Applied Computing and Informatics xxx (xxxx) xxx

Contents lists available at ScienceDirect

Applied Computing and Informatics

journal homepage: www.sciencedirect.com

An audio encryption scheme based on Fast Walsh Hadamard Transform

and mixed chaotic keystreams
F.J. Farsana a,⇑, V.R. Devi a, K. Gopakumar b
a
Department of Electronics and Communication, LBS Centre for Science and Technology, Kerala University, Trivandrum, Kerala, India
b
Department of Electronics and Communication, TKM College of Engineering, Kollam, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper introduces an audio encryption algorithm based on permutation of audio samples using dis-
Received 5 August 2019 crete modified Henon map followed by substitution operation with keystream generated from the mod-
Revised 4 October 2019 ified Lorenz-Hyperchaotic system. In this work, the audio file is initially compressed by Fast Walsh
Accepted 7 October 2019
Hadamard Transform (FWHT) for removing the residual intelligibility in the transform domain. The
Available online xxxx
resulting file is then encrypted in two phases. In the first phase permutation operation is carried out using
modified discrete Henon map to weaken the correlation between adjacent samples. In the second phase it
Keywords:
utilizes modified-Lorenz hyperchaotic system for substitution operation to fill the silent periods within
Audio encryption
Fast Walsh Hadamard Transform
the speech conversation. Dynamic keystream generation mechanism is also introduced to enhance the
Modified Henon map correlation between plaintext and encrypted text. Various quality metrics analysis such as correlation,
Modified Lorenz hyperchaotic system signal to noise ratio (SNR), differential attacks, spectral entropy, histogram analysis, keyspace and key
sensitivity are carried out to evaluate the quality of the proposed algorithm. The simulation results
and numerical analyses demonstrate that the proposed algorithm has excellent security performance
and robust against various cryptographic attacks.
Ó 2019 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction developing cryptographic algorithms. Claude Shannon introduces

two basic elements for secure cryptographic algorithms [1]. These
Voice based communication becomes prominent in several elements are confusion (permutation) and diffusion (substitution)
areas such as military, phone banking, confidential voice confer- operations. In confusion operation, data samples are permuted
encing, education etc. With the increasing need for secure speech according to some specific key parameters to destroy the local cor-
communication, data encryption protocols are critically important relation between adjacent samples. While in diffusion operation,
for storage and transmission of sensitive information over exposed data samples are substituted with pseudo random numbers (PRNs)
systems. Unlike text and message signals, adjacent samples of generated by some entropy sources, to change the sample values.
voice signals are highly correlated and slowly time-varying. More- Both these operations eventually strengthen the complex relation-
over, the presence of redundant and unvoiced samples in audio sig- ship among plaintext, ciphertext and symmetric key parameters.
nal demands the need for efficient compression techniques in the While developing symmetric key encryption algorithm, designers
transform domain. Therefore, the conventional cryptographic algo- utilize substitution- permutation network as the basic structural
rithms are poorly suited for speech encryption. With the advance- element [2]. Chaos theory plays a significant role in developing
ment of security level, chaotic-systems play a significant role in encryption algorithms, due to its inherent properties such as topo-
logical transitivity, ergodicity, sensitive dependence on initial con-
ditions and deterministic pseudo randomness. These eventually
⇑ Corresponding author.
satisfy the basic requirement for theoretical cryptography. Most
E-mail addresses: [email protected] (F.J. Farsana), gopakumar@
of the chaos based algorithms such as multiple iterations of chaotic
tkmce.ac.in (K. Gopakumar).
map [3,4], bit-level scrambling approaches [5,6] bit-level confusion
Peer review under responsibility of King Saud University.
methods [7] and hybrid key methods [8] were designed based on
the above mentioned properties. Also, single and multiple
round of permutation-then-diffusion without substitution-box
Production and hosting by Elsevier (PT-DWOS), DNA encoding methods are introduced based on the

https://doi.org/10.1016/j.aci.2019.10.001
2210-8327/Ó 2019 The Authors. Production and hosting by Elsevier B.V. on behalf of King Saud University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
2 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx

reproducibility and deterministic nature of chaotic functions, since observed in the above mentioned literature are summarized as
the process can be repeated for the same function and same initial follows:
conditions. To improve the complexity and randomness of encryp-
tion scheme various chaotic systems are introduced e.g. discrete 1. Inefficient compression method to remove the unvoiced data
time [9], continuous time [10], hyper chaotic [11] and time delayed segments.
systems [12]. Cascading of different chaotic systems [13], itera- 2. Suggested Permutation methods are not strong enough to break
tively expanding lower dimensional chaotic map to higher dimen- the correlation between adjacent samples in the audio file.
sion [14], parametric perturbations of chaotic trajectories [15] are 3. Lower Dimensional chaotic maps have periodic window prob-
the common methodologies followed by the designers to develop lems like smaller chaotic range and non-uniform distribution.
algorithm based on chaos theory. Furthermore, chaos theory have 4. In substitution-permutation rounds, permutation matrix and
been incorporated in many conventional cryptographic approaches keystream generated from the chaotic function depends only
like S-box design [16], RC5 stream cipher [17] and elliptical curve on the initial condition and control parameters of the chaotic
cryptography [18] to strengthen the security of the encryption pro- map.
cesses. But these methods are flawed with limited keyspace and
computational complexity. Recently, researchers have attempted To overcome the above mentioned drawbacks, we propose an
to develop computationally efficient and unconditionally secure audio encryption algorithm based on chaotic maps based on mod-
chaotic-quantum algorithms suitable for cloud and Internet of ified Henon map and modified Lorenz-hyperchaotic system. To
Things (IoT) environments [19–22]. overcome the first drawback, the algorithm employs a signal com-
Several audio encryption algorithms have been introduced to pression mechanism by Fast Walsh Hadamard Transform (FWHT),
provide secure data transmission. Among these techniques voice which reduces the sample size for further encryption process by
encryption algorithm based on chaos theory are considered to be removing higher order coefficients. Unlike Fast Fourier Transform
effective to handle with redundant and bulky audio files. An over- (FFT) and Discrete Cosine Transform (DCT), FWHT has excellent
view of speech encryption algorithm based on chaos theory is dis- energy compression properties. Furthermore, rectangular basis
cussed here after. Long Jye Sheue et al. [23] proposed a speech functions of FWHT can be realized more efficiently in digital cir-
encryption algorithm based on fractional order chaotic systems. cuits rather than the trigonometric basis functions of the Fast Four-
It is based on two-channel transmission method where the original ier Transform. The second problem can be eliminated by modified
speech signal is encoded using a nonlinear function of the Lorenz Henon map, which weakens the strong correlation between adja-
chaotic system. Moreover, they analyzed the conditions for syn- cent coefficients. The samples are shuffled through the strong per-
chronization between fractional chaotic systems theoretically by mutation matrix generated with the modified Henon map. The
using the Laplace transform. Mosa et al. [24] introduced an algo- output data is then diffused by XOR-ring the permuted coefficients
rithm based on permutation and substitution of speech segments with keystream generated by the modified Lorenz-hyperchaotic
using chaotic Baker map. They used Discrete Cosine Transform system. The encryption process in higher dimensional space elim-
(DCT) to remove the residual intelligibility in order to compress inates periodic window problems such as limited chaotic range
the signal. Maysaa abd ulkareem et al. [25] proposed a method and non-uniform distribution. It also extends the keyspace and
based on logistic map and blowfish encryption algorithm. They consequently enhances the security. To eliminate the fourth draw-
employed partial encryption method by wavelet packet transform back, a dynamic keystream selection mechanism is introduced. If
for splitting the raw signal to improve the speed of encryption pro- the initial conditions remain the same, the introducer can easily
cess. Halto et al. [26] presented a hybrid chaotic speech encryption acquire the keystream by known chosen plain text and chosen
algorithm in which Arnold cat map is utilized for permutation and ciphertext attacks. This possibility can be prevented by dynamic
logistic map for substitution operation. They used Discrete Cosine keystream generation, in which the keystream generated will be
Transform (DCT) for compressing the audio signal to minimize relevant to audio segments. Therefore different audio segment
residual intelligibility. In [27] Halto et al. presented a method generates completely different keystream which eliminates chosen
where the Lorenz system generates the keystream for substitution plain text and chosen ciphertext attacks.
operation and Rossler chaotic system for permutation process. Ela- The rest of this paper is organized as follows: The preliminary
shy at al. [28] proposed a two level audio encryption algorithm studies of the proposed audio encryption algorithm are presented
based on chaotic Baker maps and double phase random coding. in Section 2. Theoretical framework of the proposed approach is
In the first level it utilizes Baker map and in the second level it uti- given in Section 3. Numerical simulations and performance evalu-
lizes optical encryption using double random phase encoding ations are discussed in Section 4. Comparison of proposed work
(DRPE) for providing physical security which is hard to break. with other state-of-art is discussed in Section 5, followed by con-
Sathya murthi et al. [29] introduced an algorithm based on chaotic clusion in Section 6.
shift keying. In [29], audio signals are sampled and its values are
segmented into four levels, and then the samples are permuted 2. Preliminary studies
using chaotic systems such as Logistic map, Tent map, Quadratic
map, and Bernoulli’s map. Sheela et al. [30] proposed an algorithm In this section, we describe mathematical models of chaotic
based on two-dimensional modified Henon map (2D-MHM) and maps used for encryption, i.e., parametric perturbated Lorenz-
standard map. They introduced Hybrid Chaotic Shift Transform hyperchaotic system and modified Henon map. Periodic, quasi-
(HCST), and deoxyribonucleic acid (DNA) encoding rules to periodic, chaotic and hyper-chaotic behavior of parametric pertur-
enhance the security level. Aissa Belmeguenai et al. proposed a bated Lorenz map is discussed by means of bifurcation diagram.
method, where only relevant part of the speech segment under- One dimensional signal compression by FWHT is also discussed.
goes encryption process [31]. Animesh Roy et al. [32] presented
an algorithm based on audio signal encryption using chaotic Henon 2.1. Hyperchaotic system
map and lifting wavelet transform. Z. Habib et al. presented a paper
based on amplitude scrambling and Discrete Cosine Transform Higher dimensional chaotic systems show distinct advantages
coefficient scrambling. In [33], they designed a permutation net- over lower dimensional chaotic system due to its complex dynamic
work using TD-ERCS chaotic map. Some of the constraints random behavior. In this work, we adopt a parametric perturbated

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx 3

Lorenz- hyperchaotic system [34]. Lorenz system shows chaotic used to generate permutation matrix for encryption process. This
behavior for the control parameters, a = 10, b = 8/3 and c = 28 dynamical system (2) takes a point ðxn ; yn Þ in the two dimensional
[35]. Parametric perturbation in Lorenz system may be given to plane and map this point to a new point given by ðxnþ1 ; ynþ1 Þ. The
all of the parameters a, b & c or on selected parameters. In the pro- map is dependent on two bifurcation parameters að> 0Þ and
posed system (1) the control parameter ‘a’ is selected for paramet- bð> 0Þ: Moreover, the parameter b measures contraction rate of
ric perturbations by adding a PI controller in the feedback path of the 2D quadratic Henon map which is independent of xn and yn .
the Lorenz system. Fig. 2 depicts the bifurcation diagram of both the systems. Bifurca-
tion diagram of modified Henon map shows wider range of output
x
¼ aðy  xÞ þ w
distribution for the control parametera compared to its seed map.
Therefore, modified henon map increases the range of permutation
y
¼ cx  y  xz
operation compared to its seed map.
z
¼ xy  bz
2.3. Fast Walsh Hadamard Transform (FWHT)
w
¼ ki aðy  xÞ þ kp x ð1Þ
Walsh-Hadamard transform is used in different applications,
where x; y; z; w are the state variables and a; b; c; kp ; ki are the system such as data compression, processing of speech and image signals,
parameters. kp and ki are control parameters of the PI controller. coding and communications. It is an orthogonal transformation
Parametric perturbation changes the three dimensional autono- that decomposes input signals into rectangular waveforms called
mous system to non-autonomous system (1), which is equivalent Walsh functions. The Walsh functions forms a system of orthogo-
to a four dimensional hyperchaotic system. When the control nal functions and have only two values +1 and 1. This transfor-
parameters, a ¼ 10; b ¼ 83 ; c ¼ 10; kp ¼ 3:6; ki ¼ 5:2;and initial con- mation is computationally simple since it has no multiplication
ditions (1,1,1,1), then the Lyapunov exponent obtained are and division operations. To implement FWHT of order n ¼ 2m
L1 ¼ 0:01000, L2 ¼ 0:421905, L3 ¼ 0:326781, L4 ¼ 13:385272. requires only nlogn addition and subtraction. Generally, Fast Walsh
Since more than two Lyapunov exponents are positive, the system Hadamard transformation follows the recursive definition of sym-
is hyperchaotic. The evolution of periodic, chaotic, and hyperchaotic metric Hadamard matrix. Let H be the Hadamard of order n ¼ 2k
attractors in this system can be generated by varying ki [20, 15] by described as follows:
fixing all other parameters constant. Fig. 1 illustrates the bifurcation

H2k1 H2k1
diagram and Lyapunov exponents of the modified system. H 2k ¼ ¼ H2  H2k1
H2k1 H2k1
2.2. Henon map dynamical system
k ¼ 1; 2; 3;

















;
Henon map was proposed by Michel Henon in 1976 as a com-
prehensible approach of the Poincare map that results from the H1 ¼ 1 ð3Þ
solution of complex Lorenz equation [36]. Modified Henon map

was developed to improve the complex dynamic behavior and 1 1
when k ¼ 1; H2 ¼ H2  H1 ¼
bifurcation range [30]. In this map, x2n term of the seed map is 1 1
replaced with nonlinear termcosðxn ). Modified Henon map can be
2 3
mathematically modeled as follows: 1 1 1 1
6 1 1 1 1 7
xnþ1 ¼ 1  acosðxn Þ  byn 6 7
when k ¼ H4 ¼ H2  H1 ¼ 6 7 ð4Þ
4 1 1 1 1 5
ynþ1 ¼ xn ð2Þ 1 1 1 1
In the proposed method, the system (2) is taken as a computa- where  denotes Kronecker product.
tional tool to generate permutation matrix. The data set resulted The Walsh transform is modeled as in Eq. (5) for one dimen-
from dynamic system (2) for some specific values of a and b are sional signal.

15

10
L4
L1
Lyapunov exponent

-5 L3 L2

-10

-15

-20

-25
0 5 10 15 20 25
b

(a) (b)
Fig. 1. (a) Lyapunov exponents b) Bifurcation diagram for ki [20, 15].

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
4 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx

1 10

0.5 5

max (x)
max (x)

0 0

-0.5 -5

-1 -10
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 0 1 2 3 4 5 6 7 8
System parameter(a) System parameter(a)

Fig. 2. (a) Bifurcation of diagram Henon map (b) Bifurcation diagram of modified Henon map.

" #
1 X
N1 Y
n1 with bi ðxÞ ¼ 0 or 1 for i ¼ 0;


:n  1 ð6Þ
W ðuÞ ¼ f ðxÞ ð1Þbi ðxÞbn1i ðuÞ ð5Þ
N x¼0 i¼0 In this transformation, Walsh kernel forms an array of matrix
having orthogonal rows and columns. Therefore, both the forward
where x and u are independent variable represented in n bits. The
and inverse transformations are identical operations except for the
binary representation of x and u can be written as:
constant multiplicative factor of 1/N for 1D signal.
ðxÞ10 ¼ ðbn1 ðxÞbn2 ðxÞ . . . b0 ðxÞÞ2 " #
X
N1 Y
N 1
bi ðxÞbn1i ðuÞ
f ðxÞ ¼ WðuÞ ð1Þ ð7Þ
ðuÞ10 ¼ ðbn1 ðuÞbn2 ðuÞ . . . bo ðuÞÞ 2 u¼0 i¼0

m (N,1) C2 (P, 1)

Input –speech FWHT Cipher

m (R, Q)
Section Keys

Permutation Substitution

C (R, Q)
1

Fig. 3. Block diagram of the proposed algorithm.

1 0.01
0.000,0.001,0.014,0.0.011,.................0.025,0.032,0.325,0.526,0.456,0.258,..............0.001 0.012,0.091,0.014,0.0.009,.................0.001,0.0052,0.003,0.0026,0.004..............0.001

0.008
0.5

0.006
N
0
0.004

-0.5
N 0.002

-1 0
0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 500 1000 1500 2000 2500 3000 3500 4000 4500
(a) (b)
0.01
0.012,0.091,0.014,0.0.009,.................0.001,0.0052,0.003,0.0026,0.004, 0.001

0.008

0.006
N-L
0.004

0.002 L

0
0 500 1000 1500 2000 2500 3000 3500 4000 4500
(c)

Fig. 4. (a) Original speech (b) Compressed speech signal (c) Compressed signal with zeroing out the coefficients.

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx 5

reshaped plaintext permuted signal

200 200

100
100

0
0

-100
-100

-200
-200 0 500 1000 1500 2000 2500 3000 3500 4000 4500
0 10 20 30 40 50 60 70
(a) (b)

-0.948 7.4009 67.077 ….. 20.27 -20.27 -26.60 -15.92 ….. 40.646
-2.993 9.0728 -13.18 ….. -2.693 -2.693 0.0396 -1.148 ….. 67.077
1.307 -17.25 -56.17 ….. 7.4009 7.4009 0.950 -4.465 ….. -13.18

…. ….. ….. ….. …. …. ….. ….. ….. 40.646

-2.706 40.646 -15.92 …. -20.27 0.948 -9.238 -1.109 ….

7.4009 67.077 7.4009 67.077 -26.60 40.646 34.646 54.646 9.646 100.646

9.0728 -13.18 9.0728 -13.18 0.0396 67.077 -2.706 67.077 -2.706 67.077

(c) (d)

Fig. 5. (a) Original speech signal in 2D vector space (b) Permuted speech signal (c) Original data samples in 2D vector space (d) Permutated sample values.

3. Proposed method regime by slowly varying the control parameters. Since there is a
slow transition from periodic to chaotic, there may be a chance
The overall idea of the encryption process is depicted in Fig. 3. to produce periodic or redundant samples for the first few itera-
The input speech signal is initially compressed by FWHT followed tions. In this scenario, the first few iterations in the permutation
by permutation operation. Then, generate keystream for substitu- process seems fairly close together, hence it should be discarded.
tion operation relevant to characteristics of plain speech signal. Therefore, the total number of iterations is p2 + 1000. Algorithm 1
The various steps in encryption process are systematically demon- describes the entire permutation process in detail. Iterate the data
strated as follows: samples as follows:
 
xi þ 1 ¼ mod ð1  acos xi Þ þ byj ; Q Þ i 2 ð1; Q Þ
3.1. Encryption process
yjþ1 ¼ modððxi Þ; RÞ j 2 ð1; RÞ ð9Þ
3.1.1. Audio signal compression by FWHT


Initially, we divide the input audio file into frames, each with N where ðxi ; yj Þ is the initial position of the data sample and xiþ1 ; yjþ1
samples. Assume the message signal is m ¼ fm1 ; m2 ; is the first iterated position. Fig. 5 displays the values of original and
::mi ;


mN gi ¼ 1; 2;





N. We compress the audio signals based permuted data samples.
on Eq. (5) and reduce the sample space by discarding the higher
order coefficients (8). Original signal, compressed signal with Algorithm 1. Permutation of speech samples
numerical values are displayed in Fig. 4.
Input: Audio files, system parameters of modified Henon map
mw ¼ FWHTðmÞ Output: permuted speech samples
1: Set audio file with frequency 8 KHz for a duration of 0.5 sec
mw ðN  L : lengthðmw ÞÞ ¼ 0; lengthðmw Þ ¼ N ð8Þ and system parameters a = 3.58, b = 0.56
2: Read and segment the audio file
where mw ðP; 1Þ is the compressed data signal and P is the sample
3: %%Compression, Compress the signal and Zeroing out the
size of the compressed signal.
coefficient.
4: mw ¼ whtðaudio:wav Þ
3.1.2. Permutation process 5: mw ðN  l : lengthðmw ÞÞ ¼ 0;
Prior to permutation process, the audio samples are reshaped 6: %%permutation
mw ðP; 1Þ into two dimensional vector space mw ðQ ; RÞ. In this step 7: Reshape(mw ; Q ; R)
permute the audio samples depending on the random matrix gen- 8: for i = 1:Q do
erated from the modified Henon map by performing some trans- 9: for j = 1:R do
formation on the original data samples that produce ciphered 10: xi þ 1 ¼ 1  a  cosðxi Þ þ byj
audio samples C 1 ðQ ; RÞ, with uniform histogram. The minimum
11: yj þ 1 ¼ bxi
number of iterations for Henon map should be greater than P 2 to
12: end for
completely permute the data samples. A chaotic system shows
13: end for
gradual evolution from Periodic to quasi-periodic and to chaotic

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
6 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx

3.1.3. Dynamic keystream selection mechanism Algorithm 2: Dynamic keystream Generation

In order to make initial conditions dependent on each other and Input: compressed signal coefficient, set the system parameter as
be sensitive on plain audio data, the initial conditions are derived a = 10; b = 10/3; c = 28; kp = 3.5; ki = 5.2. Time step = 0.005
from Eq. (10) and by performing basic arithmetic operations as in Output: Pseudo random number sequences
Eq. (11). The sub keys so obtained are the initial conditions 1. %% initial conditions for hyperchaotic system
xð0Þ; yð0Þ; zð0Þ; wð0Þ for the hyperchaotic system (1). In each round 2. for i = 1:N
of operation the keys are updated, and it will avoid the possibility 3. keyð1Þ ¼ N2 sumð2i þ 1Þ
of various differential attacks. 4. keyð2Þ ¼ N2 sumð2iÞ
5. end for
1 X N
keyð1Þ ¼ mw ð2i þ 1Þi ¼ 1; 2;


:N 6. %%%sub key generation
N=2 i¼0
7: subkeyð1Þ ¼ keyð1Þ þ keyð2Þ; mod1
8: subkeyð2Þ ¼ keyð1Þ  keyð2Þ
1 X N 9: subkeyð3Þ ¼ keyð1Þ  keyð2Þ
keyð2Þ ¼ mw ð2iÞ ð10Þ ~
keyð2Þ
N=2 i¼0 10: subkeyð4Þ ¼ keyð1ÞA
11: %%% keystream generation
12: %%set the initial conditions for hyperchaotic system as x(0)
subkeyð1Þ ¼ keyð1Þ þ keyð2Þ; mod1
= subkey(1); y(0) = subkey(2) z(0) = subkey(3); w(0) = subkey
(4)
subkeyð2Þ ¼ keyð1Þ  keyð2Þ 13: for k = 1:N
14: fx ¼ a  yðkÞ  a  xðkÞ þ wðkÞ
subkeyð3Þ ¼ keyð1Þ  keyð2Þ 15: fy ¼ xðkÞ  zðkÞ þ c  xðkÞ  yðkÞ
16: fz ¼ xðkÞ  yðkÞ  b  zðkÞ
~
keyð2Þ
subkeyð4Þ ¼ keyð1ÞA ð11Þ 17: fw ¼ kp  a  ðyðkÞ  xðkÞÞ  kixðkÞ
18: %%update the values of x, y, z, w
After computing the initial conditions [xð0Þ; yð0Þ; zð0Þ; wð0Þ] for 18: xðk þ 1Þ ¼ xðkÞ þ dt  fx
the modified Lorenz hyperchaotic system, generate the keystream 19: yðk þ 1Þ ¼ yðkÞ þ dt  fy
by iterating the system (1) by Runge-Kutta method. Convert the
20: zðk þ 1Þ ¼ zðkÞ þ dt  fz
four hyper-chaotic sequences into integer sequences
      21: wðk þ 1Þ ¼ wðkÞ þ dt  fz
xi ; yi ; zi ; and fwi g as follows: 22: tðk þ 1Þ ¼ tðkÞ þ dt
23. end for
xi ¼ modððabsðxi Þ  floorðxi ÞÞ1014 ; 1Þ
24: %% Convert the sequences into integer sequences
25: for i = 1:N
yi ¼ modððabsðyi Þ  floorðyi ÞÞ1014 ; 1Þ 26: xi ¼ modððabsðxi Þ  floorðxi ÞÞ104 ; 1Þ
27: yi ¼ modððabsðyi Þ  floorðyi ÞÞ104 ; 1Þ
zi ¼ modððabsðzi Þ  floorðzi ÞÞ10 ; 1Þ 14
28: zi ¼ modððabsðzi Þ  floorðzi ÞÞ104 ; 1Þ
29: wi ¼ modððabsðwi Þ  floorðwi ÞÞ104 ; 1Þ
wi ¼ modððabsðwi Þ  floorðwi ÞÞ10 ; 1Þ 14
ð12Þ 29. end
30: for k = i:N
th
xi ; yi ; zi ; wi indicate the i element of keystreams. The absðx) returns 31: X = XOR(x(i),y(i),z(i),w(i))
the absolute value of x and floorðxÞ returns the largest integer less 32: end for
than or equal to x. Generate the normalized keystream for substitu- 33: %%Normalized key stream generation
tion operation as follows: 34: key_norm=X  X min =X max  X min

X ¼ XORðxi ; yi ; zi ; wi Þ

X  X min 3.1.4. Substitution operation

key norm ¼ ð13Þ After first level encryption process (permutation operation), the
X max  X min
two dimensional data samples (C 1 ðQ ; RÞ) are reshaped to one
where key norm is the normalized keystream generated. X max; X min dimensional data samples (ðC 1 ðP; 1Þ). Then, generate keystream
are the maximum and minimum values present in the array X. for substitution operation based on Algorithm 2 and eliminate first
The sequence of operation is elaborated in Algorithm 2. few samples of the keystream since the samples are redundant.

400 2000

200 1000

0 0

-200 -1000

-400 -2000
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
(a) (b)

Fig. 6. Encrypted speech signal after second level of encryption (a) Permuted speech signal after p2 þ 1000 rounds of iterations (b) Speech signal after substitution operation.

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx 7

Substitution operation is then carried out by XOR-ing the per- Table 1

muted samples with the keystream generated using Eq. (12). Correlation coefficient (r xy ).

Fig. 6(a) and (b) show the speech samples after p2 þ 1000 permu- Audio files Henon map Modified Henon map
tation operation and speech signal after substitution operation Audio.wav1 0.00163 0.000991
respectively. Pseudo code for substitution operation is given as in Audio.wav2 0.00451 0.001359
Algorithm 3. Audio.wav3 0.00856 0.000651
Audio.wav4 0.00421 0.001569
X
P Audio.wav5 0.00131 0.001865
C2 ¼ C 1 ðiÞ
key norm ð14Þ Audio.wav6 0.00969 0.002634
i¼0 Audio.wav7 0.005624 0.001695

PNs
cov ðx; yÞ i¼1 ðxi  EðxÞÞðyi  EðyÞÞ
1
Algorithm 3. Substitution operation NS
rxy ¼ ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffisffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð15Þ
Input: permuted audio file, dynamic keystream rx ry PNs
2 PNs

Output: encrypted speech file

1
NS
ðxi  EðxÞÞ NiS ðyi  EðyÞÞ2
i¼1 i¼1
1: for i = 1:p do
2: C 2 ¼ xorðC 1 ðiÞ:key subÞ rx ; ry –0
3: end for
where EðxÞ and EðyÞ are mean and rx ; ry are the standard deviation
of the encrypted and decrypted speech signal.

3.2. Decryption process 4.2. Signal to noise ratio (SNR)

The procedure of decryption process is just reverse of the The signal to noise ratio is one of the straight forward methods
encryption process. The decryption process can be performed to validate the performance of data encryption algorithm. SNR
easily by means of the pre-shared keys. The decryption process measures the noise content in the encrypted data signal. Cryptan-
can be briefly described as follows: alyst always try to increase the noise content in the encrypted sig-
Step 1: Generate the same keystream bits according to the steps nal so as to minimize the information content in the encrypted
3.1.3 in the encryption process. data [38]. Fig. 8 shows the original and encrypted speech signal.
Step2: Xoring the samples C2 ðP; 1Þ with the keystreams. Fig. 8(c, d) illustrates the randomized nature of encrypted signal
Step3: Do the inverse permutation process according to the step after permutation and substitution operation. The SNR values of
3.1.2. encrypted audio files are calculated based on the Eq. (16) and given
Step 4: Take Inverse Fast Walsh Hadamard Transform (IFWHT). in Table 3.
PN S 2
i¼1 xi
4. Simulation result and analysis SNR ¼ 10  log10 PN ð16Þ
i¼1 ðxi
S
 yi Þ2
The proposed algorithm is simulated on a classical computer
with MATLABR 2013a (version) software. Different voice samples 4.3. UACI and NSCR analysis
of male and female audio files with sampling rate of 8000 sam-
ples/sec are selected for the test. We evaluated the performance In data encryption, resistace to differential attacks is generally
of this algorithm through various statistical and differential analyzed through the NSCR (number of samples change rate) and
analyses. UACI (unified average changing intensity) tests [39]. In this analy-
sis two different speech segments are encrypted with same key-
4.1. Correlation analysis streams, where the original speech segments are differed by one
sample space. Then the encrypted speech segments are compared
Correlation analysis is a statistical method to evaluate the per- by the number of sample change rate (NSCR) and the unified aver-
formance of cryptographic algorithm over various statistical age changing intensity (UACI). Both these parameters can be
attacks [37]. Correlation coefficient analysis measures the mutual expressed as follows:
relationship between similar segments in the plain and encrypted
X Di
audio file. A secure data encryption algorithm converts original NSCR ¼  100%i ¼ 1; 2


:N
data into random-like noisy signal with low correlation coefficient. i
N
Low correlation coefficient indicates the narrow correlation

P 0 
between original and encrypted speech files. In this work Correla- 1  xi
i xi
UACI ¼
tion analysis is carried out for both Henon map and modified N 65535
Henon map. Correlation coefficient (rxy ) between original and
encrypted audio samples are calculated and listed in Table 1. Cor- Table 2
relation coefficient between original data samples in the same Correlation coefficient (r xy ).
audio file, and original and encrypted data samples are also calcu-
Audio files Original Encrypted
lated and given in Table 2. Analysis shows that there is an improve-
ment of permutation operation since the correlation coefficients of Audio.wav1 0.98153 0.000991
Audio.wav2 0.96421 0.001359
modified Henon map is smaller than its seed map. Fig. 7(a) shows
Audio.wav3 0.98816 0.000651
the scatter plot diagram of original speech signal. Randomized nat- Audio.wav4 0.97461 0.001569
ure of speech signal after permutation and substitution operation Audio.wav5 0.99141 0.001865
is illustrated in Fig. 7(b) and (c). Correlation coefficient can be cal- Audio.wav6 0.98979 0.002634
Audio.wav7 0.97562 0.001695
culated as follows:

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
8 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx

-3 5
x 10 x 10
3 1.5 600

400
2 1

200
1 0.5

0 0

-200

-1 -0.5
-400

-2 -1
-600

-3 -1.5
-250 -200 -150 -100 -50 0 50 100 150 200 -800
-3 -2 -1 0 1 2 3 -800 -600 -400 -200 0 200 400 600
(a) x 10
-3
(b) (c)

Fig. 7. Scatter plot diagram of (a) Original signal (b) Encrypted signal after permutation (c) Encrypted signal after substitution.

4 4
x 10 x 10
4 1.5 100

1 80
2
0.5
60
0 0
40
-0.5
-2 20
-1

-4 -1.5 0
0 1 2 3 4 5 0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
(a) 5
x 10
(b) (c)
4
x 10
100 1.5
1500
1
1000
50
0.5
500
0 0
0
-0.5
-500
-50
-1
-1000
-100 -1.5
0 1000 2000 3000 4000 5000 -1500 0 1000 2000 3000 4000 5000
0 1000 2000 3000 4000 5000
(d) (e) (f)

Fig. 8. (a) Original signal (b) Compressed signal (c) Data samples after permutations (d) Data samples after substitution (e) Decrypted signal.

Table 3 4.4. Spectral entropy

Signal to noise ratio.

Audio files SNR (dB) Spectral entropy measures the randomness in both encrypted
Audio.wav1 133
and original speech signal. Its measurement is based on the
Audio.wav2 154 assumption that the spectrum of meaningful speech segment is
Audio.wav3 124 correlated than the noisy signal [1]. The spectral measurement
Audio.wav4 110 compares the entropy where the amplitude component of the
Audio.wav5 135
power spectrum is taken as a probability parameter in entropy cal-
Audio.wav6 137
Audio.wav7 144 culation. The amount of information can be calculated as the neg-
ative of entropy or the negative logarithm of probability. Thus,
meaningful speech segments show low entropy since it contains
where organized data samples. However the encrypted speech signals

Z
1; xi –x;i
di ¼ ð17Þ
0; otherwise Table 4
Differential analysis.
ci and c;i denotes the the audio samples at ith position of the
Audio files NSCR UACI
encrypted speech samples and N corresponds to the length of the
speech segments. The upper-bound for NSCR and UACI are 100% Audio.wav1 99.9997 33.33
Audio.wav2 99.9999 33.28
and 33.3% respectively. For a secure encryption scheme these Audio.wav3 99.9998 33.34
parameters should be close to the upper bound ideal values. NSCR Audio.wav4 99.9996 33.25
and UACI values of the proposed algorithm is calculated and listed Audio.wav5 99.9992 33.31
in Table 4. The results show that the values obtained through pro- Audio.wav6 99.9993 33.29
Audio.wav7 99.9989 33.30
posed algorithm is considerably closer to ideal values.

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx 9

have high entropy and large spectral peaks similar to noisy signal. correct key. Fig. 11(b, c, d, e) shows the decrypted signal with slight
The entropyEi can be measured as follows: variations in the initial conditions in the range of 1015 :
X
Ei ¼ PSDn ðf i Þ log ðPSDn ðf i ÞÞ; i ¼ 1; 2; 3





; n ð18Þ
n 4.6. Histogram analysis
where PSDn is the normalized power spectrum and f i is the fre-
Histogram analysis is one of the accurate methods to evaluate
quency of the signal. Irregularities of amplitude in original and
the quality of encrypted speech signal. Since a practical encryption
encrypted signals are shown in Fig. 9.
algorithm is likely to encrypt original speech file into random like
noise, it is desirable to obtain an encrypted speech file with equally
4.5. Keyspace and key sensitivity analysis
probable sample values. Therefore, the encrypted speech furnishes
no information that would facilitate the possibility of any statisti-
In the proposed algorithm, system parameters of modified
cal attacks on the encrypted domain. Histogram of both encrypted
Henon map (a ¼ 3:58; b ¼ 0:56), system parameters of modified
and original speech signal is illustrated in Fig. 12. Fig. 12(b) dis-
Lorenz-hyperchaotic system
plays the uniformly distributed histogram of encrypted speech sig-
(a ¼ 10; b ¼ 3:33; c ¼ 28; kp ¼ 3:5; ki = 5.2) and initial conditions
nal, which indicates the randomness of encrypted speech signal.
of the hyperchaotic system (xð0Þ; yð0Þ; zð0Þ; wð0Þ) constitutes the
From the histogram, it is clear that the proposed algorithm is
keyspace. If the precision of each system parameter and initial con-
highly secure against various statistical attacks.
dition is set to 15 decimal points, the key space of the proposed

6
algorithm is 1015 ¼ 2548:11 . It is sufficiently large enough to 4.7. Computational complexity
resist the brute force attack. Key sensitivity is the essential quality
for any good data encryption algorithm, which make sure that the Big-oh Notation is a unified way to express the complexity of an
security level of the algorithm against the brute-force attack [40]. It algorithm. In classical computation, computational complexity is
means that, a small variation for any key parameter bring an evaluated by the elementary operations involved in the encryption
apparent change in both encrypted and decrypted speech signal. process. Since speed of an algorithm depends on the target com-
The effect of variation in keyparameter on encryption process is puter processor, it is difficult to estimate the exact runtime of an
verified by encrypting the signal with slightly different initial con- algorithm. Big-oh notation measures the execution time of an algo-
ditions. The simulation result shows that the slight variations in rithm in terms of the input array size and the nature of arithmetic
keyparameter will result in completely different encrypted signal. operations. In the proposed algorithm, the encryption process con-
Fig. 10 shows the encrypted signals with two different initial con- sists of p2 þ 1000 round of permutation operations and a single
ditions. To evaluate the key sensitivity of decrypted signal, encrypt round of substitution operations. Computational complexity of
the speech file with one fixed secret key then decryption is per- permutation operation is Oðn2 Þ time or quadratic time, since the
formed with slightly different keys. The resulting speech files time complexity of the operation grows quadratically with respect
decrypted with wrong keys apparently looks different and reveals to input array sizen. However, the computational complexity of
no information. Fig. 11(a) shows the decrypted speech signal with substitution operation is independent of input array size, since this

0 94 80
Power Spectrum Magnitude (dB)

Power Spectrum Magnitude (dB)

Power Spectrum Magnitude (dB)

93
-20 75
92
-40 70
91
-60 65
90
-80 89 60

-100 88 55
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(a) (b) (c)

Fig. 9. Power Spectral Density (a) Original speech signal (b) Permuted speech signal (c) Substituted speech signal.

3
x 10
1 2000 1500

0.5 1000
1000
500
0

0 0
-0.5
-500
-1000
-1
-1000

-1.5 -2000 -1500

0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600
(a) (b) (c)

Fig. 10. Key sensitivity on encryption process (a) Original speech signal (b) Encrypted speech signal for key x0 = 0.413, y0 = 0.931, z0 = 0, w0 = 0.825 (c) Encrypted speech
signal for x0 = 0.913, y0 = 0.131, z0 = 0.123, w0 = 0.925.

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
10 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx

4
x 10
2 2000 1500

1000
1 1000
500

0 0 0

-500
-1 -1000
-1000
-2 -2000 -1500
0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 500 1000 1500 2000 2500 3000 3500 4000 4500
(a) (b) (c)
2000 1500

1000
1000
500

0 0

-500
-1000
-1000

-2000 -1500
0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 500 1000 1500 2000 2500 3000 3500 4000 4500
(d) (e)

Fig. 11. Key sensitivity on decryption process (a) decrypted signal with correct key, decrypted signal with incorrect key (b) x0 þ 1015 (c) y0 þ 1015 (d) z0 þ 1015 (e)
w0 þ 1015 .

800 140

700 120

600
100
500
80
400
60
300

200 40

100 20

0 0
-1.5 -1 -0.5 0 0.5 1 1.5 2 0 10 20 30 40 50 60 70 80 90 100
(a) 4
x 10 (b)

Fig. 12. Histogram of (a) Original speech signal (b) Encrypted speech signal.

Table 5
Quality metrics comparison of various encryption methods.

Method Key length Key space CC SNR NPCR UACI

AES 128, 192, 256 2128 ; 2192 ; 2256 0.00971 1.4461 99.6032 33.3218
Triple DES 168 2168 0.1704 0.250 99.5985 33.3419
Ref. [31] 521 >2512 0.0011 10.6357 99.6399 33.8085
Ref. [29] 533 2744 0.0233 33.7464 99.9982 33.1197
Ref. [8] 477 2477 0.0029 23.89 99.8725 33.1160
Ref. [16] >264 >2624 0.0491 44.8 99.6399 33.1085
Ref. [20] 2107 2 2107 0.0321 54.89 99.6317 33.2782
Ref. [33] 398.66 2398:66 8 0.016 12,830 130 99.6521 33.2122
[Prop: Meth] 548.11 2548:11 0.0.0009 133 99.9989 33.3421

process takes single step to complete the operation irrespective of between original and encrypted signals and tabulated in Table 5.
array lengthn. Thus the computation complexity of substitution We have compared our proposed algorithm with advanced encryp-
round is Oð1Þ. Computational complexity of entire process can be tion standard (AES), Data Encryption Standard (triple DES), algo-
 
expressed as O n2 þ Oð1Þ ¼ Oðn2 Þ. rithm based on quantum chaotic system [16] and an algorithm
based on substitution-permutation chaotic network [20]. Speech
encryption algorithm based on Zaslavsky map [8], TD-ERCS chaotic
5. Comparison with existing works map [33], multiple chaotic shift keying [29] and a non-chaotic [31],
method are also considered for comparative analysis. The size of
The proposed speech encryption algorithm differs from other the proposed method’s key space is greater than 2540 (Section 4.5).
methods, in terms of data compression, permutation and substitu- It is clear from the simulation results (Table 3) that the encrypted
tion operations. Therefore comparison of proposed method with speech signal contains more noise content than in original speech
other state-of-the-art approaches is difficult. However, we have signal. Correlation coefficient (rxy ) is evaluated to be almost zero
analyzed various quality metrics such as key length, keyspace, sig- (Table 2). Also NSCR and UACI analysis validates the capacity of
nal to noise ratio (SNR), NPCR, UACI and correlation coefficient the proposed algorithm against various differential attacks. From

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001
 F.J. Farsana et al. / Applied Computing and Informatics xxx (xxxx) xxx 11

this analysis, we can found that proposed algorithm shows consid- [10] X. Zhang, Z. Zhao, Chaos-based image encryption with total shuffling and
bidirectional diffusion, Nonlinear Dyn. 75 (2014) 319–330.
erable improvement in almost all the encryption quality metrics.
[11] B. Norouzi, S. Mirzakuchaki, A fast color image encryption algorithm based on
hyper-chaotic systems, Nonlinear Dyn. 78 (2014) 995–1015.
6. Conclusion [12] S. Lakshmanan, M. Prakash, C.P. Lim, R. Rakkiyappan, P. Balasubramaniam, S.
Nahavandi, Synchronization of an inertial neural network with time-varying
delays and its application to secure communication, IEEE Trans. Neural
In this paper, a novel approach for speech encryption algorithm Networks Learn. Syst. 29 (2018) 195–207.
based on parametric perturbated Lorenz-hyperchaotic system and [13] L. Teng, X. Wang, J. Meng, A chaotic color image encryption using integrated
bit-level permutation, Multimedia Tools Appl. 77 (2018) 6883–6896.
modified Henon map is introduced. Modified Henon map maxi- [14] Y. Luo, R. Zhou, J. Liu, S. Qiu, Y. Cao, An efficient and self-adapting colour-image
mized the permutation operation compared to its seed map, which encryption algorithm based on chaos and interactions among multiple layers,
eventually decreased the correlation between original and Multimedia Tools Appl. 77 (20) (2018) 26191–26217.
[15] K. Muhammad, R. Hamza, J. Ahmad, J. Lloret, H. Wang, S.W. Baik, Secure
encrypted data samples. Selection of hyperchaotic system elimi-
surveillance framework for IoT systems using probabilistic image encryption,
nated weak chaotic trajectories and smaller chaotic ranges, which IEEE Trans. Ind. Inf. 14 (2018) 3679–3689.
is commonly observed in lower dimensional chaotic system. Due [16] Akram Belazi, Majid khan, Ahmed A. Abd El-Latif, Safya Belghith, Efficient
cryptosystem approaches: S-boxes and permutation-substitution based
to the hyperchaotic nature, the proposed system has more key-
encryption, Nonlinear Dyn. 87 (1) (2017) 337–361.
space, which protects the proposed algorithm against various sta- [17] Mohamed Amin, Ahmed A. Abd, El-Latif, Efficient modified RC5 based on
tistical attacks like brute force attack. Moreover, dynamic chaos adapted to image encryption, J. Electron. Imaging 19 (1) (2010) 013012.
keystream generated with hyperchaotic system eliminated the [18] Ahmed A. Abd El-Latif, Li Li, Xiamu Niu, A new image encryption scheme based
on cyclic elliptic curve and chaotic system, Multimedia Tools Appl. 70 (3) (2014)
possibility of differential attacks. Furthermore, Fast Walsh Hada- 1559–1584.
mard Transform (FWHT) improved the efficiency of algorithm by [19] Ahmed A. Abd El-Latif, Bassem Abd-El-Atty, Muhammed Taha, Robust
reducing the computational complexity while doing the compres- encryption of quantum medical image, IEEE Access 6 (2018) 1073–1081.
[20] Ahmed A. Abd El-Latif, Li Li, Ning Wang, Qi Han, Xiamu Niu, A new approach to
sion process. Various simulations and numerical analysis have chaotic image encryption based quantum chaotic system, exploiting color
been carried out on classical computer to evaluate the performance spaces, Signal Process. 93 (11) (2013) 2986–3000.
of the proposed algorithm. Finally, we have made a comparison of [21] Ahmed A. Abd El-Latif, Bassem Abd-El-Atty, M. Shamim Hossain, Md. Abdur
Rahman, Atif Alamri, B.B. Gupta, Efficient quantum information hiding for
proposed algorithm with chaotic, non-chaotic and standard remote medical image sharing, IEEE Access 6 (2018) 21075–21083.
encryption algorithms. From the comparative study, it can be con- [22] Ahmed A. Abd El-Latif, Bassem Abd-El-Atty, M. Shamim Hossain, Samir
cluded that the proposed algorithm shows improvement over Elmougy, Ahmed Ghoneim, Secure quantum steganography protocol for fog
cloud Internet of Things, IEEE Access 6 (2018) 10332–10340.
some of the existing algorithms and it is an excellent choice for
[23] Long Jye Sheu, A speech encryption using fractional chaotic systems, Nonlinear
voice encryption in practical applications. Dyn. 63 (6/7) (June 2011) 103–108.
[24] E. Mosa, N.W. Messiha, O. Zahran, F.E. Abd El-Samie, Chaotic encryption of
speech signal, Int. J. Speech Technol. 14 (2011) 285–296.
Author contribution [25] Maysaa abd ulkareem, Iman Qays Abdul jaleel, Speech encryption using
chaotic map and blowfish algorithm, J. Basrah Res. 39 (2) (2013).
Farsana F J, Devi V R and Gopakumar K have equally contributed [26] S.N.Al. Saad, E. Halto, A speech encryption based on chaotic map, Int. J.
Comput. Appl. 93 (4) (2014).
in proposing the ideas, discussing the results, and writing and [27] E. Halto, D. Shihab, Lorenz and Rossller chaotic system for speech signal
proofreading the manuscript. The authors have implemented the encryption, Int. J. Comput. Appl. 128 (11) (Oct 2015).
algorithms and carried out the experiments. [28] E.M. Elshamy, El-Sayed M. Rabaie, H.S. El-Sayed, Efficient audio cryptosystem
based on chaotic maps and double phase random coding, Int. J. Speech
Technol. 14 (2015) 285–296.
Declaration of Competing Interest [29] P. Sathiyamurthi, S. Ramakrishnan, Speech encryption using chaotic shift
keying for secured speech communication, EURASIP J. Audio Speech Music
Process., Springer Open Access (2017), https://doi.org/10.1186/s13636-017-
The authors declare that they have no known competing finan-
0118-0, ISSN: 1687-4722, Online First.
cial interests or personal relationships that could have appeared [30] S.J. Sheela, K.V. Suresh, Deepa knath Tandur, A novel audio cryptosystem using
to influence the work reported in this paper. chaotic maps and DNA encoding, J. Comput. Networks Commun. (2017).
[31] Aissa Belmeguenai, Zahir Ahmida, Salim Ouchtati, A novel approach based on
stream cipher for selective speech encryption, Int. J. Speech Technol. 20 (July
References 2017) 685–698.
[32] Animesh Roy, A.P. Misra, Audio signal encryption using chaotic Henon map
[1] C.E. Shannon, Communication theory of secrecy systems, Bell Syst. Tech. J. 28 and lifting wavelet transforms, Eur. Phys. J. Plus (2017).
(1949) 656–715. [33] Z. Habib, J.S. Khan, J. Ahmad, M.A. Khan, F.A. Khan, Secure speech
[2] C. Li, When an attacker meets a cipher-image in 2018: A year in review, in communication algorithm via DCT and TD-ERCS chaotic map, in: Electrical
arXiv:1903.11764, 2019, [online] Available: https://arxiv.org/abs/1903.11764. and Electronic Engineering (ICEEE) 2017 4th International Conference on. IEEE,
[3] J.X. Chen, Z.L. Zhu, C. Fu, H. Yu, L.B. Zhang, A fast chaos-based image encryption 2017, pp. 246–250.
scheme with a dynamic state variables selection mechanism, Commun. [34] V.R. Devi, Farooq Farsana, K. Gopakumar, Hyperchaos generated from 3D
Nonlinear Sci. Numer. Simul. 20 (2015) 846–860. chaotic systems using PI controller, AIP Conf. Proc. 1859 (020030) (2017).
[4] G.D. Ye, J. Zhou, A block chaotic image encryption scheme based on self- [35] Edward Norton Lorenz, Deterministic non-periodic flow, J. Atmos. Sci. 20 (2)
adaptive modeling, Appl. Soft. Comput. 22 (2014) 351–357. (1963) 130–141.
[5] X.L. Chai, An image encryption algorithm based on bit level Brownian motion [36] M. Henon, A two-dimensional mapping with a strange attractor, Commun.
and new chaotic systems, Multimed. Tools Appl. 76 (1) (Jan 2017) 1159–1175. Math. Phys. 50 (1) (1976) 69–77.
[6] C.L. Fu, B.B. Lin, Y.S. Miao, X. Liu, J.J. Chen, A novel chaos-based bit- level [37] J. Ahmed, S.O. Hwang, A secure image encryption scheme based on chaotic and
permutation scheme for digital image encryption, Opt. Commun. 284 (23) affine transform, Multimedia Tools Appl. 75 (2016) 13951–13976.
(2013) 5415–5423. [38] B.T. Bosworth, W.R. Bernecky, J.D. Nickila, B. Adal, G.D. Carter, Estimating
[7] W. Zhang, K.W. Wong, H. Yu, Z.L. Zhu, A symmetric color image encryption signal-to-noise ratio (SNR), IEEE J. Oceanic Eng. 33 (2008) 414–418.
algorithm using the intrinsic features of bit distributions, Commun. Nonlinear [39] F.A. Khan, J. Ahmed, J. Ahmad, A novel image encryption based on Lorenz
Sci. Numer. Simul. 18 (3) (2013) 584–600. equation, Gingerbreadman chaotic map and S8 permutation, J. Intell. Fuzzy
[8] F.J. Farsana, K. Gopakumar, A novel approach for speech encryption: Zaslavsky Syst. 33 (2017) 3753–3765.
map as pseudo random number generator, in: Proceedings of International [40] Ecrypt, II Yearly Report on Algorithms and Keysizes; 2010. http://www.ecrypt.
Conference on Advances in Computing and Communications, ICACC 2016, eu.org/documents/D.SPA.13.pdf.
2016, pp. 816–823.
[9] C. Li, G. Luo, Ke Qin, C. Li, An image encryption scheme based on chaotic tent
map, Nonlinear Dyn. 87 (2017) 127–133.

Please cite this article as: F. J. Farsana, V. R. Devi and K. Gopakumar, , Applied Computing and Informatics, https://doi.org/10.1016/j.aci.2019.10.001