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-42
SLIDE 42

Group Group Action ction Systems ystems (III) (III)

Note that:

L:= { x ∈ X | (Kerχ)(x) = x }; Kerχ ⊆ Stab(L); H leaves L invariant; We will be interested in systems for which the

(Kerχ)−orbits of elements in X\L are large.

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-43
SLIDE 43

L X Π

~(k)(x) x

X*

~(k)(x*)

χ(~(k))

AcPHFs AcPHFs

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-44
SLIDE 44

Useful Useful AcPHFs. AcPHFs.

A group action system (X, H, χ, S) is p-diverse if |(Kerχ)(x)| ≥ p, ∀ x ∈ X\L.

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-45
SLIDE 45

Useful Useful AcPHFs. AcPHFs.

A group action system (X, H, χ, S) is p-diverse if |(Kerχ)(x)| ≥ p, ∀ x ∈ X\L.

  • Lemma. If (X, H, χ, S) is p-diverse, then (X, H, K, S, χ, ~)

is (1/p)-universal.

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-46
SLIDE 46

Useful Useful AcPHFs. AcPHFs.

A group action system (X, H, χ, S) is p-diverse if |(Kerχ)(x)| ≥ p, ∀ x ∈ X\L.

  • Lemma. If (X, H, χ, S) is p-diverse, then (X, H, K, S, χ, ~)

is (1/p)-universal. Moreover…

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-47
SLIDE 47

Useful Useful AcPHFs. AcPHFs.

A group action system (X, H, χ, S) is p-diverse if |(Kerχ)(x)| ≥ p, ∀ x ∈ X\L.

  • Lemma. If (X, H, χ, S) is p-diverse, then (X, H, K, S, χ, ~)

is (1/p)-universal. Moreover… …there´s a “dedicated” way of upgrading it to (1/p)-universal2 !!

M.I. González-Vasco, Bochum 05 Group Action Based PHFs

slide-48
SLIDE 48

Examples Examples

slide-49
SLIDE 49

An An example example using using linear groups inear groups

Let X be Fq

n, {α1,…, αn} and Fq basis for X.

M.I. González-Vasco, Bochum 05 Examples

slide-50
SLIDE 50

An An example example using using linear groups inear groups

Let X be Fq

n, {α1,…, αn} and Fq basis for X.

Let H≤ GL(n, q), leaving a d-dimensional space L invariant.

M.I. González-Vasco, Bochum 05 Examples

slide-51
SLIDE 51

An An example example using using linear groups inear groups

Let X be Fq

n, {α1,…, αn} and Fq basis for X.

Let H≤ GL(n, q), leaving a d-dimensional space L invariant. Define χ : H a GL(d, q) M a Md

M.I. González-Vasco, Bochum 05 Examples

slide-52
SLIDE 52

An An example example using using linear groups inear groups

Let X be Fq

n, {α1,…, αn} and Fq basis for X.

Let H≤ GL(n, q), leaving a d-dimensional space L invariant. Define χ : H a GL(d, q) M a Md …How to achieve p-diversity?

M.I. González-Vasco, Bochum 05 Examples

slide-53
SLIDE 53

An An example example using sing non-abelian non-abelian groups roups

Take X non-abelian, H ≤ Aut(X),

Examples M.I. González-Vasco, Bochum 05

slide-54
SLIDE 54

An An example example using sing non-abelian non-abelian groups roups

Take X non-abelian, H ≤ Aut(X), L ≤ X, H-invariant (h(L) = L ∀ h∈ H)

Examples M.I. González-Vasco, Bochum 05

slide-55
SLIDE 55

An An example example using sing non-abelian non-abelian groups roups

Take X non-abelian, H ≤ Aut(X), L ≤ X, H-invariant (h(L) = L ∀ h∈ H) Construct a projection χ: H a H|L by means of a “group base” of L; i.e., a sequence [α1,…, αn], with each αi= (αi1,…, αiri ), αiji ∈ G, so that each g∈ L can be expressed as a product: g = α1j1 ··· αsjs, where αiji∈ αi.

Examples M.I. González-Vasco, Bochum 05

slide-56
SLIDE 56

An An example example using sing non-abelian non-abelian groups roups

Take X non-abelian, H ≤ Aut(X), L ≤ X, H-invariant (h(L) = L ∀ h∈ H) Construct a projection χ: H a H|L by means of a “group base” of L; that is, a sequence [α1,…, αn], with each αi= (αi1,…, αiri ), αiji ∈ G so that each g∈ L can be expressed as a product: g = α1j1 ··· αsjs, where αiji∈ αi . Then, χ : H a H|L h a (h(α1j1),…,h(αsjs))

Examples M.I. González-Vasco, Bochum 05

slide-57
SLIDE 57

An An example example using sing non-abelian non-abelian groups roups

Seems simple but…

Examples M.I. González-Vasco, Bochum 05

slide-58
SLIDE 58

An An example example using sing non-abelian non-abelian groups roups

Seems simple but… further requirements are needed!

Examples M.I. González-Vasco, Bochum 05

slide-59
SLIDE 59

An An example example using sing non-abelian non-abelian groups(II) roups(II)

Seems simple but… further requirements are needed! For instance, for realising Cramer and Shoup´s scheme:

random elements from L must be hard to distinguish from

random elements from X.

“factoring” x∈ L with respect to the group base α should be hard

(without trapdoor information)

(for details, see G-V, Martínez, Steinwandt, Villar [TCC 05])

Examples M.I. González-Vasco, Bochum 05

slide-60
SLIDE 60

A Geometric A Geometric Example xample

Let p be a finite projective plane over a prime field Fq, let X be the point-set of p , L a fixed line in p , and c a fixed point on L.

M.I. González-Vasco, Bochum 05 Examples

slide-61
SLIDE 61

A Geometric A Geometric Example xample

Let p be a finite projective plane over a prime field Fq, let X be the point-set of p , L a fixed line in p , and c a fixed point on L. Take H the group of elations with center c (note that every elation induces a permutation in the L points).

M.I. González-Vasco, Bochum 05 Examples

slide-62
SLIDE 62

A Geometric A Geometric Example xample

Let p be a finite projective plane over a prime field Fq, let X be the point-set of p , L a fixed line in p , and c a fixed point on L. Take H the group of elations with center c (note that every elation induces a permutation in the L points). Define χ as the group homomorphism χ : H a SL ζ a ζ|L

M.I. González-Vasco, Bochum 05 Examples

slide-63
SLIDE 63

A Geometric A Geometric Example xample

M.I. González-Vasco, Bochum 05 Examples

L c Kerχ = elations with axis L, thus |Kerχ| = q p a

slide-64
SLIDE 64

Final Remarks Final Remarks

slide-65
SLIDE 65

Final Remarks Final Remarks

Given a suitable group action system, we know how to

construct “good” PHFs.

M.I. González-Vasco, Bochum 05 Final Remarks

slide-66
SLIDE 66

Final Remarks Final Remarks

Given a suitable group action system, we know how to

construct “good” PHFs.

Unfortunately, so far “good” ≠ “good enough”, as the main

cryptographic constructions require aditional properties.

M.I. González-Vasco, Bochum 05 Final Remarks

slide-67
SLIDE 67

Final Remarks Final Remarks

Given a suitable group action system, we know how to

construct “good” PHFs.

Unfortunately, so far “good” ≠ “good enough”, as the main

cryptographic constructions require aditional properties.

However, this framework sheds some light on how to use

(robust enough) problems not yet exploited.

M.I. González-Vasco, Bochum 05 Final Remarks

slide-68
SLIDE 68

Thank Thank you!!!

  • u!!!