Information Technology Reference
In-Depth Information
k
1
/(
1
+
ρ
)
1
/(
1
+
ρ
)
E
(
ρ
)
=
ρ
(
+
ρ
)
log(
p
+
(
p
)
),
R
=
.
0
m
m
k
+
r
In the next Section 2 we apply these results in order to show how can be achieved
the key-capacity in asymptotic case for the special key-sharing protocol and how can
we approach to this key-capacity in non-asymptotic case depending on the string
length k and channel parameters p , p . In Section 3 we discuss these results in
the framework of implementation complexity and present some open problems.
2
The Proof of the Lower Bound for the Key-Rate
Let us consider the following key-sharing algorithm (KSA):
1. Bob generates truly random binary string
of the length k and sends it to Alice
through BSCm.
2. Alice generates truly random binary string x , computes
x
, where
is a
noisy version
received by Alice in the step 1 of KSA ,
- is modulo 2 addition
and sends it to Bob, through noiseless public channel.
3. Bob computers
x
=
x
.
y
x
4. Alice
sends
to
Bob
the
check
string
to
the
string
, where P is the matrix of the reduced echelon form
representing of the generator matrix of some linear
y
=
zH
=
xAH
=
xP
2
2
(
k
,
k
+
r
)
code. This code
(properly the matrix P ) can be chosen by legal parties to be good (with point of
view error correcting capability and the existence for it a constructive algorithm of
error correction).
5. Bob corrects errors in the noisy version x
of the Alice's string x . Denote by ~
this string after correction of errors.
6. Both Alice and Bob compute the key strings
K ,
A K
applying a multiplication by
B
to the strings x and ~
H 1
A
, respectively.
Let us prove the following asymptotic statement
Theorem 1. The KSA presented above allows to achieve asymtotically (as
matrix
k )
the key-capacity given by (1) if the legal channel is BSCm and the wire-tap channel is
BSCw for any
< w p .
Proof. By definition of the key-capacity it is such maximum possible key-rate
k
0
p
<
1
/
2
0
1
/
2
,
m
k . We can see from KSA that after a per-
formance of its steps 1-3 there exists a virtual BSC between Alice and Bob with the
same error probability
R k
=
l
/
I ,
P
0
that
as
p
as in the original legal BSCm. Hence we can use (4) in
m
k . Since Eve receives the sequence
sent by Bob to Alice in the step 1 of KSA through BSCw and
P
0
order to provide the condition
as
x
sent by
Search WWH ::




Custom Search