Generation Multi-Scroll Chaotic Attractors Using C

Phys. Scr. 99 (2024) 045221 https://doi.org/10.
1088/1402-4896/ad2b3f
PAPER
Generation multi-scroll chaotic attractors using composite sine

RECEIVED
26 December 2023
function and its application in image encryption
REVISED
10 February 2024
ACCEPTED FOR PUBLICATION
Pengfei Ding∗ , Ke Li and Zixuan Wang
20 February 2024 School of Electronic and Engineering, Xi’an University of Posts and Telecommunications, Xi’an, Shaanxi 710121, People’s Republic of
PUBLISHED China
∗
12 March 2024 Author to whom any correspondence should be addressed.
E-mail: [email protected]
Keywords: composite sine function, multi-scroll chaotic attractors, complexity analysis, image encryption, circuit simulation, DNA coding
Abstract
In this paper, a composite sine function is proposed and applied in a chaotic system, which is capable
of generating definite number of chaotic attractors. The proposed composite sine function possesses
infinite breakpoints, but it can produce a fixed number of scrolls by adjusting its parameters.
Compared to other chaotic systems with multiple scrolls chaotic attractors, the realization circuit of
the chaotic system with compound sine function allows obtaining different numbers of scrolls by
adjusting only one resistance value. As a result, the circuit structure remains unchanged despite
variations in the number of scrolls. Various analytical methods are applied to study the dynamical
behaviors of the proposed chaotic system, including Lyapunov exponent, equilibrium point,
bifurcation diagram, phase diagram, spectral entropy and C0-algorithm. Furthermore, based on the
analysis of dynamical characteristics, the electronic circuits of the proposed system are given on
Multisim circuit simulation software, and the multi-scroll chaotic attractors exhibit consistency with
the numerical simulation results. Finally, we incorporated the proposed chaotic system into a
Deoxyribonucleic acid coding algorithm for image encryption, and this method exhibits excellent
encryption efficiency and high level of security.
1. Introduction
The motion trajectory of state variables of a chaotic system is similar to that of a random signal, where slight
changes of initial values lead to different trajectories. However, in contrast to random motion, chaotic motion is
inherently deterministic in a dynamic way, its unpredictability due to the extreme instability of the system’s
internal dynamics. Essentially, chaotic systems present a high sensitivity to a minute variations of their initial
conditions or a tiny perturbations, such that even a very slight disturbance can lead to that the motion track
deviates completely from the system’s original trajectory after a certain period of evolution. so chaotic system is
widely employed in digital watermarking [1–3], pseudo noise code generation [4–6], image encryption [7–9],
weak signal detection [10–12] and so on.
The quantity of scrolls generated by chaotic systems is an indication of their complexity to some extent.
Hence, multi-scroll chaotic system became a widely research topic. Wang et al [13] suggested a chaotic system
with different number of scrolls by changing its system parameter. Zhang et al [14] used triangular waves, saw-
tooth waves and hysteresis sequences as nonlinear functions for generating chaotic attractors with varying
amounts of scrolls. Daras et al [15], recommended a four-dimensional autonomous chaotic system, which can
generate chaotic attractors with either a single scroll, two scrolls, or four scrolls, depending on its system’s
parameters, and it can be utilized for various applications in chaos-based cryptography and secure
communications. Luo et al [16] employed sign function and triangular waves for generating two-scroll grid
multi-scroll attractor. Echenausía-Monroy et al [17] devised a fractional-order chaotic system characterized by a
multi-scroll chaotic attractor. Zhang et al [18] recommended a novel chaotic system incorporating with a
nonlinear functions of sign function and saturated function series. Hu et al [19] introduced multi-wing and
© 2024 IOP Publishing Ltd

Phys. Scr. 99 (2024) 045221 P Ding et al
multi-scroll hidden attractors through a 5-dimensional memristive system, which scroll quantity is not related
to the system equilibrium points. Dong et al [20] proposed a method to generate multiple scrolls by utilizing a
sine function. M Yalcin et al [21] suggested a technique to produce chaotic attractors with numerous scrolls by
leveraging a sine function.
The dynamical state and the quantity of scrolls of a dynamical system is directly influenced by its nonlinear
term. For producing more scrolls, the nonlinear function used for chaotic system with more breakpoints is the
key element. One of the widely studied functions is sine function, which has infinite breakpoints, for this reason,
various sine functions are used to generate multiple scroll attractor. However, if the non-modified sine function
is directly employed in a chaotic system, the longer the simulation time, the greater number of scrolls there are.
The amounts of scrolls can be locked at a constant value by reshaping the sine function’s expression. Lü et al [22]
modulated a sine function to produce chaotic attractors with constant amounts of scrolls. Luo [23] proposed a
novel approach to generate various types of chaotic attractors by manipulating the angular frequency of the
utilized sine function. The quantity of scrolls is not varied by using normal sine function and negative feedback
[24]. Ding et al [25] multiplied a sine function by a gate function to produce stable number of scrolls. Motivated
by the above mentioned works, we designed a innovational composite sine function and seamlessly integrated it
into a Jerk system, giving birth to a fresh chaotic system. What sets this system apart is its steadfast characteristic:
its number of scrolls remains unwavering, and unaffected by the passage of simulation time.
The advancement of information technology has underscored the significance of data transmission and
storage, particularly with regard to images. As a result, a series of encryption algorithms have been put forth for
digital images [26]. To enhance encryption efficiency and security, numerous scholars have incorporated chaotic
systems into image encryption techniques. Hagras et al [27] pioneered an encryption scheme rooted in
Deoxyribonucleic acid (DNA) computing, whereas Zhang et al [28] proposed an image encryption algorithm
leveraging a 6D hyperchaotic system and DNA coding. Nevertheless, encryption methods relying on low-
dimensional chaotic systems frequently lack well-defined chaotic characteristics and are constrained by limited
key space. Despite the abundance of encryption algorithms founded on high-dimensional chaotic systems, a
notable concern arises from certain approaches, which exhibit excessively high dimensions. This phenomenon
tends to result in a reduction in encryption efficiency. We enhance the existing DNA coding encryption method
by incorporating multiple chaotic systems, which combines the proposed chaotic system and Chen
hyperchaotic system with DNA coding. The one of three state variable sequences generated by the proposed
chaotic system is utilized in DNA computation while the others to generate the scrambling matrix. Additionally,
Chen hyperchaotic system controls the operation rules of encoding, calculation and decoding of DNA. The
encryption scheme combining the proposed chaotic system and DNA coding offers the benefits of resistance to
statistical attacks, high efficiency of encryption, and a large key space.
The structure of the remaining part is outlined as follows: section 2 presents the construction of a chaotic
system mathematical model based on a composite sine function, along with a comparison of various types of
sine functions for multi-scroll chaotic systems. In section 3, the dynamical properties including phase portraits,
Lyapunov exponents, bifurcation diagrams and equilibrium points are researched. Moreover, to confirm the
chaotic attributes of the proposed system, we apply complexity algorithms, including Spectral Entropy (SE) and
the C0-algorithm for system feature analysis. In section 4, the electronic circuits are carried out on Multisim 10,
and the validity of the novel chaotic system model is demonstrated by the circuit simulation results. In section 5,
based on the multi-scroll chaotic attractors by using composite sine function, an encryption scheme combining
Chen hyperchaotic system and DNA coding is given. In section 6, we provide an analysis of the performance of
the proposed encryption algorithm, while section 7 encapsulates the key conclusions of our study.
2. Mathematical model of the novel chaotic system
The nonlinear term of the mathematical model of the novel chaotic system is a composite sine function, which is
described by differential equations as follows:
⎧ x1 = x2
x2 = x3 (1)
⎨
⎩ x3 = - mg (x1) - nx2 - cx3
where
g (x1) = - sin [2pp (x1 - x13/ d 2)] (2)
From the mathematical expression of g (x1), it can be seen that g (x1) is a composite sine function about state
variable x1. In equation (1), m, n and c are positive real number, while x1, x2 and x3 are state variables of system
(1). p, d are real constants of equation (2), and the quantity of scrolls generated by system (1) is related to d, which
2
is the parameter of sine function of g (x1). For generating multiple scrolls in chaotic system, many different types
of sine functions are proposed, such as normal sine function h (x ) [21, 24, 29], modulated sine function j (x )
[22], mathematical expression changed sine function k (x ) [19], s (x ) [30] and t (x ) [31, 32]. These sine functions
are represented as follows.
h (x ) = sin (2ppx ) (3)
j (x ) = - (x - ∣A sin ( px )∣sgn (x )) (4)
1
k (x ) = sin (2ppx )[sgn (x - a) - sgn (x - b)] + x [2 - sgn (x - a) + sgn (x - b)] (5)
2
⎧ b 2 p (x - 2b1 b3) , x  2b1 b3
⎪ 2b1
⎪
⎪ px
s (x ) = - b 2 sin ⎛
⎜ + b4⎞ , - 2b1 b3 < x < 2b1 b3
⎟ (6)
⎨ ⎝ 2b1 ⎠
⎪
b p
⎪ 2 (x + 2b1 b3) , x  - 2b1 b3
⎪ 2b1
⎩
⎧- sin (2ppx ) , - m1/ p < x < m2/ p
t (x ) = + sin (2ppx ) , x  - m1/ p (7)
⎨
⎩+ sin (2ppx ) , x  m2/ p
The oscillograms of sine function of g (x ), h (x ), j (x ), k (x ), s (x ) and t (x ) are displayed in figure 1.
When system (1) under chaotic state, keep m, n, c and d unchanged, the quantity of scrolls of the generated
chaotic attractor by system (1) varies with the parameters of p and d. For instance, let m = n = c = 0.3 and d = 10,
system (1) with p = 0.5 and p = 1 respectively, the corresponding 4-scroll and 8-scroll chaotic attractors can be
obtained, which are depicted in figures 2(a) and (b).
In figures 2(c) and (d), let m = n = c = 0.3 and d = 20, system (1) with p = 0.5 and p = 1 respectively, the
6-scroll and 12-scroll chaotic attractors can be obtained. However, comparing the chaotic attractors in figure 2,
it can be seen that the parameter of d can be employed to fine-tune the quantity of the scrolls generated by System
(1), while keeping all other parameters unchanged.
3. Dynamical characteristics of the chaotic system
This section explores the dynamical characteristics of Chaotic System (1), covering an analysis of steady states
and equilibrium points, in addition to an examination of Lyapunov exponents and bifurcation diagrams. For
better understanding of the behavior of system (1), we utilized SE algorithm and C0-algorithm to analyze its
complexity.
3.1. Equilibrium point

The equilibrium points of System (1) can be determined by solving for the roots of equation (8), which is
described as follows.
⎧ x2 = 0
x3 = 0 (8)
⎨
⎩mg (x1) + nx2 + cx3 = 0
In equation (8), it can be seen that the distribution of equilibrium points is related to the nonlinear function
of g (x1), and the equilibrium points can be written as Ep (x1, 0, 0), where x1 satisfies the following equation ,in
which k is an integer.
2pp (x1 - x13/ d 2) = kp (9)
For simplify the analysis, the parameters of p and d in equation (9) are set as p = 0.5 and d = 10, then
x13 - 100x1 + 100k = 0 (10)
In equation (10), k is integer, which is related to the position of the equilibrium point, and the solutions of
equation (10) are displayed in figure 3. Let k = 1, then x11 = 1.010, x12 = 9.456 and x13 = −10.467. It can be seen
that the point (x11,0,0) is one of the equilibrium points ‘o’, and the equation expression for solving the
equilibrium point at Ep (x11, 0, 0) is:
3
Figure 1. The oscillograms of the different sine functions: (a) g(x) with p = 0.5 and d = 10; (b) h(x) with p = 0.5; (c) j(x) with A = 6 and
p = 0.5; (d) k(x) with p = 0.5, a = −4 and b = 4; (e) s(x) with b1 = 1, b 2 = 0.1, b 3 = 3 and b 4 = 0; (f) t(x) with p = 0.5 and
m1 = m2 = 3.
⎡ 0 1 0⎤
Jac = ⎢ 0 0 1⎥ (11)
⎢
⎣ 0.3g ¢ ( x11) 0.3 0.3⎥
⎦
According to equation (11), the characteristic equation is given:
l3 - 0.3l2 - 0.3l - 0.3g ¢ (x11) = 0 (12)
its eigenvalues are l1 = -1.192, l2 = 0.446 + 0.751i and l3 = 0.446-0.751i, so the equilibrium point
Ep (x11, 0, 0) is saddle points of index 2. Let k = 0, then x′11 = 0, x′12 = 10 and x′13 = −10, and the the point
(x′11,0,0) is one of the equilibrium points ‘x’. Similarly, its eigenvalues are l¢1 = 0.9823,
l¢2 = -0.6411 + 0.7406i and l¢3 = -0.6411-0.7406i, so the equilibrium point of Ep (x ¢11, 0, 0) is saddle
points of index 1.
In figure 3, Ep1 and Ep2 are two unique equilibrium points, the distance of Ep1 to its left adjacent equilibrium
point is L1, while the distance of Ep1 to its right adjacent equilibrium point is L2. The distance of Ep2 to its left
4
Figure 2. The chaotic attractors generated by system (1) with (x1(0),x2(0),x3(0)) = (0.1,0,0). (a) 4-scroll chaotic attractor with p = 0.5,
m = n = c = 0.3 and d = 10. (b) 8-scroll chaotic attra-ctor with p = 1, m = n = c = 0.3 and d = 10. (c) 6-scroll chaotic attractor with
p = 0.5, m = n = c = 0.3 and d = 20. (d) 12-scroll chaotic attractor with p = 1, m = n = c = 0.3 and d = 20.
Figure 3. The distribution of equilibrium points of system (1) with p = 0.5 and d = 10.
adjacent equilibrium point is L3, while the distance of Ep2 to its right adjacent equilibrium point is L4. It can be
seen that L1>L2, L4>L3 and upon comparison of figures 2 and 3, it is evident that the scrolls generated by
system (1) are closely related to the equilibrium points ‘o’. In fact, the equilibrium points ‘o’ are saddle point of
index 2, and the equilibrium points ‘x’ are saddle point of index 1.
3.2. Bifurcation diagram and Lyapunov exponents

By using control variable method, the parameters of system (1) with equation (2) are fixed as p = 0.5, d = 10 and
n = c = 0.3, and the range of parameter m is chosen as m Î (0, 1). As a result, the Lyapunov exponent spectrum
and bifurcation diagram are obtained and displayed in figure 4.
From figure 4(a), it can be seen that LE1  0, LE 2  0 and LE3 < 0, and the dynamical properties of system
(1) with m Î (0, 1) are conclude as follows:
5
Figure 4. The Lyapunov exponents and bifurcation diagram of system (1) with d = 10, p = 0.5 , n = c = 0.3 and (x1(0),x2(0),
x3(0)) = (0.1,0.1,0.1). (a) Lyapunov exponents with m Î (0, 1) (b) bifurcation diagram with m Î (0, 1).
1. when m Î (0, 0.08], LE1 < 0, LE 2 < 0, LE3 < 0, the system (1) is in stable state.
2. when m Î (0.08, 0.12] È [0.13, 0.76] È [0.96, 1), LE1 > 0, LE 2 = 0, LE3 < 0, system (1) at chaotic state.
3. when m Î (0.12, 0.13) È (0.76, 0.96), LE1 = 0, LE 2 < 0, LE3 < 0, system (1) at periodic state.
The bifurcation diagram depicted in figure 4(b) for System (1) has a robust correlation with the Lyapunov
exponents, that offering valuable insights into the system’s states.
3.3. Phase portraits

The utilization of phase portraits is a highly effective approach for investigating the dynamics of chaotic systems,
thus the properties of system (1) are also analyzed by using phase portraits.
Let the parameters as p = 0.5, d = 10, n = c = 0.3 and take the initial state as x2 (0) = x3 (0) = 0,
x1 (0) = 0.1.Let m = 0.02, the phase diagrams of system (1) with equation (2) in x1-x2-x3 space and x1-x2 plane
are shown in figure 5.
According to figure 5, it is evident that in the domain of attraction, the state variable of x1 remains constant,
while the state variables of x2 and x3 exhibit minimal changes. The phase diagrams indicate the stability of System
(1), and it keep consistence with the bifurcation diagram and the Lyapunov exponents.
6
Figure 5. The phase portraits of system (1) with d = 10, p = 0.5, m = 0.02, n = c = 0.3 and (x1(0),x2(0),x3(0)) = (0.1,0,0). (a) x1-x2-x3
space. (b) x1-x2 plane.
With the parameter of m set as 0.13, the phase portraits of System (1) with equation (2) can be generated and
are illustrated in figure 6.
In figure 6, the motion trajectory of system (1) is a closed curve, which denotes that system (1) is in periodic
state.
Keep the parameter of m as 0.3, the phase portraits of System (1) with equation (2) in x1-x2-x3 space and the
x1-x2 plane can be generated and showed in figure 7.
In figure 7, the motion trajectory of system (1) indicates that System (1) is in chaotic state, and a 4-scroll
chaotic attractor is generated.
Take the parameter of m as 0.5, the phase portraits of System (1) with equation (2) can be obtained and
presented in figure 8.
From figures 7 and 8, it becomes evident that the chaotic system yields a 4-scroll chaotic attractor while
m = 0.3 or m = 0.5, the results highly match the bifurcation diagram presented in figure 4(b).
According to the mathematical expression of nonlinear function g(x) and its waveform in figure 1(a), the
number of cycles in the middle part of the composite function will vary with the parameter d. Let
m = n = c = 0.3, p = 0.5 and d = 20, the generated 6-scroll chaotic attractors in x1-x2-x3 space and x1-x2 plane are
displayed in figure 9.
From figures 5–8, it is apparent that the phase portraits of system (1) effectively represent its state. A
comparison between figures 7 and 9 leads to the conclusion that the quantity of scrolls can be adjusted through
changing the parameter d of the composite sine function g(x1).
7
Figure 8. The phase diagrams of system (1) with d = 10, p = 0.5, m = 0.5, n = c = 0.3 and (x1(0),x2(0),x3(0)) = (0.1,0,0). (a) x1-x2-x3
Figure 9. The phase portraits of system (1) with m = n = c = 0.3, p = 0.5, d = 20 and (x1(0),x2(0),x3(0)) = (0.1,0,0). (a) x1-x2-x3 space.
(b) x1-x2 plane.
3.4. Sophistication test

Sophistication test is a quantitative analysis method for complexity measurement. Different sophistication
algorithms are used for different aspects of complexity. In chaotic dynamics, sophistication test value represents
the closeness between chaotic sequences and random sequences.
Due to the advantage of fast computation speed, the sophistication algorithms of SE and C0-algorithms are
commonly used to calculate the complexity of chaotic sequences. For a sequence {x(s), s = 0,1,2,K,S−1} with
8
Figure 10. The SE and C0-complexity of the state variable of system (1) with n = c = 0.3, p = 0.5, d = 10 and (x1(0),x2(0),
x3(0)) = (0.1,0,0). (a) The SE with respect m Î (0, 1). (b) The C0-complexity with respect m Î (0, 1).
length of S, the values of SE and C0 can be defined by equations (13) and (14) respectively.
S/2-1
⎧ SE = - å k = 0 P k ln P k
⎪
⎪P = ∣X (k)∣2
k S/2-1 (13)
⎨ å k = 0 ∣X (k)∣2
⎪ 2p
⎪ X (k) = å S x (s ) e -j S sk
⎩ s=0
S-1
⎧
C0 =
å s = 0 ∣x (s) - x˜(s)∣2
⎪ S-1
⎪ å s = 0 ∣x (s)∣2
⎪ 1 S-1 ˜ 2p
⎪ x˜(s ) =
S
å k=0
X (k) e j S sk
(14)
⎨
⎪ ⎧ X (k), ∣X (k)∣2 > 1 å S - 1∣X (k)∣2
⎪˜ ⎪ S s=0
⎪ X (k ) = ⎨ 1 S-1
⎪ 0 , ∣X (k)∣  S å s = 0 ∣X (k)∣
2 2
⎪
⎩ ⎩
where Pk is the power spectrum denstiy probability of sequence x(s), and X(k) is the sequence of x(s) after a
Fourier transform in equation (13). In equation (14), X˜ (k ) is the sequence of X(k) with the irregular parts
removed, and x˜(s ) is the sequence of X˜ (k ) after an inverse Fourier transform.
To maintain consistency with Lyapunov exponents, we set the system parameters as same as it defined in the
analyses of Lyapunov exponents, where the initial values equal to (0.1, 0, 0), and the parameters of system (1)
with equation (2) are fixed as n = c = 0.3, p = 0.5 and d = 10, and the range of parameter m chosen as m Î (0, 1),
then the curve of SE and C0-complexity of state variable of x3 sequence with parameter m is shown in figure 10.
According to figures 4 and 10, when the maximum Lyapunov exponents LE1  0, the values of SE and
C0-complexity are zero or close to zero, which means the system is stable or in periodic state. When
m Î (0, 0.08) È (0.76, 0.96), the value of C0 is less than 0.01. When m Î (0.08, 0.76), the values of SE and C0
are greater than zero, which are positively correlated with Lyapunov exponents value.
4. Electronic circuit design on Multisim10
The circuit realization is an effective means to verify the availability of a chaotic system, as a consequence, we designed
the electronic circuit diagram about system (1) on Multisim10. At first, the electronic circuits of the proposed
composite sine function are designed, then the electronic circuits for the new chaotic system are carried out.
4.1. Circuit design of the composite sine function

Based on equation (2), we have designed electronic circuits for the nonlinear function of composite sine function
g(x) with d = 10 and p = 0.5, and the circuit diagram and simulation results are presented in figure 11. In
figure 11(a), vo = x (1 - x 2/100) = x - x 3/100, g (x ) = - sin (x - x 3/100). Figure 11(b) shows the variation
of the composite sine function g(x) with the input of x, where the simulation results are displayed with a
9
Figure 11. The electronic circuits and waveform of composite sine function g(x) with d = 10 and p = 0.5. (a) Electronic circuit
diagram. (b) Simulation result.
horizontal ordinate unit of 2 seconds per division (2s/div) and a vertical ordinate unit of 1 volt per division
(1V/div).
Replace the resistor of R1 with 200 k W and keep the other parameters unchanged, the waveform of
composite sine function g (x ) with p = 0.5, d = 20 can be obtained and shown in figure 12. In figure 12(b), the
units of the horizontal and vertical ordinates are 2s/div and 1v/div, respectively.
Compare figure 11 with figure 12, it can be found that the simulation results of nonlinear function g (x ) with
different value of parameter of d are similar to each other. The parameter of d can be adjusted by replacing the
resistor of R1. For example, when R1 = 100k W and R2 = 10k W, then d = R1/R2 = 10, and if R1 = 200k W,
then d = R1/R2 = 20.
4.2. Circuit design of the proposed chaotic system with composite sine function
Based on the composite sine function circuits, the complete circuits of the new proposed chaotic system are
designed according to the equations (1) and (2), which is given in figure 13. The whole circuits consist of
resistors, capacitors, analog multipliers of ad633 and operational amplifiers uA741, which are used for operation
of addition and amplification. By connecting the signal g (x ) shown in figure 11(a) to the input port g (x )
depicted in figure 13, and the circuits illustrated in figure 13 can exhibit a 4-scroll chaotic attractor. Furthermore,
by connecting the signal g (x ) from figure 12(a) to the input port g (x ) in figure 13, a 6-scroll chaotic attractor can
be generated. The chaotic attractors generated by the circuits shown in figure 13 are presented in figure 14.
The circuit equations of circuits of figure 13 can be obtained according to circuit theory, and the equations
are described as follows:
⎧ R 4 C1 dX = R3 Y
⎪ dt R1
⎪
⎪ dY R
R10 C 2 = 9Z (15)
⎨ dt R8
⎪ dZ R13 R7 R R R R
⎪ R14 C3 = ( - g (X )) + 13 2 ( - Y ) + 13 ⎛- 16 Z ⎞ ⎜ ⎟
⎪ dt R6 R21 R12 R5 R11 ⎝ R15 ⎠
⎩
10
Figure 12. The electronic circuits and waveform of composite sine function g(x) with d = 20 and p = 0.5. (a) Electronic circuit
diagram. (b) Simulation result.
Figure 13. The circuit of the proposed chaotic system.
To ensure that both the circuit equation (15) and the chaotic system (1) have identical parameters, the values
of resistors and capacitor should be selected appropriately. For example, the resistance and capacitance are
chosen as those indicated in figure 13 of equation (1) with the parameter of m = n = c = 0.3 and equation (2)
with d = 10. Figure 14(a) displays the circuit simulation outcomes for the four-scroll chaotic attractor. Keeping
11
Figure 14. Electronic circuit simulation results with parameter m = n = c = 0.3, p = 0.5 and (x1(0),x2(0),x3(0)) = (0.1,0,0). (a) 4-scroll
chaotic attractor with d = 10. (b) 6-scroll chaotic attractor with d = 20.
Figure 15. Block diagram of the encryption algorithm.
m = n = c = 0.3 and parameter d selected as 20, the 6-scroll chaotic attractor generated by the circuits of figure 13
can be obtained and revealed in figure 14(b).
Compare figure 7(b) with 14(a) and figure 9(b) with figure 14(b), respectively, it can be found that the
simulation results of the circuits have an almost identical motion trajectory to the numerical simulation results.
Additionally, the quantity of scrolls produced by system (1) can be regulated by modifying the system parameter
of d, which corresponds to the resistor R1 in circuits of the composite sine function g (x ).
5. Image encryption based on DNA coding and the proposed system
In this section, we propose an encryption scheme based on the proposed novel system combined with DNA
coding, and it is given in figure 15. In addition, the flow chart of the decryption process is given in figure 16.
5.1. DNA coding operation

There are four types of deoxyribonucleotides in DNA, and represented by A, G, C and T, where A complements
T and C complements G, which is similar to that how 0 and 1 complement each other in binary. The four bases
are represented by binary numbers 00, 01, 10 and 11, and there are 24 combinations according to the
permutation combination, among which eight met the DNA pairing mode shown in table 1.
12
Figure 16. Block diagram of the decryption algorithm.
Table 1. Encoding and decoding rules of DNA.
1 2 3 4 5 6 7 8
A 11 11 10 10 01 01 00 00
T 00 00 01 01 10 10 11 11
C 10 01 11 00 11 00 01 10
G 01 10 00 11 00 11 10 01
5.2. Encryption scheme

In the block diagram of the decryption algorithm, the plaintext image is divided into three channels R, G and B,
which is partitioned and then the DNA coding operation is carried out on each block. In addition, the proposed
chaotic system sequence participates in DNA computation of blocks as well as scrambling operation.What is
more, Chen hyperchaotic system controls the rules for encoding, calculation and decoding of DNA.
Step1. Processing of images to be encrypted: Choose the color plaintext image E (i , j ) with size M ´ N,
block the image according to the block size b ´ b, if mod (M , b ) ¹ 0, let M ¢ = floor (M /b ) + 1, otherwise let
M ¢ = M, and N Perform a similar operation, the sizes of the three matrices after processing are M ¢ ´ N ¢ , and
the new matrices represent the new image E ¢ (i , j ), furthermore, E (i , j ) and E ¢ (i , j ) have the following
relatinship.
E ¢ (i , j ) = E (i , j ) , i  M , j  N
E ¢ (i , j ) = 0, i > M , j > N (16)
the image E ¢ (i , j ) is divided into three channels R, G and B, which are represented by two-dimensional matrix
E1, E 2 and E3 with size M ¢ ´ N ¢ respectively. Then divide the three matrices into K blocks according to
equation (17).
M¢ ´ N¢
K=
b´b
E1 (1: K ) = fdivide (E1, K , b)
E2 (1: K ) = fdivide (E2, K , b)
E3 (1: K ) = fdivide (E3, K , b) (17)
where K is the number of blocks, the size of each block is b ´ b, the fdivide represent segmentation of the image
matrix, and E1 (1: K ), E 2 (1: K ) and E3 (1: K ) represent the all blocks of image matrix after segmentation,
respectively.
Step2. The generation of chaotic sequences: Set the initial values of the newly proposed chaotic system as
x1 (0), x2 (0) and x3 (0), which are calculated from equation (18).
13
¢ ¢ ¢ ¢
å i = 1å j = 1 ´ E1 (i , j ) + å i = 1å j = 1 ´ E2 (i , j )
M N M N
⎧
⎪ x1 (0) =
⎪ 2 ´ 255 ´ M ¢ ´ N ¢
¢ ¢
⎪ M¢ N¢
å i = 1å j = 1 ´ E1 (i , j ) + å i = 1å j = 1 ´ E3 (i , j )
M N
(18)
⎨ x2 (0) = 2 ´ 255 ´ M ¢ ´ N ¢
⎪ ¢ ¢
M¢ N¢
å i = 1å j = 1 ´ E2 (i , j ) + å i = 1å j = 1 ´ E3 (i , j )
⎪ M N
⎪ x3 (0) =
⎩ 2 ´ 255 ´ M ¢ ´ N ¢
where E1 (i , j ), E 2 (i , j ) and E3 (i , j ) are pixel values that the plaintext image at the positiion of (i,j) respectively,
and the obtained initial values are brought into the system (1) to generate three chaotic sequences of x1, x2
and x3.
To avoid the transient characteristics of chaotic sequences, the first 10 000 data points of the chaotic
sequence are discarded. Subsequently, replace the original data with the decimal part from the 10th to 14th of
each data in the chaotic sequence, which are shown in equation (19), and fround and f round ¢ represent the values in
the chaotic sequence are kept to 14 and 10 decimal places, respectively.
¢ (x3 (10001: length (x3)) , 10)
x3 = fround (x3 (10001: length (x3)) , 14) - f round (19)
The values of this new sequence are then constrained to the range of 0 to 255, then the sequence p is obtained
as equation (20).
p = mod (x3 ´ 1014 , 256) (20)
Finally, the sequence p is transformed into a matrix R with size of M ¢ ´ N ¢ expressed as equation (21).
R = freshape ( p , N ¢, M ¢) ¢ (21)
where freshape represent the operation that converts a sequence to a matrix. Then divide the matrix R into K
blocks, which is employed for DNA computations with plaintext images.
R (1: K ) = fdivide (R , K , s ) (22)
A combination of SHA-512 and pixel values of images is used to generate a hash value S with 512-bit, then it
is divided into 64 decimal numbers, each represented by 8 bits. The calculation process is illustrated as shown in
equation (23), in which fSHA512 represent Secure Hash Algorithm.
¢ ¢ ¢ ¢ ¢ ¢
{
S = fSHA512 å iM= 1å Nj = 1 ´ E1 (i , j ) + å iM= 1å Nj = 1 ´ E2 (i , j ) + å iM= 1å Nj = 1 ´ E3 (i , j ) }
S = {s1, s2,,s64} (23)
The initial values for the Chen hyperchaotic system are determined by using the following equations:
⎧ X (0) = (s 01 Å s 02 Å Ås16) / 256

⎪ Y (0) = (s17 Å s18 Å Ås32) / 256
(24)
⎨ Z (0) = (s33 Å s34 Å Ås48) / 256
⎪ H (0) = (s Å s Å Ås ) / 256
⎩ 49 50 64
where the symbol Å is the Exclusive-or operation and following this, these initial values are incorporated into
the Chen hyperchaotic system in the manner described below:
⎧ X = 35(Y - X ) + X
⎪Y = 7X - XZ + 12Y
(25)
⎨ Z = XY - 3Z
⎪ 
⎩ H = YZ + 0.2H
where X , Y , Z and H are state variables, then four chaotic sequences: X, Y, Z, and H can be obtained.
Step3. DNA operation: The four chaotic sequences are processed according to equation (26), and the DNA
encoding/decoding and DNA calculation are set according to the values of the chaotic sequence.
X ¢ = fmod ( floor (X ´ 104) , 8) + 1
Y ¢ = fmod ( floor (Y ´ 104) , 8) + 1
Z ¢ = fmod ( floor (Z ´ 104) , 4) + 1
H ¢ = fmod ( floor (H ´ 104) , 8) + 1 (26)
where fmod represent modulo operation, X ¢ , Y ¢ , Z ¢ and H ¢ represent the integer obtained after the modulo
operation, respectively. The encoding of sub-blocks at the same position of E1, E 2 and E3 are determined by the
operation associated with the value of X ¢. Matrix R’s encoding rules are solely contingent upon the value of Y ¢.
14
Table 2. Calculation rules of DNA.
Z 1 2 3 4
Operator Addition Subtraction Exclusive-or Exclusive-nor
The coding rules are shown in table 1.

DNA E1 (1: K ) = fencoding (E1 (1: K ) , X ¢)
DNAR (1: K ) = fencoding (R (1: K ) , Y ¢) (27)
where DNA E1 (1: K ) , DNA E2 (1: K ) , DNA E3 (1: K ) and DNAR (1: K ) are the matrices encoded according to the rules of
the corresponding sequence, and the fencoding represents DNA encoding.
The value of Z ¢ governs the operation performed between the image E and matrix R. To achieve a more
pronounced diffusion effect, aside from the initial sub-block, the DNA calculation is reapplied to the encryption
result of the matrix E and R calculated by DNA.
DNA ER1((1:1:KK)) = fcalculation (DNA E1 (1: K ) , fcalculation (DNA E1 (1: K ) , DNAR (1: K ) , Z ¢) , Z ¢)
DNA ER2((1:1:KK)) = fcalculation (DNA E2 (1: K ) , fcalculation (DNA E2 (1: K ) , DNAR (1: K ) , Z ¢) , Z ¢)
DNA ER3((1:1:KK)) = fcalculation (DNA E3 (1: K ) , fcalculation (DNA E3 (1: K ) , DNAR (1: K ) , Z ¢) , Z ¢) (28)
where DNA ER1((1: K) R (1: K ) R (1: K )

1: K ) , DNA E 2 (1: K ) and DNA E3 (1: K ) are the matrices after DNA calculation respectively, and the
fcalculation represents DNA calculation. The calculation rules are shown in table 2.
Then the value of H ¢ determines the DNA decoding rules of the subblocks after calculation operation.
E1¢ (1: K ) = fdecoding (DNA ER1((1:1:KK)) , H ¢)
E2¢ (1: K ) = fdecoding (DNA ER2((1:1:KK)) , H ¢)
E3¢ (1: K ) = fdecoding (DNA ER3((1:1:KK)) , H ¢) (29)
where E1¢ (1: K ), E 2¢ (1: K ) and E3¢ (1: K ) are the matrices after DNA decoding, and the fdecoding represents DNA
decoding.
Step4. Get the ciphertext image: Finally, the subblocks of the three image matrices are merged into a whole
block, respectively.
E1¢¢ = fmerge (E1¢ (1: K ))

E2¢¢ = fmerge (E2¢ (1: K ))
E3¢¢ = fmerge (E3¢ (1: K )) (30)
where E1¢¢, E 2¢¢ and E3¢¢ represent the ciphertext matrix of R, G and B channels respectively. and the fmerge
represents operation of combination.
The sequences x1 and x2 generated by the proposed chaotic system are arranged in descending order to
obtain the sequence of positions Gx and Gy according to the following equations.
⎧Gx = fsort (x1, ¢descend ¢) (31)

⎩Gy = fsort (x2, ¢descend ¢)
⎨
where fsort represents the sequence values of Gx , Gy and their corresponding indexes are taken as the exchange
coordinates of rows and columns, respectively, and row and column permutation are carried out on the matrix
of E1¢¢, E 2¢¢ and E3¢¢.
E1C = fscramble (E1¢¢, Gx , Gy )
E2C = fscramble (E2¢¢, Gx , Gy )
E3C = fscramble (E3¢¢, Gx , Gy ) (32)
where E1C , E 2C and E3C represent the ciphertext matrix after the scramble of R, G and B channels respectively,
and fscramble represents operation of scramble.
15
Figure 17. Visualization of the encryption and decryption results for ‘house’: (a)Original image;(b) Ciphertext image;(c) Decrypted
image.
Subsequent to this process, the three two-dimensional matrices, having undergone extensive scrambling,
and amalgamated into a three-dimensional matrix, thereby resulting in the production of the ciphertext
image C.
5.3. DNA decoding operation

Figure 16 shows the decryption process, which involves performing the exact opposite operation on the
encrypted image as during encryption. Firstly, the sequences x2 and x3 of the proposed chaotic system participate
in column scrambling and row scrambling operations of ciphertext. Then, the ciphertext image is divided into
three channels of R, G, and B, and each channel is divided into k blocks, which are participate in DNA coding
operation with the sequence x1 of the proposed chaotic system. In addition, the Chen Hyperchaotic system
controls the rules of encoding, calculation and decoding of DNA just as it does with encryption.
It is crucial to note that the key used for decryption is identical to the one used for encryption, maintaining
the constancy of initial values and parameters in all chaotic systems. Additionally, in DNA calculation, the
computational rules employed are the reverse of those used during encryption. For instance, if addition was used
in encryption, subtraction is applied in decryption. Finally, the removal of the pixels added during encryption is
necessary to obtain the original image.
6. The performance analysis of the proposed encryption algorithm
6.1. Histogram analysis

The histogram provides a visual representation of how grayscale values are distributed within a photograph The
histogram visually displays the distribution of grayscale values. Along the horizontal axis, there is the entire range
of grayscale values [0, 255], while the vertical axis displays the frequency of each grayscale value appears in the
image. Generally, a graph’s histogram distribution is erratic, in contrast to the smooth histogram distribution of
ciphertext produced by encryption methods.
Figure 17 presents both the encryption and decryption outcomes for the picture of ‘house’. Figure 18 gives
the histogram analysis of these results, and it reveals a uniform distribution after encryption.
6.2. Information entropy analysis

Information entropy quantifies the level of information disorder within a system and serves as one method for
assessing the effectiveness of encryption. The value of information entropy can be calculated by equation (33).
2L - 1
I (U ) = - å (P (u i ) ´ log 2 P (u i )) (33)
i=1
where I(U) shows the randomness of message source U, L is the length of pixels and P(ui) is the probability
occurrence of ui. If the information entropy value of an image with grayscale value from 0 to 255 is closer to 8,
there is more disorder.
Table 3 provides an analysis of information entropy for various images, while table 4 offers information
entropy values of different cryptographic systems. In table 4, the proposed encryption algorithm exhibits an
information entropy value of 7.9993, which surpasses that of the other compared encryption algorithms.
16
Figure 18. Histogram Analysis of the plain image and cipher image results for ‘house’: (a1)-(a3)Histogram of plain images; (b1)–(b3)
Histogram of corresponding cipher images.
Table 3. Information entropy.
Plaintexts Ciphertexts
Images R G B R G B
house 7.253112 7.594036 6.968458 7.999312 7.999331 7.999255

Peppers 7.338842 7.496335 7.058372 7.999323 7.999254 7.999257
Baboon 7.706755 7.474463 7.752224 7.999241 7.999289 7.999271
Table 4. Information entropy for peppers image and

comparison with other existing algorithms.
Algorithm R G B
Proposed 7.999323 7.999254 7.999257

[33] 7.9972 7.9970 7.9971
[34] 7.9970 7.9972 7.9977
[35] 7.9964 7.9969 7.9969
6.3. Correlation analysis

Generally speaking, the inter-pixel correlation of a plaintext image is highly pronounced, however, it can be
effectively reduced through the encryption algorithm, so it can be used for preventing attackers inferring the
information from the picture. Quantitative evaluation of the correlation between the two points h and q can be
expressed as follows:
cov (h , q )
rhq = (34)
D (h ) D (q )
In pursuit of conducting correlation analysis, a total of 5000 pixels was randomly selected from the ‘house’
image. The results are represented visually in figure 19, elucidating the correlation patterns in three distinct
orientations pertaining to the plaintext image. Correspondingly, the correlation distributions of the encrypted
image are shown in figure 20. To further describe the pixels correlation of the original image and its encrypted
image, table 5 shows the correlation coefficients for the R, G, and B channels of the original and encrypted
images across the three diverse orientations. Additionally, table 6 gives the comparison of correlation
coefficients of the original and encrypted Pepper in R, G, B channels. The correlation point plot reveals that the
17
Figure 19. The correlation distribution of two adjacent pixels in the horizontal, vertical and diagonal directions of the R, G and B
channels of the original ‘house’: (a1)–(a3) R channel; (b1)–(b3) G channel; (c1)–(c3) B channel.
plaintext image exhibits a strong positive correlation with a coefficient close to 1, while the ciphertext image
shows a low correlation coefficient approached to 0. The findings demonstrate the algorithm’s efficacy in
reducing image correlation, which makes it can be used as a secure and dependable solution.
In order to understand the performance of this image encryption mechanisms in reducing the correlation
between pixels, the correlation coefficient of encrypted imaged of ‘peppers’ with different method are shown in
table 6. From table 6, it can be concluded that the proposed image encryption mechanism can effectively reduce
the correlation coefficient between pixels.
6.4. Key space analysis

An optimal image encryption algorithm should possess a significantly extensive key space, which makes it
impractical and cost-prohibitive to attempt a brute-force attack. In this algorithm, we utilize X (0), Y (0), Z (0),
H (0), x1 (0), x2 (0) and x3 (0) as cryptographic keys, its key space is 1098 » 2294 , surpassing the theoretical ideal of
2100 in cryptography. Additionally, this algorithm boasts an extensive key space, endowing it with a formidable
resistance against brute-force attacks.
Meanwhile, the results of different algorithms regarding the key space are listed in table 7, which indicates
that the key space of the proposed scheme has an advantage of against resisting the brute-force attack.
6.5. Key sensitivity analysis

When the encryption key undergoes slight changes, the ciphertext image cannot be decrypted, this suggesting
that the key’s sensitivity is relatively strong. Figure 21 illustrates the decryption process applied to the ciphertext
image ‘house’ with various key value modifications. In this figure, (a) and (b) represent the original ciphertext
image ‘house’ and the resulting decrypted image when the correct key is used, respectively. Meanwhile, (c) shows
18
Figure 20. The correlation distribution of two adjacent pixels in the horizontal, vertical and diagonal directions of the R, G and B
channels of the encrypted ‘house’: (a1) -(a3) R channel; (b1)-(b3) G channel; (c1)-(c3) B channel.
Table 5. The correlation coefficient of the plaintexts and the cipertexts.
Plaintexts Ciphertexts
R G B R G B
Horizontal 0.98045 0.97134 0.93873 −0.01833 −0.02098 −0.00671

Vertical 0.98994 0.98357 0.96092 −0.01802 0.00423 −0.02095
Diagonal 0.97101 0.9565 0.91994 0.03185 0.01504 0.02280
the decrypted image obtained after introducing a perturbation of 10-14 to the key x1(0); (d), (e), (f), (g) and (h)
are the decrypted image obtained after adding the perturbation amount of 10-14 to different key, respectively.
In this algorithm, the strong sensitivity to key disturbances is evident from the fact that even a slight
disturbance results in a completely different decryption outcome compared to the original image. This
characteristic highlights its ability to effectively resist exhaustive key attacks.
6.6. Differential attack analysis

An attacker can attempt to decrypt the encryption algorithm by making small changes to the image and
observing the changes in the encryption result. Therefore, a suitable image encryption scheme must be able to
withstand differential attacks. To defend against differential attacks, the dependence of the encryption algorithm
on the plaintext image is usually evaluated by using the pixel change rate (NPCR) and the uniform average
change intensity (UACI), and they are defined by equations (35) and (36) respectively.
19
Table 6. Comparison of correlation coefficients.
Image
Channel Horizontal Vertical Diagonal
Peppers R 0.98045 0.98994 0.97101

Original
G 0.97134 0.98357 0.95650
B 0.93873 0.96092 0.91994
Proposed R −0.01833 −0.01802 0.03185
scheme
G −0.02098 0.00423 0.01504
B −0.00671 −0.02095 0.02280
[36] R 0.0002 0.0131 −0.0015
G −0.0032 0.0072 0.0071
B −0.0120 0.0080 0.0103
[37] R 0.0385 −0.0129 0.0129
G 0.0027 0.0153 −0.0007
B 0.0017 −0.0099 0.0323
[38] R −0.0048 0.0049 0.0006
G 0.0140 0.0075 −0.0012
B 0.0065 0.0182 0.0025
[39] R −0.0145 0.0071 0.0200
G −0.0149 0.0136 0.0067
B 0.0393 −0.0019 −0.0038
Table 7. Comparison of key space sizes.
Algorithms Proposed [40] [41] [42] [43] [44]
Key space 2294 2199 2194 2189 2186 2156
1 Sm Sn
NPCR (b1, b 2) = åå (∣f (b1 (i, j ) - b2 (i, j ))∣) ´ 100%
Sm Sn i = 1 j = 1 sign
(35)
1 Sm Sn (∣ b1 (i , j ) - b 2 (i , j )∣)
UACI (b1, b 2) = åå
Sm Sn i = 1 j = 1 255
´ 100% (36)
⎧ 1, i > 0
fsign (i ) = 0, i = 0 (37)
⎨
⎩ 1, i < 0
-
where Sm is the width of the plain image and Snis the height of the plain image. b1 (i , j ) and b 2 (i , j ) are the
position of cipher images before and after modifying the pixel values of the plain image, respectively. To meets
the safety standard, the ideal value of NPCR is 99.6093%, and the ideal value of UACI is 33.4635%. In our
experiments, we firstly change the pixel values of the plaintext image, then get the ciphertext image using the
proposed encryption algorithm, and finally calculate the NPCR and UACI of the ciphertext image encrypted
from the original image and the ciphertext image generated from the original image with pixel values changed.
As shown in table 8, the results indicate that the NPCR and UACI of various images are almost close to the
desired values. In additon, table 9 lists the NPCR and UACI values of the image ‘peppers’ for the proposed
method and other methods. It can be obtained that this algorithm is closer to the ideal value than the other
algorithms.
6.7. Robustness analysis

Robustness is an important indicator of anti-interference ability of the image encryption system. When the
transmission channel is affected by noise or cropping attacks, some information about the ciphertext image will
distort or lose. In security analysis of the encryption system, the main work of robustness analysis is anti-clipping
and anti-noise tests. If the ciphertext image can resist noise attacks and cropping attacks, the information loss
will not affect the decryption too much, and most of the information can still be obtained after decrypting the
affected ciphertext image.
In the process of digital image transmission, the most vulnerable interference is the addition of noise. Noise
is a typical unintentional attack, which will affect the encrypted image and cause the image quality to deteriorate.
Therefore, salt and pepper noises (SPN) with intensity of 1%, 3%, 5% and 10% were added to the ciphertext
20
Figure 21. The decryption diagram obtained by decrypting ciphertext image ‘house’ after changing different key values. (a) ciphertext
image; (b) decrypted image with correct keys; (c) x1(0) = 0.4285+10−14; (d) x2(0) = 0.5378+10−14; (e) x3(0) = 0.0672+10−14; (f) X
(0) = 0.5383+10−14; (g) Y(0) = 0.4461+10−14; (h) Z(0) = 0.5879+10−14; (i) H(0) = 0.7290+10−14.
Table 8. The NPCR and UACI for different images.
NPCR(%) UACI(%)
Images R G B Average R G B Average
House 99.6076 99.6037 99.6095 99.6069 33.4630 33.5165 33.6439 33.5411

Peppers 99.5710 99.5762 99.6104 99.5859 33.2603 33.6130 33.6413 33.5049
Baboon 99.6107 99.6039 99.6079 99.6075 33.4625 33.5167 33.6439 33.5410
image to test the ability of the proposed image encryption scheme to resist noise attacks.The test results are
shown in figure 22.
The results show that after adding noise of different intensities to the encrypted image, the quality of the
corresponding decrypted image decreases with the increase of noise intensity. After adding noise in a 10%
intensity, the proposed algorithm can still resist noise attacks. Therefore, the algorithm has strong robustness to
noise attacks.
When ciphertext image is damaged during transmission, some image data will be lost, resulting in the
received ciphertext image is incomplete. A high-quality image encryption algorithm should have the ability to
21
Figure 22. Noise attacks analysis for ‘House’. (a1)–(a4) Ciphertext under different noise attacks; (b1)–(b4) The corresponding
decrypted images.
Table 9. The NPCR and UACI of image peppers with other algorithms.
NPCR(%) UACI(%)
R G B Average R G B Average
Proposed 99.6107 99.6039 99.6079 99.6075 33.4625 33.5167 33.6439 33.5410

[45] 99.5712 99.5758 99.6094 99.5855 33.1056 30.5178 27.5385 30.3873
[46] 99.6243 99.6185 99.6280 99.6236 33.4224 33.4361 33.4603 33.4396
[47] 99.6109 99.6109 99.6375 99.6197 33.4158 30.3902 33.2420 32.3493
[48] 99.6110 99.6041 99.6084 99.6078 33.4592 33.4626 33.4548 33.4588
22
Figure 23. Cropping attacks analysis for ‘House’; (a1)–(a4) Ciphertext subjected to varying degrees of clipping; (b1)–(b4) The
corresponding decrypted images.
resist data loss. Therefore, the ciphertext image may be cropped to different degrees, so the ability of the
proposed image encryption scheme to resist cropping attacks is tested. The simulation results of the anti-
clipping attack are shown in figure 23. From the results, the decrypted images can still contain the main
information of the plaintext images.
In order to further analyze the anti-noise attack and anti-cropping attack ability of the proposed image
encryption algorithm, we calculated the Peak signal-to-noise ratio (PSNR) and the Structural similarity index
measurement (SSIM) values between the plaintext image and the corresponding decrypted image. The
calculation formulas of PSNR and SSIM are described as equations (38) and (39) respectively. It is worth noting
that the caculation of PSNR require the value of mean-square error (MSE), which fomula is presented as
equation (40). In equation (40), Sm and Sn are the height and width of the original plaintext image, P1(i, j)
23
Table 10. SSIM between the plain images and the decrypted images under the
Nosie and Cropping attacks.
Noise
Test image level Proposed Cropping size Proposed
House 0.01 0.8298 32*32 0.9616

0.03 0.6410 48*48 0.9256
0.05 0.5327 64*64 0.8791
0.1 0.3839 128*128 0. 6658
Baboon 0.01 0.9231 32*32 0.9810
0.03 0.8068 48*48 0.9589
0.05 0.7223 64*64 0.9257
0.1 0.5648 128*128 0.7644
Peppers 0.01 0.9370 32*32 0.9842
0.03 0.8363 48*48 0.9655
0.05 0.7593 64*64 0.9410
0.1 0.6112 128*128 0.8201
Table 11. PSNR between the plain images and the decrypted images under the
SPN and Cropping attacks.
SPN
Test image level Proposed Cropping size Proposed
House 0.01 38.7721 32*32 43.2211

0.03 34.0805 48*48 39.9011
0.05 31.8876 64*64 37.3086
0.1 28.9779 128*128 31.6335
Baboon 0.01 39.6716 32*32 44.3148
0.03 35.0542 48*48 40.7369
0.05 32.8321 64*64 38.1875
0.1 29.9248 128*128 32.5201
Peppers 0.01 40.4233 32*32 44.6165
0.03 35.6030 48*48 41.2589
0.05 33.4484 64*64 38.8882
0.1 30.5589 128*128 33.0529
represents the pixel value of the original plaintext image at position (i, j), and P2(i, j) represents the pixel value of
the decrypted image at position (i, j) after the image information is attacked.
2552 ⎞
PSNR = 10 log10⎛ ⎜ ⎟ (38)
⎝ MSE ⎠
(2P1 P2 + C1)(2sP1 sP2 + C 2)

SSIM = (39)
(P1 2 + P2 2 + C1)(s 2P1 + s 2P2 + C 2 )
å i= 1å j= 1[P1 (i, j ) - P2 (i, j )]2

Sm Sn
MSE = (40)
Sm ´ Sn
In equation (39), C1 = (K1 L )2 , C2 = (K2 L )2 , K1 = 0.01, K2 = 0.03, L = 255.The means of P1 and P2 are P1
and P2 , respectively. And s 2P1 and s 2P2 represent the variances of P1 and P2, respectively.
In general, the larger the PSNR value, the smaller the distortion between the decrypted image and the
plaintext image, and the better the quality of the decrypted image. In image encryption scheme, it is generally
believed that if the PSNR value is greater than 30dB, the decrypted image can be considered to be highly similar
to the plaintext image. The range of SSIM values is [0,1], and the larger the value of SSIM, the more the encrypted
image similars to its plaintex image. If the SSIM value is equal to 1, the decrypted image can be considered to be
exactly the same as the plaintext image.
The anti-noise attack and anti-cropping attack abilities of the proposed image encryption algorithm are
tested by using different images, and the simulation results are shown in tables 10 and 11. Furthermore, the anti-
noise attack and anti-cropping attack abilities of this image encryption algorithm are compared with others, and
their results are shown in table 12.
24
Table 12. Comparison of the capablility to resist SPN and Cropping attacks.
PSNR
Attack
Attack
Type Intensity [49] [50] [51] Proposed
0.0001 33.44 31.56 28.18 61.24

SPN 0.0003 33.26 30.22 28.18 55.71
0.0005 33.02 30.02 28.17 53.42
32*32 24.92 27.21 30.18 44.61
Cropping 48*48 20.13 23.96 29.01 41.25
64*64 — — — 38.88
128*128 — — — 33.05
Table 13. Runing time of different images.
Image Encryption time (s) Decryption time (s)
House 1.45 1.50

Peppers 1.43 1.49
Baboon 1.47 1.48
Table 14. Runing time of image peppers compared to different algorithms.
Algorithm Encryption time (s) Decryption time (s) Device specs. (CPU and RAM)
Proposed 1.45 1.49 1.8 GHz Intel(R) Core(TM) i5 8GB

[52] 1.65 1.63 2.9 GHz Intel(R) Core(TM) i9 32GB
[53] 1.33 1.30 2.3 GHz Intel(R) Core(TM) i5 8G
[54] 16.43 16.43 2.16 GHz Intel(R) Core(TM) 4GB
6.8. Encryption/decryption speed analysis

The time required for encryption is a measurement of the complexity of the proposed algorithm. The
experiment in this paper runs on desktop PC with Intel (R) Core (TM) 1.80 GHz, 8 GB RAM and 500 GB hard
disk. The operating system and computational plat-form are Microsoft Windows 10 and Matlab R2018b
respectively. Table 13 shows the encryption time of 3 images with dimensions of 512 × 512, and the times in the
tables are all the average time of 100 experiments. Table 14 provides the encryption time and decryption time
compared the proposed algorithm and its counterparts from the literature.
7. Conclusions
In this paper, a novel multi-scroll chaotic system with a composite sine function based on jerk system is
proposed. The number of the scrolls generated by the novel chaotic system can be controlled by the parameter d
of the composite sine function. Moreover, as to the implementing circuits of multi-scroll chaotic attractors, only
one resistor’s resistance is required to adjust for altering the number of scrolls. Compared with the chaotic
system with saturation function, triangular wave, hysteresis function, step function, or piecewise linear function,
the proposed system has the advantage of simpler circuit structure, which means that the circuit implementation
is easier. In addition, the adjustment of the scroll number is more convenient. Besides, the introduction of
complexity testing verifies the occurrence of system chaos. The numerical simulation results are coinciding with
the electronic circuit simulation ones, so the proposed new chaotic system with composite sine function is
feasible. Additionally, we combine the proposed chaotic system with Chen hyperchaotic system and DNA
coding to propose a new image encryption method, which ensures the encryption efficiency while increasing the
key space. The four DNA operations rules are controlled by the Chen hyperchaotic system, and the proposed
chaotic system is used to DNA computation and scrambled the images. Finally, the ciphertext histogram of this
encryption system is evenly distributed, the information entropy reaches 7.9993, the key sensitivity reaches the
order of 10-14 , which signifies that the algorithm exhibits robust security characteristics.
25
Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).
Future works
In this paper, the dynamic behavior of the proposed multi-scroll chaotic system as well as its application in image
encryption are analyzed theoretically. In addition, the feasibility of circuit realization and image encryption is
verified by Multisim and Matlab software. However, the time comlexity and time consumption of the proposed
image encrytion/decryption scheme is relatively high, it is difficult to meets real-time requirement. In the
future, we will look for solutions to reduce time complexity and time consumption, and try to implement its
application through FPGA.
Competing interests
The authors declare no conflict of interest.
Authors contributions
Conceptualization, P F Ding; methodology P F Ding; software, P F Ding; validation, P F Ding and K Li; formal
analysis, P F Ding, K Li. and Z X Wang; writing original draft, P F Ding; writing-review and editing, K Li. and Z X
Wang. All authors have read and agreed to the published version of the manuscript.
ORCID iDs
Pengfei Ding https://orcid.org/0000-0002-3297-4438

Ke Li https://orcid.org/0009-0001-1733-8468
References
[1] Ali Z, Imran M, Alsulaiman M, Shoaib M and Ullah S 2018 Chaos-based robust method of zero-watermarking for medical signals
Futur. Gener. Comp. Syst. 88 400–12
[2] Wu Q and Wu M 2018 Adaptive and blind audio watermarking algorithm based on chaotic encryption in hybrid domain Symmetry-
Basel 10 284
[3] Chen L, Chen J, Zhao G and Wang S 2019 Cryptanalysis and improvement of a chaos-based watermarking scheme IEEE Access 7 97549
[4] Liao T, Wan P and Yan J 2019 Design of synchronized large-scale chaos random number generators and its application to secure
communication Appl. Sci.-Basel 9 185
[5] Yu F, Li L, Tang Q, Cai S, Song Y and Xu Q 2019 A survey on true random number generators based on chaos Discrete Dyn. Nat. Soc.
2019 2545123
[6] Rezk A, Madian A, Radwan A and Soliman A 2020 Multiplierless chaotic Pseudo random number generators AEU-Int. J. Electron.
Commun. 113 152947
[7] Wang T, Song L, Wang M, Chen S and Zhuang Z 2020 A novel five-dimensional three-leaf chaotic attractor and its application in image
encryption Entropy 22 243
[8] Shao Z, Liu X, Yao Q, Qi N, Shang Y and Zhang J 2020 Multiple-image encryption based on chaotic phase mask and equal modulus
decomposition in quaternion gyrator domain Signal Process.-Image Commun. 80 115662
[9] Luo Y, Tang S, Liu J, Cao L and Qiu S 2020 Image encryption scheme by combining the hyper-chaotic system with quantum coding
Opt. Lasers Eng. 124 105836
[10] Liu J, Yang X, Gao Y and Liu F 2016 Application of similar Liu system in underwater weak signal detection Acta Phys. Sin. 65 070501
[11] Cao B, Li P, Li X, Zhang X, Ning W, Liang R, Li X, Hu M and Zheng Y 2019 Detection and parameter estimation of weak pulse signal
based on strongly coupled Duffing oscillators Acta Phys. Sin. 68 080501
[12] Wu A, Mwachaka S, Pei Y and Fu Q 2018 A novel weak signal detection method of electromagnetic LWD based on a duffing oscillator
J. Sens. 2018 5847081
[13] Wang F Q and Liu C X 2007 A new multi-scroll chaotic generator Chin. Phys.B 16 942
[14] Zhang C X and Yu S M 2009 Design and implementation of a novel multi-scroll chaotic system Chin. Phys.B 18 119
[15] Sara D and Hamid R M 2010 Generating one-, two-, three- and four-scroll attractors from a novel four-dimensional smooth
autonomous chaotic system Chin. Phys.B 19 060506
[16] Luo X H, Tu Z W, Liu X R, Cai C, Liang Y L and Gong P 2010 Implementation of a novel two-attractor grid multi-scroll chaotic system
Chin. Phys. B 19 070510
[17] Hu X, Liu C, Liu L, Yao Y and Zheng G 2017 Multi-scroll hidden attractors and multi-wing hidden attractors in a 5-dimensional
memristive system Chin. Phys. 26 110502
[18] Zhang G T and Wang F Q 2018 A novel multi-scroll chaotic generator_ Analysis, simulation, and implementation Chin. Phys. B 27
018201
26
[19] Echenausia-Monroy J L, Huerta-Cuellar G, Jaimes-Reategui R, Garcia-Lopez J H, Aboites V, Cassal-Quiroga B B and

Gilardi-Velazquez H E 2020 Multistability emergence through fractional-order-derivatives in a PWL multi-scroll system Electronics
9 880
[20] Dong E Z, Li R H and Du S Z 2021 A multi-directional controllable multi-scroll conservative chaos generator: modelling, analysis, and
FPGA implementation Chin. Phys. B 30 020505
[21] Yalcin M E 2007 Multi-scroll and hypercube attractors from a general jerk circuit using Josephson junctions Chaos Solitons Fractals 34
1659–66
[22] Yu S M, Lü J H, Leung H and Chen G R 2005 Design and implementation of n-scroll chaotic attractors from a general Jerk circuit IEEE
Trans. Circuits Syst. I-Regul. Pap. 52 1459–67
[23] Luo X 2009 Circuitry implementation of a novel nonautonomous hyperchaotic Liu system based on sine input Chin. Phys.B 18 3304
[24] Li F and Ma J 2016 Pattern selection in network of coupled multi-scroll attractors PLoS One 11 e0154282
[25] Ding P F, Feng X and Fa L 2020 Generation of 3-D multi-scroll chaotic attractors based on sign function and sine function Electronics
9 2145
[26] Wang X Y, Feng L and Zhao H Y 2019 Fast image encryption algorithm based on parallel computing system Inf. Sci. 486 340–58
[27] Hagras T, Salama D and Youness H 2022 Anti-attacks encryption algorithm based on DNA computing and data encryption standard
Alex. Eng. J. 61 11651–62
[28] Zhang Q and Han J 2021 A novel color image encryption algorithm based on image hashing, 6D hyperchaotic and DNA coding
Multimed.Tools Appl. 80 13841–64
[29] Ma J, Wu X Y, Chu R T and Zhang L P 2014 Selection of multi-scroll attractors in Jerk circuits and their verification using Pspice
Nonlinear Dyn. 76 1951
[30] Tang W K S, Zhong G Q, Chen G R and Man K F 2001 Generation of n-scroll attractors via sine function IEEE Trans. Circuits Syst.
I-Fundam. Theor. Appl. 48 1369
[31] Ding P F and Feng X 2020 Generation of multi-scroll chaotic attractors from a Jerk circuit with a special form of a sine function
Electronics 9 842
[32] Ding P F, Feng X and Wu C M 2020 Novel two-directional grid multi-scroll chaotic attractors based on the Jerk system Chin. Phys. B 29
108202
[33] Parnab D and Santanu M 2023 A physical memristor-based chaotic system and its application in colour image encryption scheme Phys.
Scr. 98 115252
[34] Khan H, Jamal S S, Hazzazi M M, Khan M and Hussain I 2023 New image encryption scheme based on Arnold map and cuckoo search
optimization algorithm Multimed Tools Appl. 82 7419–41
[35] Zhang Y Q, He Y, Li P and Wang X Y 2020 A new color image encryption scheme based on 2DNLCML system and genetic operations
[36] Cheng G, Wang C and Chen H 2019 A novel color image encryption algorithm based on hyperchaotic system and permutation
diffusion architecture Int. J. Bifurcation Chaos 29 1950115
[37] Movafegh Ghadirli H, Nodehi A and Enayatifar R 2021 Color image dna encryption using mrna properties and non-adjacent
coupledmap lattices Multimedia Tools Appl. 69 808445
[38] Kang X and Guo Z 2020 A new color image encryption scheme based on DNA encoding and spatiotemporal chaotic system Signal 80
115670
[39] Hosny K M, Kamal S T and Darwish M M 2022 A color image encryption technique using block scrambling and chaos Multimedia
Tools Appl. 81 505–25
[40] Lee W K, Phan R, Yap W S and Goi B M 2018 Spring: a novel parallel chaos-based image encryption scheme Nonlinear Dynam. 92
575–93
[41] Imran Haider M, Ali A, Shah D and Shah T 2021 Block cipher’s nonlinear component design by elliptic curves: an image encryption
application Multimed. Tool. Appl. 80 4693–718
[42] Sahari M L and Boukemara I 2018 A pseudo-random numbers generator based on a novel 3d chaotic map with an application to color
image encryption Nonlinear Dynam. 94 723–44
[43] Wang X Y and Li Z M 2019 A color image encryption algorithm based on hopfield chaotic neural network Opt Laser. Eng. 115 107–18
[44] Ullah A, Jamal S S and Shah T 2018 A novel scheme for image encryption using substitution box and chaotic system Nonlinear Dynam.
91 359–70
[45] Alexan W, Alexan N and Gabr M 2023 Multiple-layer image encryption utilizing fractional-order chen hyperchaotic map and
cryptographically secure PRNGs Fractal Fract. 7 287
[46] Zhang Y Q, He Y, Li P and Wang X Y 2020 A new color image encryption scheme based on2DNLCML system and genetic operations
[47] Alexan W, ElBeltagy M and Aboshousha A 2022 RGB image encryption through cellular automata S-box and the lorenz system
Symmetry 14 443
[48] Huang L, Sun Y, Xiang J and Wang L 2022 Image encryption based on a novel memristive chaotic system, Grain-128a algorithm and
dynamic pixel masking Journal of Systems Engineering and Electronics 33 534–50
[49] Zhu L, Jiang D, Ni J, Wang X, Rong X and Ahmad M 2022 A visually secure image encryption scheme using adaptive thresholding
sparsification compression sensing model and newly-designed memristive chaotic map Inf. Sci. 607 1001–22
[50] Wang H, Xiao D, Li M, Xiang Y and Li X 2019 A visually secure image encryption scheme based on parallel compressive sensing Signal
Process. 155 218–32
[51] Zhu L, Song H, Zhang X, Yan M, Zhang T, Wang X and Xu J 2020 A robust meaningful image encryption scheme based on block
compressive sensing and SVD embedding Signal Process. 175 107629
[52] Gabr M, Elias R, Hosny K M, Papakostas G A and Alexan W 2023 Image encryption via base-n PRNGs and parallel base-n S-boxes IEEE
Access 11 85002–30
[53] Wang X, Xu X and Sun K 2023 A color image encryption and hiding algorithm based on hyperchaotic system and discrete cosine
transform Nonlinear Dyn. 111 14513–36
[54] Jithin K C and Sankar S 2020 Colour image encryption algorithm combining Arnold map, DNA sequence operation, and a Mandelbrot
set J. Inf. Secur. Appl. 50 102428
27

Generation Multi-Scroll Chaotic Attractors Using C

Uploaded by

Copyright:

Available Formats

Generation Multi-Scroll Chaotic Attractors Using C

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Generation Multi-Scroll Chaotic Attractors Using C

Uploaded by

Copyright:

Available Formats

Phys. Scr. 99 (2024) 045221 https://doi.org/10.

Generation multi-scroll chaotic attractors using composite sine

© 2024 IOP Publishing Ltd

2. Mathematical model of the novel chaotic system

3. Dynamical characteristics of the chaotic system

3.1. Equilibrium point

x13 - 100x1 + 100k = 0 (10)

3.2. Bifurcation diagram and Lyapunov exponents

3.3. Phase portraits

3.4. Sophistication test

4. Electronic circuit design on Multisim10

4.1. Circuit design of the composite sine function

Figure 13. The circuit of the proposed chaotic system.

Figure 15. Block diagram of the encryption algorithm.

5. Image encryption based on DNA coding and the proposed system

5.1. DNA coding operation

Figure 16. Block diagram of the decryption algorithm.

Table 1. Encoding and decoding rules of DNA.

5.2. Encryption scheme

⎧ X (0) = (s 01 Å s 02 Å Ås16) / 256

Table 2. Calculation rules of DNA.

Operator Addition Subtraction Exclusive-or Exclusive-nor

The coding rules are shown in table 1.

where DNA ER1((1: K) R (1: K ) R (1: K )

E1¢¢ = fmerge (E1¢ (1: K ))

⎧Gx = fsort (x1, ¢descend ¢) (31)

5.3. DNA decoding operation

6. The performance analysis of the proposed encryption algorithm

6.1. Histogram analysis

6.2. Information entropy analysis

Table 3. Information entropy.

house 7.253112 7.594036 6.968458 7.999312 7.999331 7.999255

Table 4. Information entropy for peppers image and

Proposed 7.999323 7.999254 7.999257

6.3. Correlation analysis

6.4. Key space analysis

6.5. Key sensitivity analysis

Table 5. The correlation coefﬁcient of the plaintexts and the cipertexts.

Horizontal 0.98045 0.97134 0.93873 −0.01833 −0.02098 −0.00671

6.6. Differential attack analysis

Table 6. Comparison of correlation coefﬁcients.

Peppers R 0.98045 0.98994 0.97101

Table 7. Comparison of key space sizes.

Algorithms Proposed [40] [41] [42] [43] [44]

Key space 2294 2199 2194 2189 2186 2156

6.7. Robustness analysis

Table 8. The NPCR and UACI for different images.

Images R G B Average R G B Average

House 99.6076 99.6037 99.6095 99.6069 33.4630 33.5165 33.6439 33.5411

Proposed 99.6107 99.6039 99.6079 99.6075 33.4625 33.5167 33.6439 33.5410

House 0.01 0.8298 32*32 0.9616

House 0.01 38.7721 32*32 43.2211

(2P1 P2 + C1)(2sP1 sP2 + C 2)

å i= 1å j= 1[P1 (i, j ) - P2 (i, j )]2

0.0001 33.44 31.56 28.18 61.24

Table 13. Runing time of different images.

Image Encryption time (s) Decryption time (s)

House 1.45 1.50

Table 14. Runing time of image peppers compared to different algorithms.