1 , 2 2 3 Stphanie Delaune , Steve Kremer and Ma rk Ry - - PowerPoint PPT Presentation

1
SMART_READER_LITE
LIVE PREVIEW

1 , 2 2 3 Stphanie Delaune , Steve Kremer and Ma rk Ry - - PowerPoint PPT Presentation

Co erion-Resistane and Reeipt-F reeness in Eletroni V oting 1 , 2 2 3 Stphanie Delaune , Steve Kremer and Ma rk Ry an 1 LSV, ENS de Cahan, CNRS & INRIA, F rane 2 F rane Tlom R&D 3 Sho


slide-1
SLIDE 1 Co er ion-Resistan e and Re eipt-F reeness in Ele troni V
  • ting
Stphanie Delaune 1, 2 , Steve Kremer 2 and Ma rk Ry an 3 1 LSV, ENS de Ca han, CNRS & INRIA, F ran e 2 F ran e Tl om R&D 3 S ho
  • l
  • f
Computer S ien e, Universit y
  • f
Birmingham, UK S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
1 / 22
slide-2
SLIDE 2 Ele troni voting A dvantages: Convenient, E ient fa ilities fo r tallying votes. Dra wba ks: Risk
  • f
la rge-s ale and undete table fraud, Su h p roto
  • ls
a re extremely erro r-p rone. "A 15-y ea r-old in a ga rage
  • uld
manufa ture sma rt a rds and sell them
  • n
the Internet that w
  • uld
allo w fo r multiple votes" A vi Rubin P
  • ssible
issue: fo rmal metho ds abstra t analysis
  • f
the p roto
  • l
against fo rmally-stated p rop erties S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
2 / 22
slide-3
SLIDE 3 Exp e ted p rop erties Priva y: the fa t that a pa rti ula r voted in a pa rti ula r w a y is not revealed to any
  • ne
Re eipt-freeness: a voter annot p rove that she voted in a ertain w a y (this is imp
  • rtant
to p ro- te t voters from
  • er ion)
Co er ion-resistan e: same as re eipt-freeness, but the
  • er er
intera ts with the voter during the p roto
  • l,
e.g. b y p repa ring messages S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
3 / 22
slide-4
SLIDE 4 Summa ry Observations: Denitions
  • f
se urit y p rop eties a re
  • ften
insu iently p re ise No lea r distin tion b et w een re eipt-freeness and
  • er ion-resistan e
Goal: Prop
  • se
the rst fo rmal metho ds denitions
  • f
re eipt-freeenes and
  • er ion-resistan e
Results: F
  • rmalisation
  • f
re eipt-freenes and
  • er ion-resistan e
as some kind
  • f
  • bservational
equivalen e in the applied pi- al ulus, Co er ion-Resistan e ⇒ Re eipt-F reeness ⇒ Priva y , Case study: p roto
  • l
due to Lee et al. [Lee et al., 03℄ S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
4 / 22
slide-5
SLIDE 5 Summa ry Observations: Denitions
  • f
se urit y p rop eties a re
  • ften
insu iently p re ise No lea r distin tion b et w een re eipt-freeness and
  • er ion-resistan e
Goal: Prop
  • se
the rst fo rmal metho ds denitions
  • f
re eipt-freeenes and
  • er ion-resistan e
Results: F
  • rmalisation
  • f
re eipt-freenes and
  • er ion-resistan e
as some kind
  • f
  • bservational
equivalen e in the applied pi- al ulus, Co er ion-Resistan e ⇒ Re eipt-F reeness ⇒ Priva y , Case study: p roto
  • l
due to Lee et al. [Lee et al., 03℄ S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
4 / 22
slide-6
SLIDE 6 Outline
  • f
the talk 1 Intro du tion 2 Applied π
  • al ulus
3 F
  • rmalisation
  • f
Priva y and Re eipt-F reeness 4 F
  • rmalisation
  • f
Co er ion-Resistan e 5 Con lusion and F uture W
  • rks
S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
5 / 22
slide-7
SLIDE 7 Outline
  • f
the talk 1 Intro du tion 2 Applied π
  • al ulus
3 F
  • rmalisation
  • f
Priva y and Re eipt-F reeness 4 F
  • rmalisation
  • f
Co er ion-Resistan e 5 Con lusion and F uture W
  • rks
S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
6 / 22
slide-8
SLIDE 8 Motivation fo r using the applied π
  • al ulus
Applied pi- al ulus: [Abadi & F
  • urnet,
01℄ basi p rogramming language with
  • nstru ts
fo r
  • n urren y
and
  • mmuni ation
based
  • n
the π
  • al ulus
[Milner et al., 92℄ in some w a ys simila r to the spi- al ulus [Abadi & Go rdon, 98℄ A dvantages: allo ws us to mo del less lassi al ryptographi p rimitives b
  • th
rea habilit y and equivalen e-based sp e i ation
  • f
p rop erties automated p ro
  • fs
using ProV erif to
  • l
[Blan het℄ p
  • w
erful p ro
  • f
te hniques fo r hand p ro
  • fs
su essfully used to analyze a va riet y
  • f
se urit y p roto
  • ls
S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
7 / 22
slide-9
SLIDE 9 Motivation fo r using the applied π
  • al ulus
Applied pi- al ulus: [Abadi & F
  • urnet,
01℄ basi p rogramming language with
  • nstru ts
fo r
  • n urren y
and
  • mmuni ation
based
  • n
the π
  • al ulus
[Milner et al., 92℄ in some w a ys simila r to the spi- al ulus [Abadi & Go rdon, 98℄ A dvantages: allo ws us to mo del less lassi al ryptographi p rimitives b
  • th
rea habilit y and equivalen e-based sp e i ation
  • f
p rop erties automated p ro
  • fs
using ProV erif to
  • l
[Blan het℄ p
  • w
erful p ro
  • f
te hniques fo r hand p ro
  • fs
su essfully used to analyze a va riet y
  • f
se urit y p roto
  • ls
S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
7 / 22
slide-10
SLIDE 10 The applied π
  • al ulus
  • n
an example Syntax: Equational theo ry: de ( en ( x, y), y) = x Pro ess: P = ν s, k.( out( 1, en ( s, k)) | in( 1, y). out( 2, de ( y, k))). Semanti s: Op erational semanti s → : P → ν s,
  • k. out(
2, s) Op erational lab eled semanti s α

→ :

P

ν

x 1.
  • ut(
1, x 1)

− − − − − − − − → ν

s, k.( in( 1, y). out( 2, de ( y, k))) | {en ( s, k)/x 1}) in( 1, x 1)

− − − − − → ν

s, k.( out( 2, s) | {en ( s, k)/x 1}

. . .

S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
8 / 22
slide-11
SLIDE 11 The applied π
  • al ulus
  • n
an example Syntax: Equational theo ry: de ( en ( x, y), y) = x Pro ess: P = ν s, k.( out( 1, en ( s, k)) | in( 1, y). out( 2, de ( y, k))). Semanti s: Op erational semanti s → : P → ν s,
  • k. out(
2, s) Op erational lab eled semanti s α

→ :

P

ν

x 1.
  • ut(
1, x 1)

− − − − − − − − → ν

s, k.( in( 1, y). out( 2, de ( y, k))) | {en ( s, k)/x 1}) in( 1, x 1)

− − − − − → ν

s, k.( out( 2, s) | {en ( s, k)/x 1}

. . .

S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
8 / 22
slide-12
SLIDE 12 Stati equivalen e
  • n
frames
  • passive
atta k er F rame A frame is a p ro ess
  • f
the fo rm ν˜ n.({ M 1/ x 1} | . . . | { M n/ x n}) . Example P = ν s, k.( out( 2, s) | {en ( s, k)/x 1}

φ( P) = ν

s, k.{en ( s, k)/x 1} Stati equivalen e
  • n
frames (≈ s )

ϕ ≈

s ψ when dom(ϕ) = dom(ψ) (the frames
  • in ide
  • n
unrestri ted va riables), fo r all terms U, V , ( U = E V )ϕ i ( U = E V )ψ S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
9 / 22
slide-13
SLIDE 13 Stati equivalen e
  • n
frames
  • passive
atta k er F rame A frame is a p ro ess
  • f
the fo rm ν˜ n.({ M 1/ x 1} | . . . | { M n/ x n}) . Example P = ν s, k.( out( 2, s) | {en ( s, k)/x 1}

φ( P) = ν

s, k.{en ( s, k)/x 1} Stati equivalen e
  • n
frames (≈ s )

ϕ ≈

s ψ when dom(ϕ) = dom(ψ) (the frames
  • in ide
  • n
unrestri ted va riables), fo r all terms U, V , ( U = E V )ϕ i ( U = E V )ψ Example 1:

ν

k.({ en (a, k)/ x} | { k/ y}) ≈ s ν n.({ en (b, k)/ x} | { k/ y}) S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
9 / 22
slide-14
SLIDE 14 Stati equivalen e
  • n
frames
  • passive
atta k er F rame A frame is a p ro ess
  • f
the fo rm ν˜ n.({ M 1/ x 1} | . . . | { M n/ x n}) . Example P = ν s, k.( out( 2, s) | {en ( s, k)/x 1}

φ( P) = ν

s, k.{en ( s, k)/x 1} Stati equivalen e
  • n
frames (≈ s )

ϕ ≈

s ψ when dom(ϕ) = dom(ψ) (the frames
  • in ide
  • n
unrestri ted va riables), fo r all terms U, V , ( U = E V )ϕ i ( U = E V )ψ Example 2:

ν

k.{ en (a, k)/ x} ≈ s ν n.{ en (b, k)/ x} S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
9 / 22
slide-15
SLIDE 15 Lab eled bisimulation
  • n
p ro esses
  • a tive
atta k er Lab eled bisimulation (≈ℓ ) Lab eled bisimila rit y is the la rgest symmetri relation R
  • n
losed extended p ro esses, su h that A R B implies 1 φ( A) ≈ s φ( B) , 2 if A → A′ , then B →∗ B′ and A′ R B′ fo r some B′ , 3 if A α

A′ , then B →∗ α

→→∗

B′ and A′ R B′ fo r some B′ . Theo rem (Abadi & F
  • urnet,
01) A ≈ℓ B ⇔ no
  • ntext
an distinguish the t w
  • p
ro esses A and B . S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
10 / 22
slide-16
SLIDE 16 V
  • ting
p roto
  • ls
in the applied π
  • al ulus
Denition (V
  • ting
p ro ess) VP ≡ ν˜ n.( V σ 1 | · · · | V σ n | A 1 | · · · | A m) V σ i : voter p ro ess and v ∈ dom(σ i) refers to the value
  • f
his vote A j : ele tion autho rit y

˜

n : hannel names The
  • ut ome
  • f
the vote is made publi , i.e. there exists B su h that VP (→∗ α

− →∗)∗

B with φ( B) ≡ ϕ | { vσ 1/ x 1, . . . , vσ n/ x n} fo r some ϕ.

֒ →

S is a
  • ntext
whi h is as VP but has a hole instead
  • f
t w
  • f
the V σ i S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
11 / 22
slide-17
SLIDE 17 Outline
  • f
the talk 1 Intro du tion 2 Applied π
  • al ulus
3 F
  • rmalisation
  • f
Priva y and Re eipt-F reeness 4 F
  • rmalisation
  • f
Co er ion-Resistan e 5 Con lusion and F uture W
  • rks
S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
12 / 22
slide-18
SLIDE 18 F
  • rmalisation
  • f
p riva y Classi ally mo deled as
  • bservational
equivalen es b et w een t w
  • slightly
dierent p ro esses P 1 and P 2 , but hanging the identit y do es not w
  • rk,
as identities a re revealed hanging the vote do es not w
  • rk,
as the votes a re revealed at the end Solution: [Kremer & Ry an, 05℄

֒ →

  • nsider
2 honest voters and sw ap their votes A voting p roto
  • l
resp e ts p riva y if S[ V A{ a/ v} | V B{ b/ v}] ≈ℓ S[ V A{ b/ v} | V B{ a/ v}] . S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
13 / 22
slide-19
SLIDE 19 F
  • rmalisation
  • f
p riva y Classi ally mo deled as
  • bservational
equivalen es b et w een t w
  • slightly
dierent p ro esses P 1 and P 2 , but hanging the identit y do es not w
  • rk,
as identities a re revealed hanging the vote do es not w
  • rk,
as the votes a re revealed at the end Solution: [Kremer & Ry an, 05℄

֒ →

  • nsider
2 honest voters and sw ap their votes A voting p roto
  • l
resp e ts p riva y if S[ V A{ a/ v} | V B{ b/ v}] ≈ℓ S[ V A{ b/ v} | V B{ a/ v}] . S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
13 / 22
slide-20
SLIDE 20 Leaking se rets to the
  • er er
T
  • mo
del re eipt-freeness w e need to sp e ify that a
  • er ed
voter
  • p
erates with the
  • er er
b y leaking se rets
  • n
a hannel h W e denote b y V h the p ro ess built from the p ro ess V as follo ws: h

=

0,

( P |

Q) h

=

P h | Q h ,

  • n. P)
h

= ν

  • n. out( h,
n). P h ,

( in( u,

x). P) h

=

in( u, x). out( h, x). P h ,

( out( u,

M). P) h

=

  • ut( u,
M). P h , . . . W e denote b y V \ out( h ,·)

= ν

h .( V |!in( h , x) . S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
14 / 22
slide-21
SLIDE 21 Re eipt-freeness Denition (Re eipt-freeness) A voting p roto
  • l
is re eipt-free if there exists a p ro ess V ′ , satisfying V ′\ out( h ,·) ≈ℓ V A{ a/ v}, S[ V A{ / v} h | V B{ a/ v}] ≈ℓ S[ V ′ | V B{ / v}] . Intuitively , there exists a p ro ess V ′ whi h do es vote a , leaks (p
  • ssibly
fak e) se rets to the
  • er er,
and mak es the
  • er er
b elieve he voted S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
15 / 22
slide-22
SLIDE 22 Some results Let VP b e a voting p roto
  • l.
W e have fo rmally sho wn that: VP is re eipt-free =

VP resp e ts p riva y. Case study: Lee et al. p roto
  • l
W e have p roved re eipt-freeness b y exhibiting V ′ sho wing that V ′\ out( h ,·) ≈ℓ V A{ a/ v} sho wing that S[ V A{ / v} h | V B{ a/ v}] ≈ℓ S[ V ′ | V B{ / v}] S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
16 / 22
slide-23
SLIDE 23 Outline
  • f
the talk 1 Intro du tion 2 Applied π
  • al ulus
3 F
  • rmalisation
  • f
Priva y and Re eipt-F reeness 4 F
  • rmalisation
  • f
Co er ion-Resistan e 5 Con lusion and F uture W
  • rks
S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
17 / 22
slide-24
SLIDE 24 Intera ting with the
  • er er
T
  • mo
del
  • er ion-resistan e,
w e need to mo del intera tion b et w een the
  • er er
and the voter: 1 se rets a re leak ed to the
  • er er
  • n
a hannel 1 , and 2
  • utputs
a re p repa red b y the
  • er er
and given to the voter via 2 . W e denote b y V 1, 2 the p ro ess built from V as follo ws: 1, 2

=

0,

( P |

Q) 1, 2

=

P 1, 2 | Q 1, 2 ,

  • n. P)
1, 2

= ν

  • n. out(
1, n). P 1, 2 ,

( in( u,

x). P) 1, 2

=

in( u, x). out( 1, x). P 1, 2 ,

( out( u,

M). P) 1, 2

=

in( 2, x). out( u, x). P 1, 2 (x is a fresh va riable), . . . S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
18 / 22
slide-25
SLIDE 25 Co er ion-resistan e (1) First app ro ximation: VP is
  • er ion-resistant
if there exists a p ro ess V ′ su h that S[ V A{ / v} 1, 2 | V B{ a/ v}] ≈ℓ S[ V ′ | V B{ / v}] . Problem: the
  • er er
  • uld
  • blige
V A{ / v} 1, 2 to vote ′ = , the p ro ess V B{ / v} w
  • uld
not
  • unterbalan e
the
  • ut ome
Solution:

֒ →

a new relation w e have alled adaptive simulation (A a B) S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
19 / 22
slide-26
SLIDE 26 Co er ion-resistan e (1) First app ro ximation: VP is
  • er ion-resistant
if there exists a p ro ess V ′ su h that S[ V A{ / v} 1, 2 | V B{ a/ v}] ≈ℓ S[ V ′ | V B{ / v}] . Problem: the
  • er er
  • uld
  • blige
V A{ / v} 1, 2 to vote ′ = , the p ro ess V B{ / v} w
  • uld
not
  • unterbalan e
the
  • ut ome
Solution:

֒ →

a new relation w e have alled adaptive simulation (A a B) S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
19 / 22
slide-27
SLIDE 27 Co er ion-resistan e (2) Denition (Co er ion-resistan e) A voting p roto
  • l
is
  • er ion-resistant
if there exists a p ro ess V ′ and an evaluation
  • ntext
C satisfying S[ V A{ / v} 1, 2 | V B{ a/ v}] a S[ V ′ | V B{ x/ v}] ,

ν

1,
  • 2. C[ V
A{ / v} 1, 2] ≈ℓ V A{ / v} h ,

ν

1,
  • 2. C[ V ′]\ out( h ,·) ≈ℓ
V A{ a/ v}, where x is a fresh free va riable. Intuitively , V B{ x/ v} an adapt his vote and
  • unter-balan e
the
  • ut ome,
w e require that when w e apply a
  • ntext
C (the
  • er er
requesting V A{ / v} 1, 2 to vote ) the p ro ess V ′ in the same
  • ntext
C votes a . S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
20 / 22
slide-28
SLIDE 28 Some results Let VP b e a voting p roto
  • l.
W e have fo rmally sho wn that: VP is
  • er ion-resistant =

VP resp e ts re eipt-free.

֒ →

ree ts the intuition but the p ro
  • f
is te hni al Case study: Lee et al. p roto
  • l
Co ersion-resistan e dep ends
  • n
implementation details: en ryption with integrit y he k

֒ →

fault atta k: the p roto
  • l
is not
  • er ion-resistant
en ryption without integrit y he k

֒ →

the p roto
  • l
is
  • er ion-resistant
S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
21 / 22
slide-29
SLIDE 29 Con lusion and F uture W
  • rks
Con lusion: rst fo rmal denitions
  • f
re eipt-freeness and
  • er ion-resistan e
  • er ion-resistan e ⇒
re eipt-freeness ⇒ p riva y , a ase study giving interesting insights F uture W
  • rks:
de ision p ro edure fo r
  • bservational
equivalen e fo r p ro esses without repli ation
  • ther
p rop erties based
  • n
not b eing able to p rove individual/universal veriabilit y S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
22 / 22
slide-30
SLIDE 30 Con lusion and F uture W
  • rks
Con lusion: rst fo rmal denitions
  • f
re eipt-freeness and
  • er ion-resistan e
  • er ion-resistan e ⇒
re eipt-freeness ⇒ p riva y , a ase study giving interesting insights F uture W
  • rks:
de ision p ro edure fo r
  • bservational
equivalen e fo r p ro esses without repli ation
  • ther
p rop erties based
  • n
not b eing able to p rove individual/universal veriabilit y S. Delaune (LSV, ENS Ca han) Ele troni V
  • ting
22 / 22