This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS 1

Securing Data in Internet of Things (IoT) Using

Cryptography and Steganography Techniques
Manju Khari, Aditya Kumar Garg, Amir H. Gandomi , Senior Member, IEEE, Rashmi Gupta, Member, IEEE,
Rizwan Patan , and Balamurugan Balusamy

Abstract—Internet of Things (IoT) is a domain wherein which Constraints of the IoT include energy budget, connectivity, and
the transfer of data is taking place every single second. The computational power [2].
security of these data is a challenging task; however, security Although IoT devices have made life easier, little atten-
challenges can be mitigated with cryptography and steganog-
raphy techniques. These techniques are crucial when dealing tion has been given to the security of these devices. Currently,
with user authentication and data privacy. In the proposed the focus of developers is to increase the capabilities of these
work, the elliptic Galois cryptography protocol is introduced devices, with little emphasis on the security of the devices. The
and discussed. In this protocol, a cryptography technique is used data that is transferred over the IoT network is vulnerable to
to encrypt confidential data that came from different medical attack. This data is needed to be secured to protect the privacy
sources. Next, a Matrix XOR encoding steganography technique
is used to embed the encrypted data into a low complexity of the user. If there is no data security, then there is a possibil-
image. The proposed work also uses an optimization algorithm ity of data breach and thus, personal information can be easily
called Adaptive Firefly to optimize the selection of cover blocks hacked from the system. Some of the important concepts of
within the image. Based on the results, various parameters are IoT involve identification and authentication. These concepts
evaluated and compared with the existing techniques. Finally, are inter-related to each other as cryptographic functions that
the data that is hidden in the image is recovered and is then
decrypted. are necessary to ensure that the information is communicated
to the correct device and if the source is trusted or not. With
Index Terms—Confidential data, cryptography, data security, the lack of authentication, a hacker can easily communicate
Internet of Things (IoT), steganography, user authentication.
to any device.
Whenever two devices communicate with each other, there
is a transfer of data between them. The data can also be very
sensitive and personal. Therefore, when this sensitive data is
I. I NTRODUCTION moving from device to device over the IoT network, then there
HE INTERNET of Things (IoT) is a network of
T connected vehicles, physical devices, software, and elec-
tronic items that facilitate data exchange. The purpose of IoT
is a need for encryption of the data. Encryption also helps to
protect data from intruders. The data can be easily encrypted
with the help of cryptography, which is the process of convert-
is to provide the IT-infrastructure for the secure and reliable ing simple text into unintelligible text. The primary objectives
exchange of “Things” [1]. The foundation of IoT mainly con- of cryptography are confidentiality, integrity, nonrepudiation,
sists of the integration of sensors/actuators, radio frequency and authentication. Elliptic curve cryptography (ECC) is one
identification (RFID) tags, and communication technologies. of the cryptographic algorithms that is used in the proposed
The IoT explains how a variety of physical items and devices work. ECC is a public key cryptographic technique based on
can be integrated with the Internet to permit those objects to the algebraic structure of elliptic curves over finite fields.
cooperate and communicate with each other to reach common In addition, to the cryptographic techniques, another
goals. The IoT consists mostly of little materials that are asso- method, named steganography is used in the proposed work
ciated together to facilitate collaborative calculating situations. which helps to provide additional security to the data.
Steganography hides encrypted messages in such a way that no
Manuscript received July 22, 2018; revised October 16, 2018; accepted one would even suspect that an encrypted message even exists
February 20, 2019. This paper was recommended by Associate Editor Y. Yuan.
(Corresponding author: Amir H. Gandomi.) in the first place. In modern digital steganography, encryp-
M. Khari, A. K. Garg, and R. Gupta are with the Department of tion of data occurs using typical cryptographic techniques.
CSE, Ambedkar Institute of Advanced Communication Technology and Next, a special algorithm helps to insert the data into redun-
Research, New Delhi 110031, India (e-mail: manjukhari@yahoo.co.in;
adityagarg2607@gmail.com; rashmig71@yahoo.com). dant data that is part of a file format, such as a JPEG image.
A. H. Gandomi is with the Analytics and Information Systems, School of The proposed work uses Matrix XOR steganography to pro-
Business, Stevens Institute of Technology, Hoboken, NJ 07030 USA (e-mail: vide additional security. The image block is optimized with
a.h.gandomi@stevens.edu).
R. Patan and B. Balusamy are with the School of Computing Science the help of Adaptive Firefly algorithm in which the encrypted
and Engineering, Galgotias University, Greater Noida 201310, India (e-mail: data is hidden in a selected block from a huge image block.
prizwan5@gmail.com; kadavulai@gmail.com). The remaining part of this paper is organized as follows.
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. Section II discussed work that relates to the security of
Digital Object Identifier 10.1109/TSMC.2019.2903785 IoT communication. Section III proposes methodology for
2168-2216 c 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS

improvements in medical data security. Section IV describes 2) break-glass access. In standard situations, a medical worker
the experimental setup and results. Section V concludes can decrypt and access data if the attribute set satisfies the
this paper. access policy of a medical file. In an emergency, a break-glass
access mechanism is used that can bypass the access policy
of the medical file so that emergency medical care workers or
II. R ELATED W ORK rescue workers can access the data in a timely fashion.
This section presents an overview of existing studies of Safety and confidentiality of information sent over the IoT
healthcare data security within the IoT network. network is a priority for the healthcare and medical indus-
Daniels et al. [3] introduced security microvisor (SμV) tries. Bairagi et al. [11] developed three methods for hiding
middleware, which uses software virtualization and assem- information so that communication over the IoT network can
bly level code verification to provide memory isolation and be preserved with the help of steganography. Information is
custom security. Banerjee et al. [4] presented energy-efficient hidden in the deepest layer of the image with the help of
datagram transport layer security (eeDTLS), which is a low- minimal distortion in the least significant bit (LSB) and the
energy variant of datagram transport layer security (DTLS) sign of the information can also be utilized. This technique
that had the same security strength but a lower energy improved imperceptibility and ability when compared to the
requirement. actual method.
Manogaran et al. [5] proposed a system in which medi- Huang et al. [12] presented a steganography scheme that
cal sensor devices are embedded in the human body to collect employs vector quantization (VQ) transformation in which
clinical measurements of patients. Significant changes in respi- LSB embeds secret data into a cover image. In the first
ratory rate, blood pressure, heart rate, blood sugar, and body level, the pixels of a 4 × 4 VQ-transformed image block are
temperature that exceed standard levels are detected by the separated into two different groups: 1) the LSB group and
sensors, which generate an alert message containing relevant 2) the secret data group. In the second level, VQ indexes are
health information that is sent to the doctor, with the help embedded in the LSB group and secret data are embedded in
of a wireless network. This system uses a vital management the secret group. Shanableh et al. [13] proposed the flexible
security mechanism to protect large amounts of data in the macro-block ordering (FMO) feature of H.264/AVC to hide
industry. message bits. The macroblocks are assigned to arbitrary slice
There is an urgent need for the securing the data that is groups with reference to the content of the message bits to be
transmitted every second over the IoT network. Some of the hidden. In the proposed method, a maximum payload of three
existing studies for data security are shown below. message bits per macroblock is achieved.
Sun et al. [6] proposed CloudEyes, a cloud-based anti- Liao et al. [14] proposed a new medical JPEG image
malware system. The proposed system provided efficient and steganographic scheme that is based on the dependencies of
trusted security services to the devices in the IoT network. interblock coefficients. The basic strategy that is used in this
Ukil et al. [2] studied the requirements of embedded security, paper consists of preserving the differences among discrete
provided methods and solutions for resisting cyber-attacks, cosine transform (DCT) coefficients at the same position in
and provided technology for tamper proofing the embedded adjacent DCT blocks as much as possible.
devices based on the concept of trusted computing. The development of IoT was related to the security of
Chervyakov et al. [7] provided a data storage scheme for end-user’s privacy and communication. However, the techni-
the least probability of data redundancy, data loss, and the cal heterogeneity, materials, and asymmetric nature of com-
speed of encoding and decoding, that can cope with differ- munication between the Internet and sensor nodes created
ent objective preferences, workloads, and storage properties. challenging security issues.
This analysis showed that if the selection of redundant residue
number system (RRNS) parameters is accurate, then it not
only allows increased safety and reliability but it also helps
III. P ROPOSED A PPROACH
to increase the speed of processing the encrypted data. The
applications used on IoT platforms generally require more A. Elliptic Galois Cryptography and Steganography Protocol
data than traditional applications. Raza et al. [8] presented This paper proposes the elliptic Galois cryptography (EGC)
lightweight secure CoAP for the IoT (Lithe), which helped protocol for protection against data infiltration during trans-
in the development of a novel DTLS header compression mission over the IoT network. In the proposed work, different
scheme designed to reduce energy consumption with the help devices in the IoT network transmit data through the proposed
of 6LoWPAN. Moreover, security is not compromised with the protocol as a part of the controller. The encrypted algorithm
DTLS header compression scheme. Vučinić et al. [9] proposed within the controller encrypts the data using the EGC proto-
object security architecture (OSCAR), which is the architec- col and then the encrypted and secured message is hidden
ture for end-to-end security in the IoT. OSCAR is based on the in layers of the image, with help from the steganography
concept that the security of an object is related to the security technique. The image can then be easily transferred throughout
of the application payload. the Internet such that an intruder cannot extract the message
Yang et al. [10] proposed the lightweight break-glass hidden inside the image. Initially, the EGC technique encrypts
access control (LiBAC) system in which medical files can confidential data. Subsequently, the encoded secret message
be encrypted in two ways: 1) attribute-based access and is inserted within the image by the XOR steganography
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

KHARI et al.: SECURING DATA IN IoT USING CRYPTOGRAPHY AND STEGANOGRAPHY TECHNIQUES 3

Let P and Q be two points on an elliptic curve such that [15]

U = GR (2)
where U is the public key, R is the private key, and G is the
generator point. An elliptic curve group is formed with the
help of this point only if no repeated factors are contained in
27h2 + 4g3 = 0(modP) and x3 + gx + h.
The mathematical structure for addition over GF (P), emu-
lating the conditions are given by the following equation.
Let P(x1 , y1 ) and Q(x2 , y2 ) be components of the elliptic
curve graph. Then, P + Q = (x3 , y3 ) where

Fig. 1. EGC. x3 = λ2 − x1 − x2 and y3 = λ(x1 − x3 ) − y1 . (3)

Then, λ resolves as

technique. Next, an optimization algorithm called the Adaptive 3x12 +a
if P = Q
Firefly algorithm is used to select a block in the image. λ= 2y1
y2 −y1
(4)
x2 −x1 if P = Q.
1) Elliptic Galois Cryptography: ECC, commonly known
as the public key encryption technique, is based on elliptic After the completion of the key generation step, encryption
curve theory. The keys are generated by using the properties can be performed with the chaotic neural network, as described
of elliptic curve equations instead of traditional methods. below.
The proposed work uses EGC (Fig. 1). For improving the a) Encryption based on the chaotic neural network:
efficiency of calculations and to reduce the complexities of Suppose there is a chaotic neural network with n inputs and
rounding errors, the elliptic curve over the Galois field (Fa ) is outputs. The input to this network is the plain text that is
used. The value of the Galois field must be greater than one. converted into cipher text with the help of ECC-based Galois
The elements of a Galois field GF(P) are as follows: field. The input text is denoted as (P1 , P2 , . . . , Pn ) and the
cipher text is denoted as (C1 , C2 , . . . , Cn ).
GF(P) = (0, 1, 2, . . . , P − 1)U Step 1: Generate a chaotic sequence
(P, P + 1, P + 2, . . . , P + P − 1)U
  (g(n), g(n + 1), . . . , g(n + n + 1)).
P2 , P2 + 1.P2 + 2, . . . , P2 + P − 1 U · · · U
  This chaotic sequence is known as the cipher text.
Pn−1 , Pn−1 +1, Pn−1 +2, . . . , Pn−1 +P − 1 [15] Step 2: The plain text is input into the network
(P1 , P2 , . . . , Pn ) and is then converted into the chain of
where P ∈ P and n ∈ Z+ . The order of the Galois field is
binary data. The binary sequence thus generated is denoted by
given by Pn and P is called the characteristic of the field. GF
(b1 , b2 , . . . , bn ) and the formula used to generate this binary
stands of Galois Field. The degree is at most n − 1 for each
data is as follows:
polynomial [15].
The secret key/public key is generated by the owner of the b(8n − 8)b(8n − 7) · · · b(8n − 2)b(8n − 1)
medical data and is used to access the authorized data. The
generation of the key is crucial as both the public key and where
the private key need to be generated. The message will be n = 1, 2, . . . , n.
encrypted by the user’s public key and the message can only
be decrypted with the help of a private key. Every user on the Step 3: The binary sequence that was generated in the
network develops key pairs that are used for encryption and previous step helps to perform weight factor generation, which
decryption. varies based on input functions. The equation for weight factor
An elliptic curve over a Galois field GF (P) greater than generation is as follows:
three (P > 3) that was formed with the help of variables g  
1 if b(j + 8 × n) = 0
and h within the field GF (P) and elements such as (x, y) Wj = . (5)
−1 if b(j + 8 × n) = 1
gives the following equation [15]:
According to the above equation, there are different weight
y2 = x3 + h mod P + gx. (1) factors for different inputs and the value of j varies from zero
to seven.
For different inputs and values of h and g, different ellip- Step 4: After the creation of weight factors, a bias function
tic curve points with values x and y exist that are within the is introduced for every chaos, which is generated by using
range of the Galois field [x, y belongs to GF (P)]. The public weight factors. The problem of singularity is avoided with the
key thus generated is a random point that lies on the elliptic help of this bias function.
curve and the random number generated is the private key. The equation for the bias function is as follows:
Multiplication of the private key by the generator point G in  
the curve provides the public key. dj = f Wj × dj (6)
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS

Fig. 2. Proposed Matrix-XOR steganography technique.

Fig. 3. Adaptive Firefly optimization.

dj is the bias function, Wj is the weight factor, and dj is the where the binary data bit is D and C is the binary image
input function. bit block. Two conditions must be satisfied to carry out this
Step 5: The cipher text is generated, based on the weight embedding process.
factor and the input function Condition 1: For the two blocks, if the XOR operation results
in a zero, then there is no requirement to change the last bit
g(i) = xn (n) × (1 − yn ) + dj . (7)
position.
In the above equation, (xn , yn ) are the points on the elliptic Condition 2: For the two blocks, if the XOR operation does
curve (secret key). After creation of the cipher text, this text not result in zero, then there is a change of the cover block
is stored on a cloud platform with the help of the Matrix XOR (i.e., zero to one or one to zero).
Encoding steganography technique. After deciding the bit position for the cover block, the
b) Matrix XOR: Matrix XOR is a technique for hid- embedding process is executed based on the following
ing encrypted data in which the encrypted data is hidden equation:
inside the H.264 video file [16]. For this technique, the Firefly   

Me = (d(i) ⊕ c(i))c (i) + (d(i) ⊕ c(i)) c (i) . (9)

optimization technique is used to optimize the blocks of the
image. With the help of this optimization technique, block Optimization of the cover block is handled with the help of
selection among the whole image is possible. the Firefly algorithm, as discussed below.
The proposed OM-XOR steganography technique is shown Step 3 (Adaptive Firefly Optimization): The Adaptive Firefly
in Fig. 2. The initial image is tiled and the secret data is algorithm (Fig. 3) is described by these three standard rules.
hidden on the cover block with the help of Adaptive Firefly 1) All the fireflies are unisex so that all fireflies are attracted
optimization. The tiled image is recombined and decoded. to each other.
Finally, the encrypted message is decoded by using the secret 2) Attractiveness between the fireflies is proportional to
key. their brightness; thus, a less bright firefly will move
Step 1 (Permutative Straddling): When there is no need to toward a brighter one. With increased distance between
use the full size to hide the encrypted message, the fragment fireflies, both the attractiveness and brightness decrease.
of the image remains unused. Permutative straddling is used to 3) The brightness of a firefly is determined by the landscape
eliminate this problem. This technique scatters the secret mes- of the objective function. Two important issues persist in
sage over the complete carrier medium; i.e., over the complete the Firefly algorithm: a) formulation of the attractiveness
image. Permutation depends on a key-based password. If the and b) the variation of light intensity [17].
user has the correct key, the same permutation can be repeated. The main motive is to reduce the count of the pixel blocks.
Step 2 (Encoding): There are many algorithms for embed- For this, some control parameters are initialized, such as:
ding secret information into an image block. By intro- 1) initial brightness b = 0.4;
ducing the Matrix XOR encoding technique, the proposed 2) randomization parameter η = 0.5;
work enhances the embedding efficiency. The conversion 3) coefficient of light absorption γ = 0.5.
of triple(f , k, g(i)) to quad(e, k, g(i)) and the compression The following equation is used to calculate multiobjective
of the encrypted message enhances the efficiency of this functions [17]:
technique.
The Matrix XOR technique embeds the g(i) chaotic sequence f (x) = [min{(c)}]. (10)
(secret data) in the optimized image block (cover block). In
this process, the one-bit block from the cover block is replaced For a given medium with a fixed light absorption coefficient,
with the encrypted information block. The one-bit embedding the light intensity I varies with the distance c, according to
process is carried out using the following equation: the following equation [17]:

Me = D ⊕ C (8) I = I0 e−γ c (11)

 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

KHARI et al.: SECURING DATA IN IoT USING CRYPTOGRAPHY AND STEGANOGRAPHY TECHNIQUES 5

where I0 is the original light intensity, In (10), the cost func-

tion “c” should be minimal in order to optimize the number
of blocks of an image. As attractiveness of fireflies is directly
related to LI with a neighboring firefly, β is defined as
follows [17]:

β(r) = β0 e−γ c ;
k
k≥1 (12)
where β0 is defined as attraction per c = 0 and γ is defined as
the illuminated saturation immersion coefficient. γ = [0, ∞]
and γ = 1. In certain circumstances, β0 is expressed as β0 = 1
and k = 2 since most fireflies are visible due to a constrained
separation. (a)
Suppose that, for two fireflies i and j, room arrangement is n-
dimensional. The distance between the xi and yi , the individual
count can be create using Cartesian (c) count calculation, as
follows:
 
 d
  2  
c= x −x + y −y . (13)
j i i j
k=1

Next, a time-varying AW is developed to update the position

of the fireflies. Over the period, the AW diminishes. The AW
value is updated with the following equation:
m 
Snew (t + 1) = f (x) + β0 e−γ c xj − xi + αεi . (14)
Equation (14) shows that the ith firefly moves toward the jth (b)
firefly, which is more attractive in the optimization.
Fig. 4. Matrix XOR stegno image and cover image. (a) Cover image.
Where location of the ith firefly is Snew (t + 1), after t + 1th (b) Stegno image.
interchange; randomization parameter is α; α ∈ [0, 1] in place
of major difficulties α = 0.1. The random numbers vector
is εi . Thus, the ith element is updated as follows: H.264, and each video has 30 frames/s. Video resolution is
Snew = Sold + Snew (t + 1). (15) 304 × 204 pixels.
The cover block image, which is the converted frame for
Thus, complexity is reduced with the help of the the H.264/AVC video, is depicted in Fig. 4(a). Fig. 4(b) cor-
Firefly optimization algorithm and helps to achieve better responds to the stegno image in which medical data is hidden
performance and results in the case of block selection. in the profound layer of the cover block. Thus, with the help
c) Data retrieval: On the receiver side, data is retrieved of the proposed method, the medical data of the patient is
with the help of the OM-XOR retrieval process. Encrypted securely hidden within the image.
data that was stored in the cloud is retrieved with the OM-XOR The data that is hidden with the proposed method cannot
decoding process. Finally, the decoded information that was be extracted by any unauthorized person since only the ECC
encrypted using the steganography technique is then decrypted secret key can retrieve the original data.
with the help of CNN decryption using the private key of the
user.
B. Parameter Evaluation
IV. E XPERIMENTAL R ESULTS AND D ISCUSSION For showing the efficiency of the proposed EGC system,
embedding efficiency, carrier capacity, peak signal to noise
The MATLAB simulation tool is used for implementation
ratio (PSNR), mean square error (MSE), and time complex-
of the proposed work. The proposed work uses H.264/AVC
ity were evaluated. The results of all these parameters were
video standards.
compared to some of the existing methods, such as LSB
steganography, FMO steganography, and optimized modified
A. Database Description matrix encoding (OMME) steganography. Various graphs have
Some official videos, such as MPEG 2 and MPEG 4 help to been put-up to efficiently show the comparison between the
achieve a high compression ratio. Although high-performance proposed work and the existing methods. The parameters are
efficiency is achieved, there is a loss in the compression evaluated as follows.
ratio [18]. This problem is solved with supreme progressive 1) Embedding Efficiency (Eη ): Embedding efficiency is
film coding model up-to-date. H.264 MPEG-4 is similarly used defined as the rate of the amount of secrets bits stored
for this process and the video quality of this H.264/AVC is and embedded in an image block, which helps to show the
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS

Fig. 5. Time complexity comparison analysis. Fig. 6. MSE comparison analysis.

effectiveness of the system

k+1
Eη = n (16)
k
where k is the bit in the cover block and n is the total quantity
of secret bits within the embedding technique.
2) Carrier Capacity (C): Carrier capacity is the ability of
the system to hide encrypted data inside the cover block.
The value of carrier capacity is directly proportional to the
performance
Total number of secret bits
C= . (17)
Number of bits in the cover block
Carrier capacity is the other name for hiding capacity and is
measured in terms of bits per pixel (BPP). Fig. 7. PSNR comparison analysis.
3) Mean Square Error: MSE is the amount of similarity
and the range of distortion in an image and it also helps in
the measurement of the amount of reliability
1 
n
MSE = (X − Y)2 . (18)
N
i=X,Y

N is the total pixel within the image, X is the initial image,

and Y is the final Stegno image.
4) Peak Signal to Noise Ratio: PSNR calculates the invisi-
bility of the image. PSNR can also be used for dynamic range
images
2552
PSNR = 10 log10 . (19)
MSE
Time Complexity: Time complexity is the amount of time Fig. 8. Carrier capacity analysis.
taken between the encryption and decryption process. To
increase the efficiency of the system, time complexity must
be lowered.
Comparative analyses are shown in Figs. 4–6. respectively), as depicted in Fig. 7. Fig. 8 shows that the
The EGC protocol gave better performance regarding proposed protocol yielded better carrier capacity performance
embedding efficiency, carrier capacity, PSNR, MSE, and time as compared to LSB, FMO, and OMME (0.33%, 16.35%,
complexity as compared to other techniques, such as LSB and 9.36% better performance, respectively). As shown in
steganography, FMO steganography, and OMME steganog- Fig. 9, the proposed protocol yielded better embedding effi-
raphy. As depicted in Figs. 5 and 6, the MSE and time ciency performance as compared to LSB, FMO, and OMME
complexity of the proposed EGC protocol is very low, as (31%, 21%, and 1.16% better performance, respectively).
compared to existing methods. The proposed protocol yielded Therefore, overall the proposed method is well optimized
better PSNR performance as compared to LSB, FMO, and and yielded better results when compared to the existing
OMME (32.42%, 45.62%, and 52.24% better performance, protocols.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

KHARI et al.: SECURING DATA IN IoT USING CRYPTOGRAPHY AND STEGANOGRAPHY TECHNIQUES 7

[11] A. K. Bairagi, R. Khondoker, and R. Islam, “An efficient stegano-

graphic approach for protecting communication in the Internet of Things
(IoT) critical infrastructures,” Inf. Security J. Glob. Perspective, vol. 25,
nos. 4–6, pp. 197–212, 2016.
[12] C.-T. Huang, M.-Y. Tsai, L.-C. Lin, W.-J. Wang, and S.-J. Wang,
“VQ-based data hiding in IoT networks using two-level encoding
with adaptive pixel replacements,” J. Supercomput., vol. 74, no. 9,
pp. 4295–4314, 2018.
[13] T. Shanableh, “Data hiding in MPEG video files using multivariate
regression and flexible macroblock ordering,” IEEE Trans. Inf. Forensics
Security, vol. 7, no. 2, pp. 455–464, Apr. 2012.
[14] X. Liao, J. Yin, S. Guo, X. Li, and A. K. Sangaiah, “Medical JPEG
image steganography based on preserving inter-block dependencies,”
Comput. Elect. Eng., vol. 67, pp. 320–329, Apr. 2018.
[15] C. J. Benvenuto, Galois Field in Cryptography, Univ. Washington,
Seattle, WA, USA, 2012.
[16] T. Wiegand, G. J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview
Fig. 9. Embedding efficiency analysis. of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst.
Video Technol., vol. 13, no. 7, pp. 560–576, Jul. 2003.
[17] A. H. Gandomi, X. S. Yang, and A. H. Alavi, “Mixed variable struc-
tural optimization using firefly algorithm,” Comput. Struct., vol. 89,
V. C ONCLUSION nos. 23–24, pp. 2325–2336, 2011.
The EGC protocol generated high levels of data security to [18] R. Hegde and S. Jagadeesha, “An optimal modified matrix encoding
technique for secret writing in MPEG video using ECC,” Comput. Stand.
serve the purpose of protecting data during transmission in the Interfaces, vol. 48, pp. 173–182, Nov. 2016.
IoT. With the novel ECC over Galois field, the proposed EGC
protocol provided better security. Due to the enhanced embed-
ding efficiency, advanced data hiding capacity can be achieved.
With the help of the proposed protocol and Adaptive Firefly
optimization, any amount of data can be easily transmitted
over the IoT network securely hidden within the profound lay-
ers of images. Performance is evaluated with parameters, such
as embedding efficiency, PSNR, carrier capacity, time com-
plexity, and MSE. Finally, the proposed work is implemented Manju Khari received the B.Tech. degree from the
in a MATLAB simulator, and approximately 86% steganog- Indira Gandhi Institute of Technology under Guru
raphy embedding efficiency was achieved. Results from this Gobind Singh Indraprastha (GGSIP) University,
New Delhi, India, in 2004, the M.Tech. degree in
proposed protocol were compared to existing methods, such information security from the Ambedkar Institute of
as OMME, FMO, and LSB. Technology under GGSIP University in 2010, and
the Ph.D. degree in computer science and engineer-
ing from the National Institute of Technology Patna,
Patna, India, in 2016.
R EFERENCES She is an Assistant Professor with the Ambedkar
Institute of Advanced Communication Technology
[1] R. H. Weber, “Internet of Things—New security and privacy challenges,”
and Research, Under Govt. of NCT Delhi affiliated with Guru Gobind Singh
Comput. Law Security Rev., vol. 26, no. 1, pp. 23–30, 2010.
Indraprastha University. She is also the Professor-in-Charge of the IT Services
[2] A. Ukil, J. Sen, and S. Koilakonda, “Embedded security for Internet of the Institute and has experience of over 12 years in Network Planning and
of Things,” in Proc. 2nd Nat. Conf. Emerg. Trends Appl. Comput. Sci. Management. She has 70 published papers in refereed National/International
(NCETACS), Mar. 2011, pp. 1–6. Journals and Conferences, such as IEEE, ACM, Springer, Inderscience, and
[3] W. Daniels et al., “SμV-the security microvisor: A virtualisation-based Elsevier, 6 book chapters in a Springer. She has also coauthored two books
security middleware for the Internet of Things,” in Proc. ACM 18th published by NCERT of XI and XII. Her current research interests include
ACM/IFIP/USENIX Middleware Conf. Ind. Track, Dec. 2017, pp. 36–42. software testing, software quality, software metrics, information security,
[4] U. Banerjee, C. Juvekar, S. H. Fuller, and A. P. Chandrakasan, “eeDTLS: optimization, and nature-inspired algorithm.
Energy-efficient datagram transport layer security for the Internet of
Things,” in Proc. GLOBECOM IEEE Glob. Commun. Conf., Dec. 2017,
pp. 1–6.
[5] G. Manogaran, C. Thota, D. Lopez, and R. Sundarasekar, “Big data
security intelligence for healthcare industry 4.0,” in Cybersecurity for
Industry 4.0. Cham, Switzerland: Springer, 2017, pp. 103–126.
[6] H. Sun, X. Wang, R. Buyya, and J. Su, “CloudEyes: Cloud-based mal-
ware detection with reversible sketch for resource-constrained Internet
of Things (IoT) devices,” Softw. Pract. Exp., vol. 47, no. 3, pp. 421–441,
2017.
[7] N. Chervyakov et al., “AR-RRNS: Configurable reliable distributed data
storage systems for Internet of Things to ensure security,” Future Gener. Aditya Kumar Garg is currently pursuing the
Comput. Syst., vol. 92, pp. 1080–1092, Mar. 2019. B.Tech. degree in computer science with the
[8] S. Raza, H. Shafagh, K. Hewage, R. Hummen, and T. Voigt, “Lithe: Ambedkar Institute of Communication Technologies
Lightweight secure CoAP for the Internet of Things,” IEEE Sensors J., and Research, New Delhi, India.
vol. 1, no. 10, pp. 3711–3720, Oct. 2013. His current research interests include develop-
[9] M. Vučinić et al., “OSCAR: Object security architecture for the Internet ing Web Applications, image processing, machine
of Things,” Ad Hoc Netw., vol. 32, pp. 3–16, Sep. 2015. learning algorithms, and their applications.
[10] Y. Yang, X. Liu, and R. H. Deng, “Lightweight break-glass access
control system for healthcare Internet-of-Things,” IEEE Trans. Ind.
Informat., vol. 14, no. 8, pp. 3610–3617, Aug. 2017.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS

Amir H. Gandomi (SM’19) received the Ph.D. Rizwan Patan received the B.Tech. and M.Tech.
degree in engineering from the University of Akron, degrees in computer science and engineering
Akron, OH, USA, in 2015. from Jawaharlal Nehru Technological University
He used to be a Lecturer in several univer- Anantapur, Anantapur, India, in 2012 and 2014,
sities. He was a Distinguished Research Fellow respectively, and the Ph.D. degree in computer
in headquarter of BEACON NSF Center located science and engineering from the School of
at Michigan State University, East Lansing, MI, Computer Science and Engineering, Vellore Institute
USA. He is an Assistant Professor of Analytics and of Technology, Vellore, India, in 2017.
Information Systems with the School of Business, He is an Assistant Professor with the School
Stevens Institute of Technology, Hoboken, NJ, USA. of Computing Science and Engineering, Galgotias
He has published over 130 journal papers and four University, Greater Noida, India. He has published
books. Some of those publications are now among the hottest papers in the reputed 8 SCI journals and 20 free Scopus indexed journals, and also presented
field and collectively have been cited over 11 000 times with an H-index paper in National/International Conferences, published book chapters in CRC
of 53. He has been named as Highly Cited Researcher (top 1%) for two Press, IGI Global, Elsevier, and edited as books. He has two Indian patents.
consecutive years, 2017 and 2018, and One of the World’s Most Influential Dr. Patan is a Guest Editor of the International Journal of Grid and
Scientific Minds. He is currently ranked 19th in GP bibliography among over Utility Computing (Inderscience), Recent Patents on Computer Science, and
11 000 researchers. He is part of a NASA Technology cluster on big data, arti- Information Medical Unlock (Elsevier).
ficial intelligence, and machine learning. His current research interests include
global optimization and (big) data mining using machine learning and evolu-
tionary computations in particular.
Dr. Gandomi has also served as an Associate Editor, an Editor, and a Guest
Editor in several prestigious journals and has delivered several keynote/invited
talks.

Rashmi Gupta (M’13) received the B.Tech.

degree from the Institute of Electronics and
Telecommunication Engineers under Delhi
University, New Delhi, India, in 1997, and the M.E.
and Ph.D. degrees in electronics and communication
engineering from the Delhi College of Engineering
under Delhi University, New Delhi, in 2005 and Balamurugan Balusamy received the B.E.
2014, respectively. degree in computer science and engineering from
He was a Senior Engineer with Calcom Group Bharathidasan University, Tiruchirappalli, India,
of companies from 1991 to 1999, a Lecturer with in 2001, the M.E. degree in computer science and
the Electronics and Communication Engineering engineering from Anna University, Chennai, India,
Department, Hindu Institute of Technology, Sonipat, India, and a Senior in 2005, and the Ph.D. degree in computer science
Lecturer with the Maharaja Agrasen Institute of Technology, New Delhi, and engineering from VIT University, Vellore,
from 1999 to 2003. She is currently a Professor with the Electronics India, in 2015.
and Communication Engineering Department, Ambedkar Institute of He is a Professor with the School of Computing
Communication Technologies and Research (Govt. of NCT of Delhi). She Science and Engineering, Galgotias University,
has authored over 70 research papers in various renowned international Greater Noida, India. His current research interests
journal and conferences. Her current research interests include machine include big data, network security, and cloud computing. He is pioneer
learning, computer vision, and signal and image processing. researcher in the areas of big data and IoT and has published over 70 papers
Dr. Gupta is a Fellow Member of IETE. in various top international journals.