Received 22 April 2024, accepted 9 June 2024, date of publication 11 July 2024, date of current version 19 July 2024.

Digital Object Identifier 10.1109/ACCESS.2024.3414334

A Lightweight Image Encryption Algorithm Based

on Secure Key Generation
A’LAA HUSSEIN ALI 1 , EKHLAS KHALAF GBASHI1 , HAYA ALASKAR2 ,
AND ABIR JAAFAR HUSSAIN 3,4 , (Senior Member, IEEE)
1 Department of Computer Science, University of Technology, Baghdad 10011, Iraq
2 Computer Science Department, Prince Sattam Bin Abdulaziz University, Al-Kharj 16278, Saudi Arabia
3 Department of Electrical Engineering, University of Sharjah, Sharjah, United Arab Emirates
4 School of Computing and Mathematical Sciences, Liverpool John Moores University, L3 3AF Liverpool, U.K.

Corresponding author: Abir Jaafar Hussain ([email protected])

This work was supported in part by the Department of Electrical Engineering, University of Sharjah, under Grant 22020403199; and in part
by the Prince Sattam Bin Abdulaziz University under Project PSAU/2024/1445.

ABSTRACT Data confidentiality and security are important issues due to the sensitivity of the data
and its relationship with users’ privacy. Sensitive data includes images and texts that can be transmitted
over the Internet, Internet of Things devices and edge-fog-cloud system. These devices require speed and
accuracy responses, and they are vulnerable to hacking. To solve these problems, encryption algorithms
provide necessary solution to meet these requirements. Advanced Encryption Standard represents the best
development in data encryption; however, it is computational expensive. In this research, an improved
advanced encryption standard algorithm is proposed with advanced security and lightweight computation
utilized for encrypting of images and texts. The algorithm is improved using various steps including key
generation which is performed in two steps. First, using an innovative, proven chaotic function distinguished
by its sensitivity to any change in its variables. Second, using three-dimensional Lorenzo function. In our
research, a unique key was used for all rounds, and round key then used like advanced encryption standard.
Two new dynamic substitution boxes are used one for odd rounds and the other for even rounds in which the
speed does not exceeding a millisecond. The mix column function was replaced by a circular permutation
function at the bit level, which improved the speed and performance of the algorithm, Our extensive
simulation results indicated enhanced speed, randomness, and high efficiency in encrypting Internet of
Things data. The algorithm was evaluated using the National Institute of Standards and Technology tests.

INDEX TERMS AES, chaotic map, circular permutation, cryptography, dynamic S-box, IoT, lightweight
algorithm.

I. INTRODUCTION bility in various applications [3]. IoT devices struggle to

Internet of Things (IoT) has transformed industries, smart handle AES algorithm due to its complexity [4], [5]. Low-
homes, and healthcare, but it has also brought up new secu- powered IoT devices can benefit from lightweight algorithms
rity challenges, especially for low-resource devices [1]. Fog since they balance security and resource efficiency [6]. The
computing increases data security and privacy; however, its main purpose of data encrypting is to prevent unauthorized
high processing requirements and its drawbacks with tra- access during transmission and storage [7]. For defensive
ditional methods necessitate lightweight cryptography [2]. purposes, encryption technique or system must be nearly
Advanced encryption standard (AES) algorithm is widely unbreakable [8]. Pseudo-random key sequences have been
utilized due to its exceptional dependability and applica- generated for processing data encryption and decryption [9].
The chaotic sequence can theoretically be made unbreakable
The associate editor coordinating the review of this manuscript and by utilizing pseudo-randomness and predictive complexity
approving it for publication was Leandros Maglaras . of chaotic systems and their sensitivity to initial values to
2024 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
VOLUME 12, 2024 For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ 95871
 A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

produce the same encryption effect as a random key in a modified S-box is 51% on average showing an improvement
one-time pad [10], [11]. Because of the differences in the of 3% on average when benchmarked with the original AES
computational costs, encrypting image formats presents secu- method.
rity risks, especially when delivering images via wireless In [15], the authors introduce a novel AES encryption
networks [12]. This research aims to create an effective, S-box with superior distance to Strict Avalanche Criterion
lightweight, fast, and safe encryption framework for images (SAC), Bit Independence Criterion (BIC), and algebraic
(gray and color) and text byg safeguarding important data complexity, enhancing security measures by evaluating key
within these resource-limited devices’ networks. Our goal is cryptographic parameters. The authors claimed that their
to create a more reliable and secure IoT utilizing the com- proposed S-box performs at least as well as the standard
bined capabilities of AES improvement and chaos-based key S-boxes.
generation in the Fog computing architecture. In [16], an enhanced AES algorithm incorporates dynamic
In particular, the novelty of the proposed works includes S-box creation and key generation was proposed. The authors
the following: indicated that their algorithm enhances security against
• Chaos-based key generation: as far as the authors are attacks in online transactions, banking, and e-commerce by
concerned, this is the first-time chaotic maps are uti- adding complexity to the cipher text.
lized to generate pseudo-random key streams for image A keystream generation method utilizing the Chebyshev
and text encryption, taking advantage of their inherent map was proposed in [17], the algorithm offers real-time
resistance to statistical analysis and sensitivity to initial image encryption with strong cryptographic strength,
conditions. meeting NIST requirements. Only a small portion of a
• AES enhancement for IoT: improving AES algorithm’s minute is needed for the application, which is considered
performance for image encryption on IoT devices with a long time.
limited resources while preserving a high level of secu- The enhanced AES technique proposed in [18], uses a
rity. 256-bit random number generator to generate a randomized
S-box, enhances security in military applications and IoT
Simulation results indicated that the proposed approach security with low overheads.
is faster at encrypting data when benchmarked with state- In [19], Small Lightweight Cryptographic Algorithm
of-the-art algorithms. Furthermore, it is demonstrated that (SLA), a lightweight encryption technique based on the sub-
the throughput of the proposed method is 32,803.69 for stitution permutation network (SPN), offers faster security
encryption and 92,619.39 for decryption. While the algorithm than the Feistel-based ciphers, making it suitable for embed-
nonlinear characteristic is maintained, the recommended ded settings like Radio-Frequency Identification (RFID) tags
modifications did not weaken the security of the conventional and wireless sensor nodes.
AES. Cipher attacks require extensive amount of time to Based on chaotic theory, a first phase of permutation has
break the proposed algorithm. been added to the AES algorithm to improve confusion rates
The reminder of this paper is organized as follows. was proposed in [20]. Two DNA sequences are used to create
Section II provided extensive information about the literature the mixing matrix from the logistic map 3D; the first sequence
review. The suggested work and associated approach are cov- serves as a key in the add round key. The decryption output
ered in Section III. The results and analysis of the proposed is a DNA sequence sent to the recipient. Completing the
method are shown in Section IV, along with benchmark with encryption and decryption procedures after five statistical and
state-of-the-art techniques. Sections V and 6 show the results NIST checks took a few seconds.
and the conclusion, respectively. The research proposed in [21] creates S-box using shifting,
a circle map, and a password key, highlighting its dynamic
II. RELATED WORK nature, input-output interaction, and intricate production pro-
This section discusses the development of secure keys and cess.
lightweight image encryption technique. Table 1 shows various techniques for data encryptions
The authors in [13] combined Henon map with AES with their advantages and disadvantages. As it can be
algorithm in which AES algorithm encrypts plain image, noted, majority of these techniques suffer from low speed,
while the Henon map generates a random key for the encryp- complexity, and low security. The aim of this work is to
tion stage, enhancing security and withstanding more attacks. overcome these issues by using lightweight algorithm with
The results showed that their approach performs satisfactorily improved process complexity while preserving the security
when encrypting images [13]. of the data.
The authors in [14] introduced a technique based on AES
to enhance the S-box generation, in which they claim that
their algorithm demonstrates its effectiveness in generating III. BACKGROUND AND METHODOLOGY
encrypted messages with a larger average avalanche impact In this section, basic background about the well-known
when benchmarked with the original AES method. Avalanche Advanced Encryption standard as well as our proposed
effect shows that the outcomes of encrypted texts by the methodology will be provided.

95872 VOLUME 12, 2024

A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

TABLE 1. Benchmarking various techniques for data encryptions.

VOLUME 12, 2024 95873

 A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

TABLE 1. (Continued.) Benchmarking various techniques for data encryptions.

95874 VOLUME 12, 2024

A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

TABLE 1. (Continued.) Benchmarking various techniques for data encryptions.

A. ADVANCED ENCRYPTION STANDARD ALGORITHM of transformations, including Add Round Key, Sub Bytes,
(AES) Shift Rows, and Mix Columns, with round 10 having all
The AES algorithm, a symmetric key algorithm, established transformations except for Mix Column, and the decryption
by the US National Institute for Standard and Technol- process is the exact inverse.
ogy (NIST) as the standard for digital data encryption
method [22]. AES comprises three types, which are: AES- B. PROPOSED METHOD OF LIGHTWEIGHT AES
128, AES-192, and AES-256, each with varying key sizes that This work proposes a lightweight, fast, and secure improved
determine the allocation of rounds [23], [24]. AES’s functions AES encryption method for IoT sensor data security called
are Add-Round-Key, Shift-Rows, Mix-Columns, and Sub- simple swift IoT guard (ss IoT g). It creates secret keys using
Bytes [25], [26]. The 128-bit AES algorithm uses 08 rounds an innovative map and Lorenzo 3Dimension. The algorithm

VOLUME 12, 2024 95875

 A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

Algorithm 1 Key Generation by the Proposed Map Algorithm 2 Generating Key Using the Lorenzo Map
Inputs: p, x, n: The parameter ‘p,’ the most important variable Inputs: p, r, t (parameters for the Lorenzo model), xold , yold ,
in deciding how the logistic map behaves; x: The initial value, zold (initial random values within the range [0, 1].), n: The
n: The number of iterations to generate the key number of iterations to generate the key
Output: key: A byte array containing the generated key Output: key: A byte array containing the generated key
- Loop ‘n’ times: Loop ‘n’ times
Step1: Run the new chaotic equation for n steps and define Begin

(x × r) + (1 − x)
 Run Lorenzo Model for n Steps:
f (x) = %1 xnew = p. (x − y),
0.9 × r
ynew = (x. z) + (r. x) − y,
Step 2: If ‘x’ is larger than 0.5, put ’1’ to ’binary string’; znew = (x. y) + (t. z)
Else, append ’0’. Return the Key
Step 3: If the length of the ‘binary string’ reaches 8: End
a. Convert ‘binary string’ to an integer value.
b. Add the integer value (byte) to the ‘key’ byte array.
c. Reset ‘binary string’ to an empty string for the next accurate output data as follows.
set of bits.
Step 4: Return the generated ‘key’ byte array. x = p · (x − y), (2)
y = (x.z) + (r.x) − y, (3)
generates two S-boxes and uses circular permutations instead z = (x.y) + (t.z). (4)
of mixed columns. Figure 1 shows our proposed algorithm.
As illustrated in Figure 1, our proposed algorithm has the where x, y and z represent the input values p, r, t are parame-
following improvements. ters with p = 10, r = 28, t = 8/3 while x1 , x2 and x3 are the
input values.
1) KEY GENERATION OF SYSTEM Algorithm 2 shows the process of generating security keys
key generation is performed using two proposed methods. using the proposed 3D-Lorenz map.

a: A NOVEL CHAOTIC MAP FOR KEY GENERATION 2) AES ENHANCEMENT

This work presents a new chaotic map that has been derived This study presents a novel method for creating improved
from logistic maps and experiments, which allows the devel- S-boxes with enhanced cryptographic features, including
opment of a chaotic pseudo-random number generator. bit independence criteria, periodicity, algebraic complexity,
The generated pseudo-random numbers pass randomness strict avalanche criteria, and distance to SAC. Furthermore,
tests with uniform distribution. A technique for s-box pro- it substitutes circular permutation at the bit level for the mix
duction is proposed, using an encryption key with a high column function of the conventional AES algorithm.
correlation with s-box generation.
Expanding the map enhances system complexity and key a: S-BOX GENERATION
space, improving cryptosystem security. Additional statistical The 1ST Proposed Dynamic S-Box and inverse S-Box
analyses and computer simulations validate the suggested generation for odd rounds
map’s high level of security. The First Step: The s-box is generated by selecting two
The proposed function that generates random pseudo num- shared bytes from the sender and recipient and converting
ber is define as follows. them to a binary representation of these bytes. The binary

(x × r) + (1 − x)
 representation is then subjected to an XOR with two adjacent
f (x) = %1 (1) bits until 15 bits are obtained and stored in a 16 × 16 blank
0.9 × r
matrix. Then, bit number 16 is obtained by XOR bit number
where x is the generated values selected between 0 and 1, r is 16 with the first bit of the new array, and bit number 17 is
the initial value. obtained by XOR with the first and second bit of the new
Algorithm 1 shows the full key generation process using array, and so on until 256 bits are obtained as illustrated in
the proposed f function. Figure 2, for example, the two bytes ‘rz’ can be presented
in binary as ‘0111001001111010. The primary bytes in the
b: LORENZO MAPS 3D FOR KEY GENERATION banner representation are ignored to generate fresh starting
A chaotic map in general has three categories of dimensions. values, even if the generation method is announced and even
The 3D-Lorenzo chaotic map approach uses three variables if the s-box itself is known, enhancing the algorithm’s secu-
to represent the geometric position of points in space [28]. rity and making it difficult to predict the s-box values. The
In our study, this approach is employed due to its simplicity, binary representation is converted to bytes, yielding a 2 ×
allowing the utilization of the user input, and generating 16-byte (32-byte) matrix.

95876 VOLUME 12, 2024

A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

FIGURE 1. Proposed enhanced AES structure.

FIGURE 2. First step to S-box generation.

The Second Step: Another 2 × 16 bytes (32) are obtained The Third Step: The first and second phases are combined
by shifting to the right. The shifting is made by the values to create a matrix (16 × 4) of 64 bytes. A mask is generated
of the second digit after the decimal point of the value f(x) on this matrix, where its value helps make equal or nearly
defined in Equation (1); to avoid revealing the main values equal numbers of ones and zeros if the matrix is converted to
that were used to generate keys using 16 values, where each binary representation. This step creates a matrix of (16 × 8),
row is shifted with the corresponding value of these digits as or 64 bytes.
illustrated in Figure 3. This step obtained a new (16 × 2) byte The Fourth Step: The matrices that have been obtained
matrix (32 bytes). from the first, second, and third steps will be merged, and a

VOLUME 12, 2024 95877

 A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

TABLE 2. Time Comparative with standard AES.

Plain Text = {’00’ ‘11’ ‘22’ ‘33’ ‘44’ ‘55’ ‘66’ ‘77’ ‘88’ ‘99’‘aa’ ‘bb’ ‘cc’ ‘dd’ ‘ee’
‘ff’} Cipher Text=]Ù _D Ùå ù/ $û > -ç ãv LK Ï), Þ Ð@ -ž Èñ í p¦

FIGURE 3. The process of shifting the values using the second step of the
proposed algorithm.

Each number’s row and column values are returned to

complete the inverse S-box.

b: REPLACING THE MIX COLUMN WITH CIRCLE

PERMUTATION AT THE BIT LEVEL
The time-consuming mix column function has been replaced
with an inventive circular permutation function at the bit level
due to its speed and high randomness, making it suitable
FIGURE 4. The process of shifting the values using the fourth step of the for IoT. It reduces complexity and achieves the concepts of
proposed algorithm.
confusion and diffusion; this is accomplished by performing
a circular permutation between locations in the 4 × 4 byte text
(16 × 8) byte matrix is obtained. A shift is made to the left by matrix after converting it to its binary representation, which
the values in third digit after the decimal point as illustrated will be 4 × 32 bits in size. The change will include text in
in Figure 4 resulting in a new (16 × 8) matrix. the values, not just the locations, after returning it to its byte
The Fifth Step: In this step the matrices from four phases representation, as shown in Figure 5.
will be used to create a 256-byte matrix. This matrix is
subjected to simple equations to enhance complexity and IV. SIMULATION RESULTS AND DISCUSSION
security. Even if the two bytes in the first step are exposed, the The dimensions of plaintext blocks, key and algorithm sen-
s-box remains difficult to identify because creating the s-box sitivity, and comparative analysis lengths are addressed in
is related to knowing the key. this section. Simulations are performed on a Lenovo laptop
running Windows 10 (64-bit OS) with Python 3.7, equipped
x = 255 − expanded key, (5) with an Intel(R) CoreTM i5-5200U processor running at
y = 255 − final step[16][16], (6) 2.20GHz and 8 GB of RAM.
sbox = (x + y)%256. (7) Key generation is based on the parameters of the chaotic
map. The key will change drastically with even the smallest
where x, and y represent the stored key value and the final alteration. In terms of the algorithm, the s-box plays a crucial
s-box stored step, respectively. role. Two-byte inputs are necessary for the process of gener-
The SHA-256 hash method expands a key if its size is less ation.
than 256 bytes, eliminating duplicates and adding remain- The algorithm and the outcomes are very different because
ing integers between 0 and 255. This generation starts with any changes to these two bytes or the chaotic map’s param-
two bytes, expands them, and then forms an S-box. Note eters will result in a big change. The sbox cryptographic
(Appendix A illustrates the 1st S-box generation and its criteria tests (Balancedness, Nonlinearity,
inverse). Algebraic Complexity, Strict Avalanche Criterion, differ-
The 2nd proposed S-Box and S-Box Inverse for even ential cryptanalysis) were tested for both S-boxes for even
rounds generated by this equation: and odd cycles.
After deleting any duplicates, the remaining integers are Our study uses a chaotic map to generate a primary with
added in the range of [0, 255]. a length based on algorithm setting. The text is divided into
sbox[row][column] = (3 × (row × 16 + column) + j)%255 blocks of 16 bytes, with block padding applied if the final text
block is smaller than the permitted size.
(8)
Tests were acceptable showing that the two boxes enjoy
The 1st and 2nd S-Boxes Inverse Generation: high efficiency, safety and that the s-boxes implementation

95878 VOLUME 12, 2024

A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

FIGURE 5. The permutation process using the proposed algorithm.

FIGURE 7. Time Comparison (KB/s) with Other Algorithms(taken from [1],

[29], [30], and [31]).

column with circular permutation at the bit level. Table 2

shows the encryption and decryption process time for plain
text and its corresponding cipher text, which clearly indicated
that our proposed technique shows improved time measure.
FIGURE 6. The encryption and decryption images using our proposed Figure 6 shows our proposed techniques benchmarked with
algorithm.
the literature.
Table 3 shows various quality measures of NIST test for
our proposed technique the NIST data test is a standard-
time is short, only takes one millisecond. Also, the encryption ized evaluation process used to appreciate the performance
and decryption time of the algorithm is considered ideal. Our of different algorithms, especially those connected to data
proposed algorithm is applied to encrypt images for secu- handling, analysis, and security.
rity purposed while transmitted over wireless communication
channels. Figure 6 shows the results of the encrypted and V. CONCLUSION
decrypted images. The proposed work results in enhancing the AES algorithm
The effectiveness of our algorithm is evaluated by assess- for IoT devices by making it lightweight due to the rapid and
ing the CPU time for encryption, decryption, key setup, and stochastic nature of these devices, utilizing two key genera-
S-Box creation, comparing standard methods and enhanced tion methods: an innovative chaotic and Lorenzo’s 3D chaotic
AES with two dynamic S-Boxes, and replacing the mix maps; using two separate dynamism boxes, for even and odd

VOLUME 12, 2024 95879

 A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

TABLE 3. NIST test outcome. The process involves performing XOR for each two-bit adja-
cent bit, as explained in the first step.
[1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 0, 1, 1, 1, 0]
[1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 0, 1, 0, 0, 1, 1]
[1, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0]
[0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 0, 0]
[1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1]
[0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 1]
[0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1]
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1]
[0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 0, 1, 0, 1]
[0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 0, 1, 1, 1, 1, 1]
[0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1]
[1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0, 0, 1, 0]
[1, 0, 0, 1, 1, 0, 1, 1, 0, 1, 1, 0, 0, 1, 1, 1]
[1, 0, 1, 0, 1, 1, 0, 1, 1, 0, 1, 0, 1, 0, 0, 0]
[1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1]
[0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 1]
It is important to note that the two bytes used for the
beginning are not inserted into the S-box values; only the new
rounds. With high dynamism, its generation time does not values are obtained from them. In this step we get 16∗ 2(32
take more than 1.2 milliseconds. Changing one of the two byte).
bytes that have been used in generating the s-box or one of Second Step: The shift to the right is adjusted by the
the parameters of the chaotic map will lead to changing the number of values produced by the new chaotic function used
output of the s-box. The mix column function was replaced by to create the key, with values coming in second after the
an inventive function that involves cyclic permutations at the comma.
bit level. The updated approach offers flexibility and speed in Shift values: [1,8,7,2,8,3,2,6,5,3,7,3,4,6,6,2]
image retrieval, enhancing performance and efficiency in IoT The shift list represents the shift amount for each row in
data encryption, particularly for color and grayscale images, the matrix, resulting from the first step.
with an identical level of precision.
Moreover, NIST testing of our method revealed that it [0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0, 0, 1, 1, 1]
complies with the accepted encryption requirements. The [1, 0, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1]
updated AES algorithm, which complies with encryption [0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1]
requirements, has the potential to ensure safe encryption of
IoT data. The key generation approach, particularly in IoT [0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1, 1, 1]
applications, offers security and efficiency and meets encryp- [0, 1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0]
tion criteria, demonstrating how cryptographic algorithms [1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1]
can be improved to meet evolving cybersecurity needs. [0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0]
Future works will involve the use of the proposed tech-
niques for the encryption and decryption of video data. [0, 1, 0, 0, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]
Another direction for research will involve proposing new [1, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 1]
lightweight appropriate for IoT devices have the same prop- [1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 0, 1, 1]
erties as this method. [1, 1, 0, 0, 0, 0, 1, 0, 1, 1, 1, 1, 0, 0, 0, 1]
[0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0, 0]
APPENDIX A
Example of 1st S-Box Generation for Odd Rounds: [0, 1, 1, 1, 1, 0, 0, 1, 1, 0, 1, 1, 0, 1, 1, 0]
First Step: choose two bytes = ‘rz’ and convert them to it [1, 0, 1, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 1, 0]
is binary representation: [1, 1, 1, 0, 0, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1]
[1, 1, 0, 0, 0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 0]
[0, 1, 1, 1, 0, 0, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0] In this step, we get 16∗ 2 bytes (32 bytes).

95880 VOLUME 12, 2024

A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

Third Step: The two matrices that come from the first and
second steps should be combined to create a matrix that is
16 ∗ 4 bytes. Apply a mask to the new matrix.
Mask matrix = [[3, 10, 9, 7], [4, 6, 5, 4], [2, 8, 14, 1], [13,
15, 11, 12]]
The resulting array will be new 16 ∗ 4 bytes

Fifth Step: To create a 16 ∗ 16-byte matrix, com-

bine matrices from the first, second, third, and fourth
phases, then implement simple equations using a 256-byte
array.
Key=(b’\x8d\xc2_\x88\x14\xc83\xd8\xe5DF\xc7
Fourth Step: To create a 16 ∗ 8-byte matrix, merge \x7f\xceWp’)
the matrices from the first, second, and third steps. Key expansion by the SHA-256 hash algorithm.
Shift the matrix to the left by the new chaotic function values x = 255 – expanded key
in the third place after a comma. y = 255 – final step [16][16]
Shift values: [2, 6,2,3,7,4,0,6,2,2,6,1,1,0,0,6] s-box = (x + y) % 256
The shift list represents the shift amount for each row in We obtain this s-box after removing duplicate numbers and
the matrix, resulting from the first step. adding the remaining numbers between [0, 255].

VOLUME 12, 2024 95881

 A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

REFERENCES [14] R. B. Antonio, A. M. Sison, and R. P. Medina, ‘‘A modified generation of

S-box for advanced encryption standards,’’ in Proc. 2nd Int. Conf. Inf. Sci.
[1] S. A. Jassim and A. K. Farhan, ‘‘Designing a new lightweight AES
Syst., Mar. 2019, pp. 280–283.
algorithm to improve the security of the IoT environment,’’ IRAQI J.
Comput. Commun. Control Syst. Eng., vol. 22, no. 2, pp. 96–108, 2022. [15] A. Nitaj, W. Susilo, and J. Tonien, ‘‘A new improved AES S-box with
enhanced properties,’’ in Proc. Inf. Secur. Privacy, 25th Australas. Conf.,
[2] S. Y. Tarabay, A. Twakol, A. S. Samrah, and I. Yasser, ‘‘A secure and
2020, pp. 125–141.
efficient cryptography system based on chaotic maps for securing data
image in fog computing,’’ Int. J. Comput. Netw. Inf. Secur., vol. 15, no. 1, [16] F. J. D’souza and D. Panchal, ‘‘Advanced encryption standard (AES)
pp. 64–80, Feb. 2023, doi: 10.5815/ijcnis.2023.01.06. security enhancement using hybrid approach,’’ in Proc. Int. Conf. Comput.,
Commun. Autom. (ICCCA), May 2017, pp. 647–652.
[3] A. Ahmed, M. M. Madboly, and S. K. Guirguis, ‘‘Securing signal encryp-
tion based on reduced round homomorphic AES,’’ Int. J. Intell. Eng. Syst., [17] A. Yousif and A. H. Kashmar, ‘‘Key generator to encryption images based
vol. 16, no. 3, 2023, doi: 10.22266/ijies2023.0630.35. on chaotic maps,’’ Iraqi J. Sci., vol. 1, pp. 362–370, Feb. 2019.
[4] M. S. Fadhil, A. K. Farhan, and M. N. Fadhil, ‘‘A lightweight AES [18] D. S. Jat and I. S. Gill, ‘‘Enhanced advanced encryption standard with
algorithm implementation for secure IoT environment,’’ Iraqi J. Sci., vol. 1, randomised S box,’’ in Proc. 5th Int. Conf. Innov. Technol. Intell. Syst. Ind.
pp. 2759–2770, Aug. 2021. Appl. (CITISIA), Nov. 2020, pp. 1–6.
[5] M. Abd Zaid and S. Hassan, ‘‘Proposal framework to light weight cryptog- [19] N. Ibrahim and J. Agbinya, ‘‘Design of a lightweight cryptographic scheme
raphy primitives,’’ Eng. Technol. J., vol. 40, no. 4, pp. 516–526, Apr. 2022. for resource-constrained Internet of Things devices,’’ Appl. Sci., vol. 13,
[6] F. T. A. Hussien, A. M. S. Rahma, and H. B. A. Wahab, ‘‘A secure environ- no. 7, p. 4398, Mar. 2023.
ment using a new lightweight AES encryption algorithm for e-commerce [20] A. Kadhim and R. S. Ali, ‘‘Enhancement AES based on 3D chaos theory
websites,’’ Secur. Commun. Netw., vol. 2021, pp. 1–15, Dec. 2021. and DNA operations addition,’’ Karbala Int. J. Modern Sci., vol. 5, no. 2,
[7] A. T. Maolood, E. K. Gbashi, and E. S. Mahmood, ‘‘Novel lightweight pp. 1–23, Jul. 2019.
video encryption method based on ChaCha20 stream cipher and hybrid [21] A. T. Khudhair, E. K. Gbashi, and A. T. Maolood, ‘‘Novel dynamic
chaotic map,’’ Int. J. Electr. Comput. Eng. (IJECE), vol. 12, no. 5, p. 4988, S-box based on password key and circle map,’’ Iraqi J. Sci., vol. 1,
Oct. 2022, doi: 10.11591/ijece.v12i5.pp4988-5000. pp. 4767–4778, Sep. 2023.
[8] J. Rokan Naif, G. H. Abdul-majeed, and A. K. Farhan, ‘‘Internet of Things [22] O. C. Abikoye, A. D. Haruna, A. Abubakar, N. O. Akande, and E. O. Asani,
security using new chaotic system and lightweight AES,’’ J. Al-Qadisiyah ‘‘Modified advanced encryption standard algorithm for information secu-
Comput. Sci. Math., vol. 11, no. 2, pp. 45–52, Sep. 2019. rity,’’ Symmetry, vol. 11, no. 12, p. 1484, Dec. 2019.
[9] A. K. Farhan and E. K. Gbashi, ‘‘Chaotic system and DNA com- [23] R. Saha, G. Geetha, G. Kumar, and T.-H. Kim, ‘‘RK-AES: An
puting operations for image encryption based on pixels shuffling,’’ improved version of AES using a new key generation process
Al-Qadisiyah J. Pure Sci., vol. 26, no. 2, pp. 1–14, Apr. 2021, doi: with random keys,’’ Secur. Commun. Netw., vol. 2018, pp. 1–11,
10.29350/qjps.2021.26.2.1265. Nov. 2018.
[10] D. S. Jat and I. S. Gill, ‘‘Enhanced advanced encryption standard with [24] A. M. Abdullah, ‘‘Advanced encryption standard (AES) algorithm to
randomised round keys,’’ in Proc. 4th World Conf. Smart Trends Syst., encrypt and decrypt data,’’ Cryptogr. Netw. Secur., vol. 16, no. 1, p. 11,
Secur. Sustainability, Jul. 2020, pp. 381–386. Sep. 2017.
[11] Y. Hussain Ail and Z. A. H. Alobaidy, ‘‘Images encryption using chaos [25] N. Dalakoti, N. Gaur, and A. Mehra, ‘‘Hardware efficient AES for image
and random generation,’’ Eng. Technol. J., vol. 34, no. 1, pp. 172–179, processing with high throughput,’’ in Proc. 1st Int. Conf. Next Gener.
Jan. 2016. Comput. Technol. (NGCT), Sep. 2015, pp. 932–935.
[12] A. Gupta and A. Gupta, ‘‘A new technique of image encryption using mod- [26] A. Moneem S. Rahma and M. Abid Ali K, ‘‘To modify the partial audio
ified AES algorithm,’’ Int. J. Multidiscip. Innov. Res., vol. 1, pp. 228–2583, cryptography for Haar wavelet transform by using AES algorithm,’’ Eng.
2021. Technol. J., vol. 32, no. 1B, pp. 169–182, Jan. 2014.
[13] V. Ashqi Saeed and B. Haval Sadiq, ‘‘Image encryption based on AES [27] K. Shahbazi and S. Ko, ‘‘High throughput and area-efficient FPGA
algorithm and XOR operation,’’ Academic J. Nawroz Univ., vol. 12, no. 3, implementation of AES for high-traffic applications,’’ IET Comput. Digit.
pp. 533–539, Aug. 2023. Techn., vol. 14, no. 6, pp. 344–352, Nov. 2020.

95882 VOLUME 12, 2024

A. H. Ali et al.: Lightweight Image Encryption Algorithm Based on Secure Key Generation

[28] L. M. Jawad, ‘‘A new scan pattern method for color image encryption based HAYA ALASKAR received the M.Sc. degree in applied artificial intelligence
on 3D-lorenzo chaotic map method,’’ Multimedia Tools Appl., vol. 80, from the University of Exeter, in 2009, and the Ph.D. degree in computer
no. 24, pp. 33297–33312, Oct. 2021. science from Liverpool John Moores University, in 2014. She is currently an
[29] G. Manjula and H. S. Mohan, ‘‘Improved dynamic S-box generation using Assistant Professor with the College of Computer Science and Engineering,
hash function for AES and its performance analysis,’’ in Proc. 2nd Int. Prince Sattam Bin Abdulaziz University, Saudi Arabia. She has several
Conf. Green Comput. Internet Things (ICGCIoT), Aug. 2018, pp. 109–115. publications concentrated on using machine learning in various medical
[30] L. Sleem and R. Couturier, ‘‘Speck-R: An ultra light-weight cryptographic data, such as signals and images. Her research interests include artificial
scheme for Internet of Things,’’ Multimedia Tools Appl., vol. 80, no. 11, intelligence applications and data science.
pp. 17067–17102, May 2021, doi: 10.1007/s11042-020-09625-8.
[31] S. Niveda, A. S. Sakthi, S. Srinitha, V. Kiruthika, and R. Shanmugapriya,
‘‘A novel Simon light weight block cipher implementation in FPGA,’’
in Pervasive Computing and Social Networking. Cham, Switzerland:
Springer, 2022, pp. 159–170.
[32] R. S. Mohammed, ‘‘Design a lightweight authentication encryption based
on stream cipher and chaotic maps with sponge structure for Internet of
Things applications,’’ Int. J. Intell. Eng. Syst., vol. 16, no. 1, pp. 1–16, 2023.
[33] J. Daemen, S. Hoffert, M. Peeters, G. Van Assche, and R. Van Keer,
‘‘Xoodyak, a lightweight cryptographic scheme,’’ IACR Trans. Symmetric
Cryptol., vol. 1, pp. 60–87, Jun. 2020.

A’LAA HUSSEIN ALI received the bachelor’s

degree in computer sciences data security branch
from the University of Technology (UOT), Bagh-
dad, Iraq, in 2014, where she is currently pursuing
the degree in computer science, specializing in
data security. She has been a Programmer with the
Health Ministry of Iraq, since 2019. ABIR JAAFAR HUSSAIN (Senior Member, IEEE)
received the Ph.D. degree from The University
of Manchester (UMIST), U.K., in 2000. Her
Ph.D. thesis titled Polynomial Neural Networks
for Image and Signal Processing. She is currently
a Professor of image and signal processing with
EKHLAS KHALAF GBASHI received the bach- the University of Sharjah, United Arab Emirates,
elor’s and master’s degrees in computer sciences and also a Visiting Professor with Liverpool John
from the University of Technology (UOT), Bagh- Moores University, U.K. She is the Ph.D. Supervi-
dad, Iraq, in 1998 and 2005, respectively, and sor and an External Examiner of research degrees,
the Ph.D. degree in networks security from the including the Ph.D., and M.Phil. She has published numerous referred
Department of Computer Sciences, Technology research papers in conferences and journals in the research areas of neural
University. She has been a Faculty Member with networks, signal prediction, telecommunication fraud detection, and image
the Computer Sciences Department, UOT, since compression. She has worked with higher order and recurrent neural net-
2000, where she was the Head of Computer Secu- works and their applications to e-health and medical image compression
rity Branch, UOT, from 2016 to 2020. Her research techniques. She has developed with her research students a number of
interests include networks security (intrusion detection systems), data secu- recurrent neural network architectures. She is one of the initiators and chairs
rity, computer networks, comparative education and computer architecture, of the Development in e-Systems Engineering (DeSE) conference series.
image processing, and artificial intelligence (AI).

VOLUME 12, 2024 95883