You are on page 1of 3

Research on Reconciliation Algorithm

in Quantum Key Distribution



Chang-qing Gong
Computer College
Shenyang Institute of Aeronautical
Engineering
Shenyang, China
E-mail: gongcq@syiae.edu.cn
Hong-yuan Zhou
Computer College
Shenyang Institute of Aeronautical
Engineering
Shenyan,, China
E-mail:shifters@163.com
Jian-li Feng
Computer College
Shenyang Institute of Aeronautical
Engineering
Shenyang , China
E-mail:fengfeng000123@126.com


AbstractThe quantum cryptography is an emerging technology,
which uses quantum mechanics properties of micro-particles to
encode information. In the theory and practice, it has been
proved to be totally safe. Because of quantum channel being
susceptible to be interfered by noise, error correction in quantum
communication is facing a big problem today. This article will
focus on Cyclic Redundancy Check (CRC) replacing parity check
in the protocol binary mainly in order to improve the efficiency
of error correction and reduce the frequency of communication.
And then its performance will be contrasted with the protocol
binary with parity check by the self-designed simulation
platform. The result shows that protocol binary with CRC can
improve efficiency by up to around 50%.
Keywords-quantum key distribution; reconciliation; CRC;
protocol binary; simulation
I. INTRODUCTION
Quantum cryptography ensures the security of quantum
communication with the property of quantum key distribution
not being able to be tapped and one time pad which can not be
deciphered [1]. Both sides of the communication can detect any
eavesdropper by bit error rate (BER) increasing abnormally.
However, practical quantum key system can not be the ideal
one and noise can also lead to a certain BER even without
eavesdroppers. Therefore, what algorithm is used in the
reconciliation is very important.
II. THE INTRODUCTION OF QUANTUM CRYPTOGRAPHY
Quantum information science is based on physical science
and information science and its basic principle is the
Heisenberg uncertainty principle of quantum mechanics. Any
micro-particles can not be accurately measured and any
measurement of the particles will affect the state of particles.
Quantum communication bases on this point to detect the
existence of eavesdroppers [2]. Principle of quantum
mechanics rule out the possibility of cloning an unknown
quantum state, as a result, if eavesdropper copied the sample,
communication would have subjected to interference. So
quantum cryptography communication is better than classical
communication. Without altering their original state, the
unknown quantum states can not be copied. If quantum states
are known, we can copy it repeatedly. But the difficulty lies in
a single measurement not being able to obtain the exact
properties of quantum system. Because once it is measured, the
original quantum state changes, what measured is only one sate
of the various possible quantum states, unless the state happens
to be measured is the quantum eigenstate. Other measurements
will inevitably and irreversible change in the original quantum
state [3]. Quantum key communication use the principle of
quantum mechanics to encode information and both sides of
the communication monitor closely the abnormal increase of
quantum BER in order to determine whether the eavesdropper
exists. Once the eavesdropper is found, this communication
will be abandoned.
At present the most widely used classic encryption
technology is RSA cryptography [4].This public key system is
based on the complicated calculation of some function (such as
a factorization of large number) to the classic computer, but if
practical quantum computer is successful developed, this
technology will not have practical effect.
III. THE IMPROVEMENT TO PROTOCOL BINARY
A. Protocol binary
After measuring the estimated BER, the steps of protocol
binary are following [5]:
(1) A (sender) and B (receiver) rearrange their data by the
same sequence in order to make the errors uniformly random
distributed.
(2) Both of A and B packet their data and make sure that
the average number of errors of each group is less than one.
(3) A and B do the parity check in each group and check
the result through the classical channel. If the results are
different, it is supposed that this group has mistakes and the
number of mistakes is odd. This group will be divided into two
for parity check. If it continues until the last bit, the last bit is
considered as the error bit. In order to prevent eavesdroppers
from obtaining information, after each parity check the last bit
of each group will be discarded. Finally, the bit which is found
to the error one is also discarded.
(4) After the error correction above, each group may also
have mistakes because the even number of mistakes cant be
found. From the remaining data a subset will be selected
randomly and have parity check. If the check results are
identical, repeat this step. If this step is repeated for twenty
times continuously, the error correction is done. If the check
results are different, repeat step (1) (2) (3) to start a new round
of error correction and the length of each group will be longer.
This paper is supported by Science Foundation of Aeronautics of China. The
foundation No. is 2007ZC54002
2009 Ninth International Conference on Hybrid Intelligent Systems
978-0-7695-3745-0/09 $25.00 2009 IEEE
DOI 10.1109/HIS.2009.102
496
B. Protocol binary after improvement
Protocol binary uses the method of parity check to detect
mistakes, but this method can not find even-digit mistakes.
Solution to the problem is rearranging the original key to find
more errors. In order to solve the problem, this article improves
protocol binary by replacing parity check with CRC.
CRC is particularly suited to detect an error, because it has
strong error detection ability and the encoder and error
detection circuit are easy to achieve. Its basic idea is linear
coding theory and any string of binary digital code can set up a
one-to-one relationship with a polynomial whose coefficient
only contains 0 and 1. For example the code 1011011
corresponds the polynomial X
6
+X
4
+X
3
+X+1, while the
polynomial X
5
+X
4
+X
2
+X corresponding code 110110. The
two sides of the communication use a pre-agreed generated
polynomial G(x) to generate the CRC code. Let us suppose that
the information which is intended to send is recorded as T and
the length of G(x) is recorded as K. Then T is added a string
that contains K-1 zeros behind it, and after that it must be
divided by G(x) and the remainder is the CRC code [6]. What
following is that two sides of the communication check the
CRC code to verify the consistency of their data. The CRC
error detection capability has a great relationship with G(x). In
order to detect all the even-digit mistakes, G(x) should not
contain x factor, and except that it must get a remainder after
divided by X
e
+1. The e above must satisfy 0 <e n-1. And
G(x) should have (x+1) factor in order to detect all odd-digit
mistakes [7].It is so difficult that for the eavesdroppers to get
the original data from CRC code. And because its re-rate is
very low, CRC greatly improves the efficiency of error
detection while not reducing the security. Although the CRC
code can detect all even-digit mistakes and odd-digit mistakes
but because of the group being divided into two for error
correction, if two or more mistakes exist in this group, only one
mistake can be corrected. Therefore protocol binary after
improvement is still needed to be implemented repeatedly in
order to achieve zero error rate.

Figure 1. The program flow chart of protocol binary with CRC
Considering communications security, the two sides of the
communication can generate generated polynomial G(x) as
follows: First two sides of the communication use protocol
binary with parity check to generate a shared key. And one of
the parties determines which generated polynomial G(x) is
used, after that it encrypts the polynomial G(x) with the shared
key and sends it to the other. Finally they get the same
generated polynomial G(x) and in theory anyone else can not
get it. After that the two sides of the communication can use
protocol binary with CRC to generate a shared key to encrypt
information. To enhance security, the generated polynomial
G(x) must be updated regularly with the new shared key.
The program flow chart of protocol binary with CRC shows
in Figure 1.
IV. THE SIMULATION OF PROTOCOL BINARY WITH CRC
The interface of self-designed simulation platform shows in
Figure2.

Figure 2. The interface of self-designed simulation platform
The simulation platform includes three parts. First part is
quantum key communication, in which two sides of the
communication generate an original shared key. The second
part is the detecting eavesdroppers: if the BER is higher than a
certain value (here set to 5%), the eavesdroppers are considered
to exist and the communication will be discard. When the BER
is less than the certain value, the data will be corrected. The
third part is the reconciliation. Here includes protocol binary
with parity check and CRC, and the performance of the two
methods will be compared.
Use the two methods separately to record the number of
actual error bits and BER at the same number of repeat of error
correction. The simulation process is following: First of all
generate the original key and get the length of the original key
is 2464 and the estimated BER is 2.3 %, less than the certain
value 5%. The reconciliation starts. Record separately the
number of actual error bits and BER in two methods when the
number of repeat is 2,4,6,8,10,12,14,16,18,20, as shown in
Table.
497
According to the table above mapping line chart was drawn
as shown in Figure3.

0
0.
!
!.
?
? + b 8 !0 !? !+ !b !8 ?0
th uum| o ut
b
l
k
otoo' |`uu) `th u`t) hl
otoo' |`uu) `th CkC

Figure 3. Mapping line chart of the two methods
Based on the chart, at the same number of repeat, protocol
binary with CRC can improve efficiency by up to around 50%
than protocol binary with parity check.























V. CONCLUSION
This article focuses on the reconciliation in quantum key
communication and improves the protocol binary with CRC
replacing parity check. Theoretical analysis and the result of
simulation shows that the protocol binary with CRC can
improve efficiency greatly while not reducing the security.
REFERENCES
[1] Guang-can Guo, some considerations of quantum information.
Innovation and Technology, vol. 3, pp.42-43, 2007.
[2] Kui Dai, Hui Song,Yun Liu and so on. Introduction to quantum
information technology. National University of Defense Technology
Press, pp.62-63, 2001.
[3] Han-Wu Chen. Quantum Information and Quantum Computation
concise tutorial. Southeast University Press, pp.28-29, 2006.
[4] Xi-ren Xie. Computer Network, 2nd ed. Posts & Telecom Press, pp.272-
273, 2006.
[5] Rui-lin Ma. Quantum cryptography communication, Science Press,
pp.91-92, 2006.
[6] Wei-zhen Ni. Data Communication Theory, China People's University
Press, pp.189-190, 1999.
[7] Chuan-shan Gao, Di-lin Mao,Xiu Cao. Data Communication and
Computer Networks, 2nd ed. Higher Education Press, pp.84-86, 2004.
the number of repeat 2 4 6 8 10 12 14 16 18 20
Protocol binary
with parity check
the number
of error bits
41 33 31 23 20 17 14 11 8 6
BER (%) 1.72 1.41 1.34 1.00 0.87 0.75 0.62 0.45 0.38 0.25
Protocol binary
with CRC
the number
of error bits
36 22 17 15 11 7 6 4 2 0
BER (%) 1.53 0.96 0.75 0.66 0.47 0.38 0.23 0.19 0.09 0
TABLE I. THE NUMBER OF ACTUAL ERROR BITS AND BER AT THE
SAME NUMBER OF REPEAT IN THE TWO METHODS
498

You might also like