Chaos, Solitons and Fractals 35 (2008) 408–419

www.elsevier.com/locate/chaos

A novel algorithm for image encryption based on mixture

of chaotic maps
a,*
S. Behnia , A. Akhshani a, H. Mahmodi a, A. Akhavan b

a
Department of Physics, IAU, Urmia, Iran
b
Department of Engineering, IAU, Urmia, Iran

Accepted 14 May 2006

Communicated by Prof. M.S. El Naschie

Abstract

Chaos-based encryption appeared recently in the early 1990s as an original application of nonlinear dynamics in the
chaotic regime. In this paper, an implementation of digital image encryption scheme based on the mixture of chaotic
systems is reported. The chaotic cryptography technique used in this paper is a symmetric key cryptography. In this
algorithm, a typical coupled map was mixed with a one-dimensional chaotic map and used for high degree security
image encryption while its speed is acceptable. The proposed algorithm is described in detail, along with its security
analysis and implementation. The experimental results based on mixture of chaotic maps approves the effectiveness
of the proposed method and the implementation of the algorithm. This mixture application of chaotic maps shows
advantages of large key space and high-level security. The ciphertext generated by this method is the same size as
the plaintext and is suitable for practical use in the secure transmission of confidential information over the Internet.
 2006 Elsevier Ltd. All rights reserved.

1. Introduction

Chaos theory has been established since 1970s by many different research areas, such as physics, mathematics, engi-
neering, and biology, etc. [1]. Since 1990s, many researchers have noticed that there exists the close relationship between
chaos and cryptography [2,3]; many properties of chaotic systems have their corresponding counterparts in traditional
cryptosystems. Chaotic systems have several significant features favorable to secure communications, such as ergodic-
ity, sensitivity to initial condition, control parameters and random like behaviour, which can be connected with some
conventional cryptographic properties of good ciphers, such as confusion/diffusion. With all these advantages scientists
expected to introduce new and powerful tools of chaotic cryptography. Cryptography is the art of achieving security by
encoding messages to make them non-readable. Cryptosystems are divided between those that are symmetric (secret-
key) and those that are asymmetric (public-key). The chaotic cryptography technique which we are going to use in this

*
Corresponding author.
E-mail address: s.behnia@iaurmia.ac.ir (S. Behnia).

0960-0779/$ - see front matter  2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chaos.2006.05.011
 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419 409

paper belongs to symmetric key cryptography. There is a further division of symmetric cryptosystems into block ciphers
and stream ciphers. Two main approaches to the application scenario of chaotic systems in designing cryptographic
systems can be found in the literatures: analog and digital. Most analog chaos-based cryptosystems based on the con-
cept of chaotic synchronization, first shown by Pecora and Carrol [4]. Also many different chaotic modulations have
been proposed. These techniques may be classified into the following main families:chaotic switching or chaos shift key-
ing (CSK), differential CSK (DCSK), and chaotic masking. Some of these methods are overviewed in [5–9].
This paper chiefly focuses on the chaotic digital encryption techniques.In the digital world nowadays, the security of
digital images becomes more important since the communications of digital products over network occur more and
more frequently. Thus, to protect the content of digital images, some specific encryption systems are needed. Encryption
of images is different from that of texts due to some intrinsic features of image, such as bulk data capacity and high
correlation among pixels, traditional cryptographic techniques such as DES, IDEA and RSA are not suitable for prac-
tical image encryption, especially under the scenario of on-line communications.
The ideas of using digital chaotic systems to construct cryptosystems have also been proposed [10–17]. Undoubtedly,
security has been a topic of increasing importance in communications as the Internet and personal communications
systems are being made accessible worldwide.
However, the recent development of chaotic cryptosystem is rather disappointing. On the other hand, various crypt-
analysis have exposed some inherent drawbacks of chaotic cryptosystems [18–21]. Here we enumerate some of them;
weakness in security even with chaotic dynamics completely hidden, slow performance speed due to analytical float-
ing-point computation and small key space, which make it difficult to promote the chaotic digital encryption into prac-
tical service.
In this paper, a new design of a class of chaotic cryptosystems is suggested to overcome the aforementioned draw-
backs by using one dimensional chaotic map and their coupling for obtaining high level security [22,23].
The motivation to investigate coupled chaotic map along with a chaotic map as a cryptosystem using coupling
parameter was to achieve some computational goal, and also using coupling parameter as connection between coupled
map and third map. An approach to further enhance the security is to use high-dimensional chaotic systems such as a
coupled map. A coupled dynamical network can be regarded as a high-dimensional dynamical system. On the other
hand, we use high-dimensional chaos as the basic structure of the cryptography, which leads to the following significant
advantages: due to the high-dimensionality and chaoticity, the output ciphertext has high complexity, long periodicity
of computer realization of chaos, and effective byte confusion and diffusion in many directions in the variable space. All
these properties are favorable to achieve high practical security. In this paper a general framework for chaos-based
cryptography has been developed. By applying the proposed technique, it is possible to design cryptosystems based
on introduced hierarchy of chaotic maps [22,23].
This paper will be arranged as follows. In Section 2 a brief description of chaotic coupled map is presented. The
encryption algorithms are described in Section 3 and in Section 4 the results are devoted. Security analysis and conclu-
sion will be discussed in Sections 5 and 6, respectively.

2. Applied coupled map model

Simulation of the natural phenomena is one of the most important research fields and coupled map lattices are a
paradigm for studying fundamental questions in spatially extended dynamical systems.
We can divide them into two categories: internal and external coupled map lattice [24]. Globally coupled map is one
of the most well-known examples of the external coupling, with different number of elements [25]. Coupling relations
such as the week coupling [26], noisy coupling [27], and functional coupling [28] used to make the new coupled map. In
order to create the new encryption model based on the internal coupling, in this section we introduce coupled map
model of two chaotic system, where they are coupled according their control parameter in order to increase the security
of encrypted massage and decrease the encryption time.

2.1. One-dimensional map

The modeling of the physical system may be most appropriate in terms of a discrete time. Some examples of one-
dimensional maps are Lorenz map, tent map, and logistic map.
We generalize Logistic maps to a Hierarchy of one parameter families of maps with invariant measure, where the
Logistic map is topologically conjugate to the first map of this hierarchy [22].
These one-parameter families of chaotic maps of the interval [0, 1] with an invariant measure can be defined as the
ratio of polynomials of degree N:
410 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419
pffiffiffiffiffiffiffiffiffi
a2 ðT N ð xðnÞÞÞ2
xN ðn þ 1Þ ¼ pffiffiffiffiffiffiffiffiffi ; ð1Þ
1 þ ða2  1ÞðT N ð xðnÞÞ2 Þ
where TN are Chebyshev polynomials of type I, a control parameter, n presents the time and N is an integer greater than
one. xN(n + 1) is (N  1)-model map, that is it has (N  1) critical points in unit interval [0, 1]. By studying Shwarzian
derivative one can show that, the maps xN(n + 1) have at most N + 1 attracting periodic orbits [22,29]. These maps have
only one single period one stable fixed points or they are ergodic.
Maps do not have any kind of n-cycle or periodic orbits for N1 < a < N , actually they are ergodic for this interval of
parameter.
  Hence all n-cycles except for possible period one fixed points x = 0 and x = 1 are unstable, where for
a 2 0; N1 , the fixed point x = 0 is stable in maps xN(n + 1) (for odd integer values of N), while for a 2 [N, 1) the
x = 1 is stable fixed point in the maps. We used their conjugate or isomorphic maps. Conjugacy means that the invert-
ible map hðxÞ ¼ 1xx
; maps I = [0, 1] into [0, 1) and transform maps xN(n + 1) into ~xN ðn þ 1Þ defined as:
1 pffiffiffiffiffiffiffiffiffi
~xN ðn þ 1Þ ¼ h  xN ðn þ 1Þ  h1 ¼ 2 tan2 ðN arctan xðnÞÞ: ð2Þ
a
In this paper we used the hierarchy of chaotic map where we increase their domain to [0, 1) in order to improve the
security by increasing the control parameter domain and freedom in choosing the initial conditions.
We have derived analytically their invariant measure for arbitrary values of the parameter a and every integer values
of N [22]
pffiffiffi
1 b
lðx; bÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; b > 0: ð3Þ
p xð1  xÞðb þ ð1  bÞxÞ

We choose the parameter a in xN(n + 1) in the following form:

P½ðN 1Þ
2 
k¼0 C N2kþ1 bk
a¼ P½N2  ; ð4Þ
N k
k¼0 C 2k b

where the symbol [ ] means the greatest integer part. By considering the invariant measure, we have analytically calcu-
lated the Kolmogorov–Sinai (KS) entropy of these maps (see for more details [22]):
0 1
Z   pffiffiffi N 1
d  B Nð1 þ b þ 2 bÞ C
hks ðl; xN Þ ¼ lðxÞ dx ln  xN  ¼ ln B
@P½N   N 1 C
A: ð5Þ
dx 2 N k P½ 2  N k
k¼0 C 2k b k¼0 C 2kþ1 b

2.2. Two-dimensional coupled map

We could create the internal coupled map by using the hierarchy of families of one-parameter chaotic maps (2) as
follows:
8 pffiffiffiffiffiffiffiffiffiffiffi
< ~x1 ðn þ 1Þ ¼ a2 ðx1 ðnÞÞ tan2 ðN 1 arctan x1 ðnÞÞ;
1 2
UN 1 ;N 2 ð~x1 ; ~x2 Þ ¼ pffiffiffiffiffiffiffiffiffiffiffi ð6Þ
: ~x2 ðn þ 1Þ ¼ 12 tan2 ðN 2 arctan x2 ðnÞÞ;
a 2

where they was coupled through salve map control parameter a:

 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
BN bð~xðnÞÞ bð~xðn þ 1ÞÞ pffiffiffiffiffi
aN ð~xðnÞÞ ¼  ; bð~xðnÞÞ ¼ ð b0 þ ~xðnÞÞ2 : ð7Þ
AN 1 bð~xðnÞÞ
bð~xðnÞÞ

Now with respect to N one could expand b0 and control parameter of slave map in respect to the control parameter of
master map (for more detail refer to [22,23]). As an illustration we give:
8 pffiffiffiffiffiffiffiffiffiffiffi
< ~x1 ðn þ 1Þ ¼ a2 ðx1 ðnÞÞ tan2 ð2 arctan x1 ðnÞÞ;
1 2
U2;2 ð~x1 ; ~x2 Þ ¼ pffiffiffiffiffiffiffiffiffiffiffi ð8Þ
: ~x2 ðn þ 1Þ ¼ 12 tan2 ð2 arctan x2 ðnÞÞ
a 2
 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419 411

with
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffi 2
2bðx2 ðnÞÞ bðx2 ðn þ 1ÞÞ a1
a1 ðx2 ðnÞÞ ¼ ; bðxðnÞÞ ¼ þ xðnÞ :
1 þ bðx2 ÞðnÞ bðx2 ðnÞÞ 2  a1

As it is proved [23] the invariant measure lUN ;N ðx1 ; x2 Þ has the following form:
1 2
pffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi
1 b2 1 b1 ðx2 Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð9Þ
p x2 ð1  x2 Þðb2 þ ð1  b2 Þx2 Þ p x1 ð1  x1 Þðb1 ðx2 Þ þ ð1  b1 ðx2 ÞÞx1 Þ

with b2 > 0 and b1(x) > 0 given in (4).

KS-entropy for heirarchy of introduced coupled map in respect to invariant measure calculated and the result pre-
sented in our previous paper [23].

3. Proposed algorithm based on coupled map

In this section, a chaotic cryptosystem is presented. The proposed cryptosystem is a symmetric key block cipher
algorithm in which chaotic Coupled maps are used along with a single chaotic map. A positive way to describe the
key space might be in term of positive Lyapunov exponents. Since it was proved that the introduced map has invariant
measure according the Bikhof ergodic theorem one could use the equally the KS-entropy and Lyapunov exponents [30].
Now refereing to analytically calculated KS-entropy, we select a suitable control parameters domain at chaotic maps
for the key space (see Figs. 1(a) and (b)). Fig. 2 shows the block diagram of the chaotic encryption algorithm. The sys-
tem consists of a chaotic coupled map and a single chaotic map. First, image Im·n is transformed into matrix M(m·n)·1
and then this matrix is encrypted using results of iteration of chaotic coupled map and the third map. Using the initial
condition and control parameters of the coupled map, the coupled map is iterated and then using a function of new
~x1 ðn þ 1Þ and ~x2 ðn þ 1Þ initial condition for the third map is made. Then the third map is iterated once and Ci is gen-
erated using.
C i ¼ M i XORð~x3 ðn þ 1Þ  1014 mod 256Þ:
Then initial condition of ~x2 ðn þ 1Þ and coupling parameter () are generated using a function of Ci and ~x3 ðn þ 1Þ. This
process continues up to Mm·n. Then Mm·n is set equal to Cm·n and the whole process is repeated for the new M from the
last element to the first one and new matrix C is the output as the ciphertext. The decryption procedure is similar to that
of encryption process illustrated above with reverse of cipher text as input instead of plain text in the encryption pro-
cedure. Since both decryption and encryption procedures have similar structure, they have essentially the same algorith-
mic complexity and time consumption.
In this scheme the coupling parameter of the coupled map and initial condition of the third map are changed in each
round and also they are sensitive to plaintext, so that the change of a pixel in the plaintext causes a completely different
cipher text. In every step control parameters change in the ~x1 ðn þ 1Þ. If we assume that attacker can find control param-
eters of ~x2 ðn þ 1Þ then parameters of ~x1 ðn þ 1Þ may be detected too. To solve this problem the coupling parameter was
related to the third map. As example in this paper we select the following maps from the hierarchy of chaotic maps.
8 pffiffiffiffiffiffiffiffiffiffiffi
< ~x1 ðn þ 1Þ ¼ a2 ðx1 ðnÞÞ tan2 ð2 arctan x1 ðnÞÞ;
1 2
U2;3 ð~x1 ; ~x2 Þ ¼ pffiffiffiffiffiffiffiffiffiffiffi ð10Þ
: ~x2 ðn þ 1Þ ¼ 12 tan2 ð3 arctan x2 ðnÞÞ
a 2

with
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2bðx2 ðnÞÞ bðx2 ðn þ 1ÞÞ
a1 ðx2 ðnÞÞ ¼ ;
1 þ bðx2 ÞðnÞ bðx2 ðnÞÞ
rffiffiffiffiffiffiffiffiffiffiffiffiffi 2
a1
bðxðnÞÞ ¼ þ xðnÞ : ð11Þ
2  a1
The last part of encryption follows by:
1 pffiffiffiffiffiffiffiffiffiffiffi
~x3 ðn þ 1Þ ¼ 2 tan2 ð10 arctan x3 ðnÞÞ: ð12Þ
a3
412 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419

Fig. 1. Lyapunov exponents: (a) U2;3 ð~x1 ; ~x2 Þ vs. a1 and a2, (b) x~3 ðn þ 1Þ vs. a3.

4. Experimental results

In this section, some experiments have been done to evaluate the performance of the proposed algorithm. Experi-
mental analysis of the new algorithm presented in this paper has been done with ‘‘Barbara’’ image. Fig. 3(a) is the 256
grey-scale Barbara plain-image of size 256 · 256. Fig. 3(b) is it’s encrypted image with the encryption key. For example
encryption keys are chosen as follows, first we choose a2 = 1.5 and substitute a1 = 1.2 in Eq. (11) and the encryption
process continues by choosing a3 = 3. It should be mentioned that the initial conditions are ~x1 ð0Þ ¼ 2, ~x2 ð0Þ ¼ 33.
The performed experiments were done on a 2.4 GHz Intel celeron Pentium(IV), 256 Mb memory and 80 Gb hard-
disk capacities. Another important feature, we observe that decryption/encryption processes duration is less than 0.2 s.
As regard the obtained results, it can be seen that the decrypted image is clear and correct without any distortion. So
it can be concluded that the chaotic encryption algorithm is sensitive to the key, it should be pointed out that a small
change of the key would generate a completely different decryption result. For the proposed scheme, the corresponding
ciphertext is the same size as the plaintext.
With a statistical analysis of ‘Barbara’ image and its encrypted image, their grey-scale histograms are given in Figs.
4(a) and (b). Fig. 4(b) shows uniformity in distribution of grey-scale of the encrypted image.

5. Security analysis

The crucial measure for the quality of a cryptosystem is its capability to withstand the attempts of an unauthorized
participant-an opponent-to gain knowledge about the unencrypted information. This measure is called security. The
 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419 413

Input Im×n
and secret keys
g and h are any simple function of Ci and x3
f is a functin of x1 and x2.
Transform I to
M(m×n)×1

set K=false
i=1
Mi=C'(m×n-i),1
Iterate Coupled Map
and calculate x1 and x2

x3=f(x1,x2)
K=ture
i=1, C'=C
yes
k=true Iterate x3 once

no
Iterate x3 once Mi=C'(m×n-i),1

Ci=Mi XOR( X3×1014 mod 256)

ε =g(Ci,x3)
x2 =h(Ci,x3)

no yes no
i=i+1 i > = m×n K=true

Output
Cipher text

Fig. 2. Block diagram.

Fig. 3. (a) Plain-image. (b) Ciphered image.

discussion of the security for discrete-value cryptosystems is based on a model which was first introduced by Shannon
[31] and was extended later by others. In this section the security analysis as follows: Key space analysis, Information
entropy, Correlation analysis of two adjacent pixels and Differential attack.
414 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419

Fig. 4. (a) Histogram of plain-image. (b) Histogram of ciphered-image.

5.1. Key space analysis

Key space size is the total number of different keys that can be used in the encryption. Cryptosystem is completely
sensitive to all secret keys. If the precision is 1014, the key space size for initial conditions and control parameters is
over than 2260. Apparently, the key space is large enough to resist all kinds of brute-force attacks.

5.2. Information entropy

Information theory is the mathematical theory of data communication and storage founded in 1949 by Claude
E. Shannon [32]. Modern information theory is concerned with error-correction, data compression, cryptography, com-
munications systems, and related topics.
To calculate the entropy H(m) of a source m, we have:
X
2N 1
1
H ðmÞ ¼ P ðmi Þlog2 ; ð13Þ
i¼0
P ðmi Þ

where p(mi) represents the probability of symbol mi and the entropy is expressed in bits. Let us suppose that the source
emits 28 symbols with equal probability, i.e., m ¼ fm1 ; m2 ; . . . ; m28 g. After evaluating Eq. (13), we obtain its entropy
 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419 415

H(m) = 8, corresponding to a truly random source. Actually, given that a practical information source seldom generates
random messages, in general its entropy value is smaller than the ideal one. However, when the messages are encrypted,
their entropy should ideally be 8. If the output of such a cipher emits symbols with entropy less than 8, there exists
certain degree of predictability, which threatens its security.
Let us consider the ciphertext of image encryption using the proposed algorithm, the number of occurrence of each
ciphertext block is recorded and the probability of occurrence is computed. The entropy is as follows:
X
255
1
H ðmÞ ¼ P ðmi Þlog2 ¼ 7:9968  8:
i¼0
P ðmi Þ

The value obtained is very close to the theoretical value of 8. This means that information leakage in the encryption
process is negligible and the encryption system is secure upon the entropy attack.

5.3. Correlation analysis of two adjacent pixels

The superior confusion and diffusion properties are shown by a test on the correlations of adjacent pixels in the
ciphered image [17]. To test the correlation between two vertically adjacent pixels, two horizontally adjacent pixels,
and two diagonally adjacent pixels, respectively, in a ciphered image, the following procedure was carried out. 1000
pairs of two adjacent (in vertical, horizontal, and diagonal direction) pixels from plain-image and ciphered image were
randomly selected and the correlation coefficients were calculated by using the following two formulas:

1 XN
covðx; yÞ ¼ ðxi  EðxÞÞðy i  EðyÞÞ; ð14Þ
N i¼1
covðx; yÞ
rxy ¼ pffiffiffiffiffiffiffiffiffiffipffiffiffiffiffiffiffiffiffiffi ; ð15Þ
DðxÞ DðyÞ

where x and y represent grey-scale values of two adjacent pixels in the image. In numerical computation, the following
discrete formulas were used:

1 XN
EðxÞ ¼ xi ; ð16Þ
N i¼1
1 XN
DðxÞ ¼ ðxi  EðxÞÞ2 : ð17Þ
N i¼1

Fig. 5 shows the correlation distribution of the two horizontally adjacent pixels in the plain-image and in the cipher
image: the correlation coefficients are 0.9574 and 0.0038, respectively, which are far apart. Similar results for diagonal
and vertical directions were obtained, which are shown in Table 1.

5.4. Differential attack

To test the influence of one-pixel change on the whole image encrypted by the proposed algorithm, two common
measures were used: NPCR and UACI [17]. NPCR means the change rate of the number of pixels of ciphered image
while one pixel of plain-image is changed. UACI which is the unified average changing intensity, measures the average
intensity of the differences between the plain-image and ciphered image.
Denote two cipher-images, whose corresponding plain-images have only one-pixel difference, by C1 and C2, respec-
tively. Label the grey-scale values of the pixels at grid (i, j) of C1 and C2 by C1(i, j) and C2(i, j), respectively. Define a
bipolar array, D, with the same size as image C1 or C2. Then, D(i, j) is determined by C1(i, j) and C2(i, j), namely, if
C1(i, j) = C2(i, j) then D(i, j) = 1; otherwise, D(i, j) = 0.
NPCR and UACl are defined by the following formulas:
P
Dði; jÞ
ij
NPCR ¼  100%; ð18Þ
W  H" #
1 X jC 1 ði; jÞ  C 2 ði; jÞj
UACI ¼  100%; ð19Þ
W  H i;j 255
416 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419

300
Correlation Coefficient= 0.0038

pixel gray value on location (x+1,y)

250

200

150

100

50

0
0 50 100 150 200 250 300
pixel gray value on location (x,y)

300

correlation coefficient = 0.9574

pixel gray value on location (x+1,y)

250

200

150

100

50

0
0 50 100 150 200 250 300
pixel gray value on location (x,y)

Fig. 5. Correlations of two horizontally adjacent pixels in the plain-image and in the cipher-image: (a) Correlation analysis of plain-
image. (b) Correlation analysis of cipher-image.

Table 1
Correlation coefficients of two adjacent pixels in two images
Plain image Ciphered image
Horizontal 0.9574 0.0038
Vertical 0.9399 0.0023
Diagonal 0.9183 0.0004

where W and H are the width and height of C1 or C2. Tests have been performed on the proposed scheme, about the
one-pixel change influence on a 256 grey-scale image of size 256 · 256. We obtained NPCR = 0.41962% and
UCAI = 0.3325%. The results show that a swiftly change in the original image will result in a significant change in
the ciphered image, so the algorithm proposed has a good ability to anti differential attack.

6. Summary and conclusion

In this paper, a way of improving the security of chaos-based cryptosystem is proposed, using a hierarchy of one
dimensional chaotic maps and their coupling, which can be viewed as a high dimensional dynamical system, which
 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419 417

it belongs to interval [0, 1) [22,23]. These maps, which are defined as ratios of polynomials of degree N, have interesting
features such as invariant
  measure, ergodicity, capability of calculation KS-entropy and variable control parameters in
the two interval: N1 ; N for odd N and [0, N] for even N [23].
The general concept proposed in this paper appears quite robust. One of the main motivations for using coupled
chaotic map along with a chaotic map as a cryptosystem used coupling parameter () is to achieve some computational
goal. Beside coupling parameter was used as connection between coupled map and third map. A complete description
of the algorithm, security properties, performance and implementation aspects were given before. We introduce the
mixture mechanism of chaotic maps, which enhance the key space and security of algorithm. In regard to the tight rela-
tionship between cryptography and chaos theory and two general principles which leads to the design of good crypto-
systems, diffusion and confusion [32], and with respect to partitioning of the state space which turns the deterministic
chaotic system into an ergodic information source that can be analyzed in terms of information theory. The source
tends to become stationary for mixing maps, in this case every initial measure, leads to the ergodic invariant measure.
Therefore, mixing maps are good candidates for encryption algorithms because both diffusion and confusion are their
immanent properties. The proposed algorithm presented several interesting features, such as a high level of security,
large enough key space, pixel distributing uniformity and an acceptable encryption speed. Besides the fast encryption
speed, production of the ciphertext has same size as its corresponding plaintext. The algorithm has been successfully
applied to and tested for the image encryption. Although the algorithm presented in this paper has focused on image
encryption, it is not just limited to this area and can be widely applied in the secure transmission of confidential infor-
mation over the Internet. It can be seen from the results that the proposed system offers a higher complexity. The high
complexity of such chaotic dynamics indicates that they could be advantageously used in chaotic cryptographic tech-
niques with enhanced security. According to the results obtained using our cryptosystem, encryption with this algo-
rithm seems to be more efficient than other methods, as it develops security of such systems. Hence it seems that
they could be helpful to reduce or even overcome cryptographical weaknesses of chaotic cryptosystems. As a conclu-
sion, we have used in the present algorithm only two prototype of the hierarchy of one-dimensional chaotic maps (one
dimensional chaotic map and their coupling). Apparently, it can be easily used for any mixing of 1D chaotic maps with
2D coupled chaotic maps. It seems that, the triple of such one dimensional maps may increase the security of crypto-
system in compared to the proposed cryptosysytem. Also this algorithm is suggested in serious applications that
requires a high level of security.

Appendix A. Derivation of entropy of coupled chaotic maps

The KS-entropy of U2,2 given in (8) can be written as

Z Z     
1 1 o~x1 ðn þ 1Þ  
~ 2;2 Þ ¼
l; U
hð~ ~ðx1 Þ dx1
l ~ ðx2 Þ dx2 ln 
l  þ ln o~x2 ðn þ 1Þ ðA:1Þ
0 0 ox ðnÞ 
1
 ox ðnÞ 
2

or
Z 1 Z 1
l; U
hð~ ~ 2;2 Þ ¼ ~ðx1 Þ dx1
l ~ðx2 Þ dx2 ;
l
0 0
     
 o 1 pffiffiffiffiffiffiffiffiffiffiffi   o 1 pffiffiffiffiffiffiffiffiffiffiffi 

ln  2 
tan ð2 arctan x1 ðnÞÞ  þ ln  tan ð2 arctan x2 ðnÞÞ ;
2
ox1 a21 ðx2 ðnÞÞ ox1 a22

where the measure l

~ related to the measure l to the following relation:
 
1 1
~ðxÞ ¼
l l :
ð1 þ xÞ2 1þx

The first integral by substituting a1(x2(n)) and b(x(n)) and change of variable x ¼ b1 tan2 ðhÞ reduces to:
Z p     
2 2 a2 ð2a1 þ 1Þ a2 ð2a1 þ 1Þ
¼ dh ln þ þ þ  cos hÞÞ
p 0 ða1 þ 1Þða2 þ 1Þ ða1 þ 1Þða2 þ 1Þ
    
a2 a2 3 1
 ln þ cos h þ ln þ cos h þ cos 2h þ lnðA þ B cos h þ C cos 2hÞ ;
a2 þ 1 a2 þ 1 4 4
418 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419

which
8 qffiffiffiffiffiffiffi  2
>
> 2a þ1 2ða þ1Þ a1
þ 3 ðaa2 2þ1Þ Þ;
> A ¼ 18 ð3ð a11þ1 Þ þ a22
> a1 þ1
>
> 
>
<  2 
B ¼ 12 1 þ a1aþ1 1
 ðaa2 2þ1Þ ;
>
>
>
>  qffiffiffiffiffiffiffi  2 
>
>
>
: C ¼ 1 2a1 þ1
þ 2ða2 þ1Þ a1
þ ða2 þ1Þ
:
8 a1 þ1 a2 a1 þ1 a2

The above expression can be calculated by using the following integral:

Z
1 2p
dh lnðA þ B cos h þ C cos 2hÞ ¼ 2 ln D;
p 0
0rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
A  3C þ ðA þ CÞ 2
 B 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1B A þ B þ C  A  B þ CC
D¼ B þ C:
p@ 2 2 A

where the above integral has been evaluated by using the well known mean values theorem of analytic function
Z
1 2p
dh ln jf ðz0 þ Reih Þj ¼ jf ðz0 Þj
p 0
by choosing f(z) = a + beih + ce2ih. Now the first part of integral (A.1) reads:
Z 1 Z 1  pffiffiffiffiffiffiffiffiffiffiffi !
2 1 sin 2ðarctan x2 ðnÞÞ 

¼ ~ðx1 ðnÞÞdx1 ðnÞ
l ~ðx2 ðnÞÞdx2 ðnÞ ln
l  2  pffiffiffiffiffiffiffiffiffiffiffi  pffiffiffiffiffiffiffiffiffiffiffi 
0 0 a2 x2 ðnÞð1 þ x2 ðnÞÞ cos3 2ðarctan x2 ðnÞÞ
with the same change of variable in second part integral (A.1) and by considering:
8  pffiffiffiffiffiffiffiffiffi
Z >  
1 p < ln aþ a2 b2 ; jaj > jbj;
ln ja þ b cos hj ¼  2 
p 0 : ln b;
>
jaj 6 jbj:
2

This part of integrals is taken:

!
1 þ a2aþ1
2
þ 2 a2aþ1
1

ln :
1 þ a2aþ1
2

Now, combining it’s results:

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qffiffiffiffiffiffiffiffi qffiffiffiffiffiffi4
a2
! 1 þ a1aþ1 1
þ 
1þ a2 þ1
þ 2 a2aþ1
1 a2
a2 þ1
ln þ ln sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !2 :
1 þ a2aþ1
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffi
a1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
1 þ a1aþ1 a1
 a1 þ1 2
1
þ a2 þ a2 þ a2 1 þ a1 þ1
a2 þ1 a2 þ1 a2 þ1

References

[1] Hao B. Starting with parabolas: an introduction to chaotic dynamics. Shanghai China: Shanghai Scientific and Technological
Education Publishing House; 1993.
[2] Brown R, Chua LO. Clarifying chaos: examples and counterexamples. Int J Bifurcat Chaos 1996;6(2):219–42.
[3] Fridrich J. Symmetric ciphirs based on two-dimensional chaotic maps. Int J Bifurcat Chaos 1998;8(6):1259–84.
[4] Pecora LM, Carroll TL. Synchronization in chaotic systems. Phys Rev Lett 1990;64:821–4.
[5] Feki M. An adaptive chaos synchronization scheme applied to secure communication. Chaos, Solitons & Fractals 2003;18:141–8.
[6] Parlitz U, Chua LO, Kocarev L, Halle KS, Shang A. Transmission of digital signals by chaotic synchronization. Int J Bifurcat
Chaos 1992;2:973–7.
[7] Morgul O, Feki M. A chaotic masking scheme by using synchronized chaotic systems. Phys Lett A 1999;251:169–76.
[8] Cuomo KM, Openheim AV. Circuit implementation of synchronized chaos with applications to communications. Phys Rev Lett
1993;71:65–8.
 S. Behnia et al. / Chaos, Solitons and Fractals 35 (2008) 408–419 419

[9] Chen JY, Wong KW, Cheng LM, Shuai JW. A secure communication scheme based on the phase synchronization of chaotic
systems. Chaos 2003;13:508–14.
[10] Xiao D, Liao X, Wong K. An efficient entire chaos-based scheme for deniable authentication. Chaos, Solitons & Fractals
2005;23:1327–31.
[11] Tang G, Liao X, Chen Y. A novel method for designing S-boxes based on chaotic maps. Chaos, Solitons & Fractals
2005;23:413–9.
[12] Xiang T, Liao X, Tang G, Chen Y, Wong KW. A novel block cryptosystem based on iterating a chaotic map. Phys Lett A
2006;349:109–15.
[13] Lü H, Wang S, Li X, Tang G, Kuang J, Ye W, et al. A new spatiotemporally chaotic cryptosystem and its security and
performance analyses. Chaos 2004;14:617–29.
[14] Huang F, Guan ZH. Cryptosystem using chaotic keys. Chaos, Solitons & Fractals 2005;23:851–5.
[15] Lee PH, Pei SC, Chen YY. Generating chaotic stream ciphers using chaotic systems. Chin J Phys 2003;41:559–81.
[16] Baptista MS. Cryptography with chaos. Phys Lett A 1998;240:50–4.
[17] Chen G, Mao Y, Chui C. A symmetric image encryption scheme based on 3d chaotic cat maps. Chaos, Solitons & Fractals
2004;21:749–61.
[18] Kocarev L. Chaos-based cryptography: a brief overview. IEEE Circ Syst 2001;1:6–21.
[19] Parker AT, Short KM. Reconstructing the keystream from a chaotic encryption scheme. IEEE Trans Circuits Syst I
2001;48(5):104–12.
[20] Zhou ChS, Lai CH. Extracting messages masked by chaotic signals of time-delay systems. Phys Rev E 1999;60:320–3.
[21] Dachselt F, Schwarz W. Chaos and cryptography. IEEE Trans Circuits Syst 2001;48(12):1498–509.
[22] Jafarizadeh MA, Behnia S, Khorram S, Nagshara H. Hierarchy of chaotic maps with an invariant measure. J Stat Phys
2001;104(516):1013–28.
[23] Jafarizadeh MA, Behnia S. Hierarchy of chaotic maps with an invariant and their coupling. Physica D 2001;159:1–21.
[24] Li Wentian. Phenomenology of non-local cellular automata. J Stat Phys 1992;68:829.
[25] Cosenza MG, Parravano A. Dynamics of coupling functions in globally coupled maps: Size, periodicity, and stability of clusters.
Phys Rev E 2001;64:036224.
[26] dos Santosa AM, Vianaa RL, Lopesa SR, Pintob SE de S, Batista AM. Chaos synchronization in a lattice of piecewise linear maps
with regular and random couplings. Physica A 2005.
[27] Monteb S, Dovidioc F, Chate H, Mosekildea E. Effects of microscopic disorder on the collective dynamics of globally coupled
maps. Physica D 2005;205:25–40.
[28] Coca D, Billings SA. Analysis and reconstruction of stochastic coupled map lattice models. Phys Lett A 2003;315:61–75.
[29] Devancy RL. An introduction to chaotic dynamical systems. Addison Wesley; 1982.
[30] Keller G. Equilibrium states in a ergodic theory. Cambridge University Press; 1998. p. 23–30.
[31] Shannon CE. A mathematical theory of communication. Bell Syst Tech J 1948;27(3):379–423, 623–56.
[32] Shannon CE. Communication theory of secrecy systems. Bell Syst Tech J 1949;28:656–715.