Cryptography Reference
In-Depth Information
1
1
1
3
1
2 .
and, from this, we obtain that now the scheme is perfectly secret because:
(
=
) =
2 ·
4 +
2 ·
4 =
Pr
C
1
1
2
Pr
(
C
=
0
|
M
=
0
) =
Pr
(
K
=
0
) =
=
Pr
(
C
=
0
).
1
2
Pr
(
C
=
0
|
M
=
1
) =
Pr
(
K
=
1
) =
=
Pr
(
C
=
0
).
1
2
(
=
|
=
) =
(
=
) =
=
(
=
).
Pr
C
1
M
0
Pr
K
1
Pr
C
1
1
2
Pr
(
C
=
1
|
M
=
1
) =
Pr
(
K
=
0
) =
=
Pr
(
C
=
1
).
In any encryption scheme with message space
M
and ciphertext space
C
the
inequality
holds because the encryption functions Enc k : M C
are injective (otherwise unique decryption would not be possible). If, moreover, the
scheme has perfect secrecy then we can show that another size limitation holds: the
key space must be at least as large as the message space.
| C |≥| M |
Proposition 3.1
If
(
Gen
,
Enc
,
Dec
)
is a perfectly secret encryption scheme with
message space
M
and key space
K
, then
| K |≥| M |
.
Proof Let c be a ciphertext that occurs with nonzero probability. Since the scheme
is perfectly secret we have that, for every m
M
,Pr
(
C
=
c
|
M
=
m
) =
Pr
(
C
=
c
0. Thus the probability that m encrypts to c is positive and this means that
there exists a key k
)>
c . Similarly, if m
, m
K
such that Enc k (
m
) =
M
=
m ,
we have a key k
k for,
otherwise, Enc k would not be injective. Thus we can define an injective map from
M
m ) =
K
such that Enc k (
c and this implies that k
=
to
K
by assigning to m akey k such that Enc k (
m
) =
c and this proves the
proposition.
In Example 3.1.1 we have seen an encryption scheme that not only satisfies the
inequality in the preceding proposition but, in fact, it also satisfies
M = K = C
without being perfectly secret. However, in Example 3.1.2 we have also seen that if
we modify this scheme by making the probability distribution on
K
uniform, then we
obtain a perfectly secret encryption scheme. This is a particular case of the following
celebrated theorem of Shannon:
Theorem 3.1 (Shannon's Theorem) Let
(
Gen
,
Enc
,
Dec
)
be an encryption scheme
such that
| M |=| C |=| K |
. Then the system has perfect secrecy if and only if the
following conditions hold:
(i)
The probability distribution on
K
is uniform.
(ii) For every m
M
and every c
C
, there exists a unique k
K
such that
Enc k (
m
) =
c.
Proof Suppose that the scheme is perfectly secret. As in the proof of Proposition
3.1 we see that, for each c
C
and each m
M
there exists at least one key k
K
such that Enc k (
m
) =
c .Ifwefix m , this means that
|{
Enc k (
m
) } k K |≥| C |
and,
 
 
Search WWH ::




Custom Search