SECRET DATA TRANSMISSION USING STEGANOGRAPHY A thesis submitted in partial fulfillment of the requirements for the Degree of Doctor of Philosophy (PhD) in Electronic Engineering By Muhammad Zeeshan Muzaffar 0909-PDEE-006 Department of Electronic Engineering School of Engineering & Applied Sciences Isra University, Islamabad Campus August 2017
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
SECRET DATA TRANSMISSION USING STEGANOGRAPHY
A thesis submitted in partial fulfillment of the
requirements for the Degree of Doctor of Philosophy
(PhD) in Electronic Engineering
By
Muhammad Zeeshan Muzaffar
0909-PDEE-006
Department of Electronic Engineering
School of Engineering & Applied Sciences Isra University, Islamabad Campus
All rights reserved. No part of the material protected by this copyright notice
may be reproduced or utilized in any form or by any means, electronic or
mechanical, including photocopying, recording or by any information storage
and retrieval system, without the permission from the author.
ii
SECRET DATA TRANSMISSION USING STEGANOGRAPHY
By
Muhammad Zeeshan Muzaffar
(0909-PDEE-006)
Names of Supervisor
Signature: _________________ Dr. Ijaz Mansoor Qureshi Professor, Department of Electrical Engineering, Air University, Islamabad.
CERTIFICATE
It is certified that the research work contained in this thesis has been carried
out under the supervision of Prof. Dr. Ijaz Mansoor Qureshi, at Isra
University, Islamabad Campus is original. It is fully adequate, in scope and
quality, as a thesis for the degree of Doctor of Philosophy.
Signature: ____________________ Supervisor Prof. Dr. Ijaz Mansoor Qureshi Professor, Department of Electrical Engineering, Air University, Islamabad.
Signature: ____________________ External Examiner Prof. Dr. Abdul Jalil Professor, Department of Electronic Engineering International Islamic University Islamabad.
Signature: ____________________ External Examiner Dr. Ihsan ul Haq Principal ICT, Faculty of Engineering & Technology International Islamic University Islamabad.
iv
DEDICATED TO
PROPHET MUHAMMAD (P. B. U. H)
THE GREATEST SOCIAL REFORMER
&
MY WORTHY PARENTS
v
ACKNOWLEDGEMENT
I am thankful to pay my heartiest praises to the Almighty, the
beneficent and compassionate Allah, who blessed me to accomplish the
dream of my beloved parents. I am quite amiably obliged to my supervisor
Dr. I. M. Qureshi, for his massive patronization, extremely fervent guidance
and transcendent vision to ensue innovative concepts entailed to
Steganography. He has always provoked me to discover new heavens of
research. He has always demonstrated pristine paternal love and care for
me as his own child. Indeed, his profuse mentoring escorted me throughout
my life even other than studies.
I pay tribute to highly worthy and exalted teachers Dr. Aamir Saleem
Chaudhry, Dr. Aqdas Naveed Malik, Dr. Tanveer Ahmad Cheema, Dr. Abdul
Jalil, Mr. Rashid Bodla, and all other teachers who have taught and trained
me from nursery to this dignified level. Mr. Rashid Bodla is one of my
teacher who help me to explore the teaching capabilities of myself by giving
a special confidence of speaking in class.
Even All expressions and feelings are inferior and insufficient to level
their profuse love, encouragement and guidance, and also, cannot translate
all the immaculate sentiments; I abode in my heart for parents and friends.
I cannot afford to overlook to tribute all the remaining family
members, including my respectable brothers, and my beloved wife who,
remained balmy, caring, and complaisant and still are in all the arduous
moments are source of love and encouragement. Particularly, I am pleased
to appreciate my selfless and committed friends: Dr. Muhammad Shakeel,
vi
Dr. Atta ur Rehman and Dr. Adnan Aziz who constantly assisted me in
study sessions and course work. In the saga to sacrifices, Ghias Malik, is
appreciable, a true fellow.
vii
ABSTRACT
Cryptography, watermarking and steganography are among the
rapidly emerging techniques pertaining to sustain, authenticate, and exhume
the hidden data, especially when it is transmitted over a public
network/channel. In cryptography, the message is executed and encrypted
artistically that the intended message becomes incomprehensible. Whereas,
water marking technique conceals the data in some cover file quite tactfully
that the data engrossed in the host signal/ entity becomes imperceptible
which is likely to be authenticated later on. Likewise, steganography
approach veils the data in the cover signal indistinctly to deflect the
interception of undesired user. The exchange of encompassed message by
cryptographic system may create suspense for the intruder. But, contrary to
it, in steganography the user is less attracted to the hidden information.
Therefore, the harmonization of both of these technologies can produce
invisible higher level of message protection. In this dissertation, the problem
of imperceptibility, data rate and robustness is formulated and different
approaches are proposed and investigated to solve it.
In first approach, a novel technique to embed information into the
audio signals is proposed. In this regard, the set of all possible values of
amplitudes of audio signals are termed as “audio sample space”. An
algorithm is proposed to subdivide this sample space into subspaces and the
information was embedded into these subspaces. On the other hand, an
algorithm for decoding on receiver side is also proposed. The algorithm has
the capability to work on real time systems and provide sufficient security at
commercial level. The amount of imperceptibility achieved is, quite a distinct
viii
benefit concerning perceptual evaluation of speech quality (PESQ).
In second approach, an innovative steganography technique has
been engineered named as weighted pattern matching (WPM) which is
utilized to collaborate the lifting wavelet transform. The message bits are
insert in the indistinguishable places that are picked from the coefficients of
detail sub-bands by taking edge of the proposed WPM technique. WPM
captures the correlation between the message data block and detail
coefficients help us to configure the exact location that can contain the data
block invisibly. The ultimate results of the experiment exhibit that the WPM
technique enhances invisibility significantly in addition to lossless massage
retrieval.
In third approach, another sustained efficient, imperishable approach
brimmed of heavy payload, named compressive weighted pattern matching
(CWPM) has been invented and applied. CWPM technique has been born
from the combination of compressive sensing (CS) with WPM. Use of CS
provides the higher level of security and bigger payload by means of
compression and encryption. CWPM holds the position where data block can
be embedded on the basis of a weighted correlation.
Our engineered techniques have been compared with the well-known
steganography elaborated in literature review. The results ingenuity proves
that all designed models are far better firm and ingratiating.
ix
ABBREVIATIONS
Abbreviation Term
AWGN ---------------------------------------------------Additive White Gaussian Noise
BER------------------------------------------------------------------------------ Bit Error Rate
CERTIFICATE --------------------------------------------------------------------------- iii
ACKNOWLEDGEMENTS ------------------------------------------------------------ v
ABSTRACT ------------------------------------------------------------------------------- vii
ABBREVIATIONS------------------------------------------------------------------------ ix
TABLE OF CONTENTS --------------------------------------------------------------- xi
LIST OF TABLES ----------------------------------------------------------------------- xiii
LIST OF FIGURES --------------------------------------------------------------------- xiv
CHAPTER I – INTRODUCTION ---------------------------------------------------- 01 1. Problem Statement ------------------------------------------------------------------ 03 2. Contribution of the Thesis --------------------------------------------------------- 05 3. Organization of the Thesis -------------------------------------------------------- 06 CHAPTER II – LITERATURE REVIEW ABOUT STEGANOGRAPHY --- 08 1. Introduction ---------------------------------------------------------------------------- 08 2. Types of Steganography ---------------------------------------------------------- 08
2.1 Audio Steganography --------------------------------------------------------- 09 2.2 Video Steganography --------------------------------------------------------- 11 2.3 Image Steganography --------------------------------------------------------- 14 2.4 Text Steganography ----------------------------------------------------------- 15
2.1 The Sensing Problem --------------------------------------------------------- 20 2.2 Signal Representation and Sparsity -------------------------------------- 21 2.3 Incoherent Sampling ----------------------------------------------------------- 22 2.4 Under-Sampling and Sparse Signal Recovery ------------------------- 23
3. Summary ------------------------------------------------------------------------------- 24 CHAPTER IV – CHANGING SLOPE METHOD: A TIME DOMAIN APPROCH FOR AUDIO STEGANOGRAPHY --------------------------------- 25 1. Introduction ---------------------------------------------------------------------------- 25 2. System Model ------------------------------------------------------------------------- 25 3. Proposed Algorithm ----------------------------------------------------------------- 27 4. Performance Graphs ---------------------------------------------------------------- 32 5. Summary ----------------------------------------------------------------------------- 39 CHAPTER V – FREQUENCY DOMAIN TECHNIQUES IN AUDIO STEGANOGRAPHY ------------------------------------------------------------------ 41 1. Introduction -------------------------------------------------------------------------- 41 2. Lifting Wavelet Based Techniques ---------------------------------------------- 41
xii
2.1 Weighted Pattern Matching Lifting Wavelets Transform -------------- 42 2.2 Compressed Sensing for Security and Payload Enhancement in
Audio Steganography -------------------------------------------------------- 55 3. Summary ----------------------------------------------------------------------------- 76 CHAPTER VI – DISCUSSION AND CONCLUSIONS ------------------------ 77 1. Summary of Results -------------------------------------------------------------- 77 2. Future Directions ------------------------------------------------------------------ 80 REFERENCES ----------------------------------------------------------------------- 82
xiii
LIST OF TABLES
Chapter Description Page No.
IV – 1 PESQ values for different values of k --------------------------- 35
IV – 2 PESQ values for different methods ------------------------------ 35
V – 1 SNR/dB in case of text messages in WPM-LWT ------------ 50
V – 2 Possible φ's for Security Enhancement ------------------------ 67
V – 3 Compressibility ratio ------------------------------------------------ 72
xiv
LIST OF FIGURES
Chapter Description Page No.
I– 1 Magic triangle - three contradictory requirements of
involves third level of LWT so that the hidden message can be more
imperceptible as shown in Figure V-10. The outputs of preparing audio
segment section are approximation and detail coefficients where and
Audio Cover Signal
Preparing each Audio
Segment
Bits embedding location
selection
Secret Image
Encryption and
Compression
Bits Embedding
Reconstruct Audio
Transmit
Audio Segmentation
Segmentation
57
are third level approximation and detailed coefficients. is
concatenation of first and second level detail coefficients vectors of
segment .
Figure V-10: Preparing Audio Segments for Embedding
2.2.5 Encryption and Compression of Secret image: For the
sake of encryption and compression of our secret message, the concept of
compressed sensing (CS) to be used here. The Shannon’s sampling
theorem states that to recover a signal, the sampling rate must be at least
the Nyquist rate. Compressed sensing is based on the interesting fact that to
recover a signal that is sparse in some domain representation, one can
sample at the rate far below the Nyquist rate. This concept used here to
Segment
LWT of
58
decrease payload and enhance security as shown in Figure V-10
mathematically.
Figure V-11: Encryption and compression of secret image
Here in Figure V-10, is an image having dimensions
containing messages of each and are the bases vectors matrix of
the domain in which can be sparsely represented with same dimensions
as . Coefficients of the new domain are represented by . Now, is
of and which is sensing matrix having dimensions where
. So, which is our output of this module having dimensions . An
important fact is that matrix is only known at transmitting and receiving
side and hence introduce security and payload reduction.
𝑓 𝜓𝑥
𝑥 𝜓𝑇𝑓
𝑦 𝜑𝑥
59
2.2.6 Segmentation: This module includes the Segmentation of the
encrypted message having dimensions . As secret image is of bit
gray scale image. So there are 𝑠 bits in total to embed. Secret
message segmented into segments such that 𝑠 where be
the length of each message segment. So, now after segmentation, the
secret message will be of the form
3 (Eq. V.9)
Where each segment is of bits long. i.e.
(Eq. V.10)
2.2.7 Bits Embedding location Selection and Bits
Embedding: Bits embedding location selection receives inputs from two
modules which are “preparing audio segments” and “segmentation”. First
module outputs are , and while second module output is .
Bits embedding location depends upon the maximum correlation
between the and message segment . The following steps are
followed to perform the functionality given in Figure V-11
Map sign vector denoted by and 3 formed to keep the
signs of and , respectively.
Change all –ve values of and with their respective +ve
values and convert it to binary.
60
Now find the maximum correlation of message segment and
by using the following steps
o Find the first most significant bit of each coefficient of
which is equal to binary 1 represented by
,where and
⁄
⁄
⁄ , where
each represents the first MSB equal to 1 of coefficient
of segment.
o Now, two more factors are used to control the robustness
against LSB attack and tolerance of change can be
represented as and respectively. Where
and size of , if these condition does not
meet in any coefficient of a segment , then this
coefficient will not consider for correlation competition.
Where points the location count from LSB of
coefficient of segment and points the location
difference from in the direction of LSB of coefficient of
segment. i.e if , and , then the bits of
coefficient of segment to find correlation can be seen
in shaded area of figure below, so area from 5 location to
location is the ROI (region of interest). A necessary
condition is that message segment. points the
5 location
o Find correlation of with each ROI present in the
coefficients of each segment of cover audio.
61
o Find the index with which the correlation of is maximum.
o After finding the maximum correlation index, replace the
message segment with that particular coefficients ROI
present in and replace this index in for recovery
purpose. Now modified detail coefficients becomes
and respectively.
After finding the maximum correlation index, replace the message segment
with that particular coefficients ROI present in and replace this index
in for recovery purpose. Now modified detail coefficients becomes
and respectively.
62
Figure V-12: Bits Embedding location selection and Bits Embedding
2.2.8 Reconstruct Audio: After embedding secret message ,
reconstruction of voice is performed. Approximate, modified detail and map
sign vectors ( , , , and 3 ) are the inputs of this
module. To reconstruct the audio, first we have to convert the binary values
Preparing each Audio Segment
Segmentation
Map sign Vector of
Map sign Vector of
Change sign of negative
Change sign of negative
Change into binary
Change into binary
Change into binaryFind Max Correlation index
Find Max Correlation index
Replace position index in
Replace in Max correlation
index of
Reconstruct Audio
63
to equivalent decimals and place negative signs by the help of and
3 and use inverse transform as shown in Figure V-12.
Figure V-13: Audio Reconstruction
𝑀𝑆𝑉 𝑖 𝑀𝑆𝑉3𝑖 𝑐𝐷 �� 𝑐𝐷 𝑖
Change signs of 𝑐𝐷 �� using
𝑀𝑆𝑉3𝑖
Change signs of using
V
ILWT of and Break
into and
𝑐𝐴 ��
𝑐𝐷 �� ILWT of and
𝑐𝐷 �� ILWT of and
Stego Sound
64
Figure V-14: Reconstruction of secret message at Receiver
Received Audio Signal
Multilevel LWT of each
Segment
Min 𝑋 𝑙
Subject to
�� 𝜑𝑋 𝑙
𝜀
Recovered Secret Image
Received Audio
Segmentation
Multilevel IDWT of each X
𝑠𝐴 𝑡 𝑠𝐷 𝑡 𝑠𝐷 𝑡
Produce ��
𝑋
65
2.2.9 Receiver: At receiver, to recover the secret message from the
received audio , first audio segmentation is performed. Size of segment
should be the same as at transmitter. Received audio segments represented
by where . Now each . Secondly, multilevel LWT performed
uptill the level performed at transmitter which produces the 𝑠 (third level
LWT approximate), 𝑠𝐷 (third level LWT details) and 𝑠𝐷 (first and second
level LWT details). After that, 𝑠𝐷 provide the indexes present in 𝑠𝐷 in
which the located. Then after getting all the ’s, reassemble having
dimensions .
Now from these, we have to estimate the having dimensions
by the use of following problem solving using CVX
Minimize
(Eq. V.11)
Subject to
This passes through the ILWT process and finally get the having
dimensions .
2.2.10 Achievements of CS based WPM-LWT: There are three
main achievements can be seen through CS based WPM-LWT scheme
Multilevel high end security
High end payload compressibility
High normalized correlation
66
First level of security is the basic characteristics of the
steganography and second level of security comes from sensing matrix .
Possible number of ’s are very huge in number and only one of them is of
use to recover secret message. Possible number of ’s shown in Table V-2.
High payload compressibility shows the increase in payload in
percent showed in Table V-3. i.e, if compressibility is 5.6888, it means the
468.88% increase in payload
Normalized correlation measure the change in stego audio before
and after passing through AWGN. This shows that signal is robust against
high attack of AWGN.
2.2.11 Results of CS based WPM-LWT: In this section, the
authenticity of the proposed scheme is depicted through the MATLAB and
CVX simulations in terms of robustness, transparency and payload etc.
For the simulation, Is taken as multilevel two dimensional Discrete
wavelet transform and is the first rows of QR decomposition of a
Gaussian matrix of with { } as entries. So, is of . Nobody can
recover estimate of on receiver side without having . There are
possible ’s and it is near to impossible to check all of them. The possible
huge number of ’s show the level of security achieved here for different
simulations shown in Table V-2
67
Table V-2 : Possible φ's for Security Enhancement
Figure
Figure V-15 (b) 512 250
Figure V-15 (c) 512 200
Figure V-15 (d) 512 150
Figure V-15 (e) 512 120
Figure V-15 (f) 512 90
Figure V-16 (b) 512 150
Figure V-16 (c) 512 120
Figure V-16 (d) 512 100
Figure V-17 (b) 512 150
Figure V-17 (c) 512 120
68
Figure V-17 (d) 512 100
Table V-2 shows the level of security by showing the possible
number of ’s. the minimum number seen in table1 is 5 which
is a big huge number for all possible ’s. this shows that our proposed
scheme meets the security needs of steganography.
Figure V-14(a) shows an original Lina image of dimension 512x512
in grayscale. After applying compressed sensing and embedding it into to
cover audio, its recovered version is shown in Figure V-14(b) with
compressed sensing parameters n=512 and m=250. With m=250 means
104% increase in payload. The recovered image is very close to the original
image in terms naked eye test.
Further in Figure V-14(c) the compressed sensing parameter n is
kept same while m is taken as 200. With m=200 means 156% increase in
payload In this way we are taking lesser information (more compressed and
secure) for embedding in the cover audio. The result shows that the image is
degraded compared to Figure V-14(b) but still it is perceptually acceptable.
Similarly, in Figure V-14(d) the compression parameter m is taken as 150.
With m=150 means 241% increase in payload.
In Figure V-14(e) the compression parameter =120 is taken very
low. With m=120 payload increases 326.67%.
69
In Figure V-14(f) the is taken extremely low as 90. With =90
payload increases 468.88%. That means much secure and compressed but
degraded compared to the former cases.
In Figure V-15 another example is shown. In Figure V-15(a) original
sketch image is shown while in Figure V-15 (b) and (c) its recovered images
are depicted. In both cases the parameter is taken as 150 and 120 and
payload in both cases increases 241.33% and 326.66% respectively. The
recovered images still considerably recognizable. On the other hand, in
Figure V-15(d), for =100, the payload increases 412% but edges of the
images effect badly. This shows that as compression and payload increases,
the quality of the recovered images decreases.
Figure V-16 presented another example of text based images. Both
uppercase and lowercase letters are used in this example and all possible
letters of English used in the phrases. Figure V-16(a) shows the original
image of the text. In Figure V-16(b), (c) and (d), parameter is taken as 150
, 120 and 100 respectively. Figure V-16(b) and (c) shows some degradations
in quality of the images but are easily readable. But Figure V-16(d) shows
degraded badly but still have the recognizable text.
From these results, it is apparent that proposed scheme provides a
good payload with significant level of and security.
Another achievement provided by compressed sensing is the
compressibility ratio which also shows the payload enhancement per second
and can be formulized as;
70
Compressibility ratio =
(Eq. V.12)
Where be the original image and is the image passes through
CS process. Compressibility ratio for different images can be seen in Table
V-3.
Figure V-15: Secret message recovery for (a) Original (b) n=512,m=250 (c) n=512,m=200 (d) n=512,m=150 (e) n=512,m=120 (f) n=512,m=90
Figure V-16: Secret message recovery (a) Original (b) for m=150 (c) for m=120 (d) for m=100
(a)
(e)
(b) (c)
(d)
(f)
(a) (b)
(c) (d)
71
Figure V-17: Secret message recovery (a) Original (b) for n=512, m=150 (c) for n=512, m=120 (d) for n=512, m=100
For the sake of robustness test, normalized correlation (NC) is used
as a figure of merit. NC used to test the level of correlation between the
original and stego audio. Its value varies between 0 and 1. Formula for NC is
Normalized Correlation ∑
√∑
∑
(Eq. V.13)
Where, L is total number of samples in audio signal. and are
original cover signal and the received stego signal respectively. Figure V-17
shows the level of NC and shows the proposed scheme is highly robust
against AWGN. The stego signal is tested at different levels of AWGN but
the NC approaches to 1 at -25dBs for all kind of audio cover signals.
Different types of voices are investigated including human voices (male,
female) from different age groups and some natural sounds. It is worth
mentioning here that the robustness of scheme is even higher in the case of
(a)
(c)
(b)
(d)
72
human voices where the NC tappers off to 1 at -50dBs. That means the
proposed scheme is quite useful for day to day human communication. In
contrast to the scheme proposed in (Shahadi et al., 2014), where users
demonstrated that their scheme is robust at 35dB and higher, our scheme
saves 60dBs for natural sounds while 85dBs for human voices
Figure V-18: AWGN Vs Normalized Correlation
Table V-3: Compressibility ratio
Figure Compressibility ratio
Figure V-15 (b) 512 250 2.048
Figure V-15 (c) 512 200 2.56
73
Figure V-15 (d) 512 150 3.4133
Figure V-15 (e) 512 120 4.2666
Figure V-15 (f) 512 90 5.6888
Figure V-16 (b) 512 150 3.4133
Figure V-16 (c) 512 120 4.2666
Figure V-16 (d) 512 100 5.12
Figure V-17 (b) 512 150 3.4133
Figure V-17 (c) 512 120 4.2666
Figure V-17 (d) 512 100 5.12
The difference between the spectrograms of original cover audio
signal and the stego signal shown in Figure V-18 and Figure V-19. Analysis
shows that there are negligible small amount of changes between the
original and stego signal in the case of different female voices. This shows
that the proposed scheme does not introduce much changes in their
74
spectrograms. It is apparent from the figures that female voices spectra have
better performance compared to male voices. This is perhaps because the
female voices exhibit higher pitch compared to male voices. So the
frequency components are more crisp and confined in female cases
compared to male voices. Hence their spectra are less vulnerable to the
changes offered because of the proposed scheme. Since, in the proposed
scheme the data is embedded in detail components of LWT.
Figure V-19: Spectrogram analysis of three male voices
Figure V-20 shows the time domain changes in cover audio before
and after embedding the secret message. Six examples are taken to
analysis the time domain changes due to our proposed scheme and all the
examples shows that no significant change can be seen in the signals before
and after embedding in time domain.
75
In this technique, the compressed sensing is applied to the message
image prior to embedding in the cover audio by means of LWT. Here,
compressed sensing provides high level security because sensing matrix
is only known at transmitter and receiver and to estimate the real
among the huge number of combinations is nearly impossible.
The simulation results show that the proposed scheme promises a great
enhancement in payload, robustness, security and imperceptibility.
Figure V-20: Spectrogram analysis of three female voices
76
Figure V-21: Time domain analysis of original and stego signal
3. SUMMARY
Two techniques investigated in this chapter, first was WPM-LWT and
second was compressed sensing based WPM-LWT. First strategy provides
the imperceptibility and robustness while second provide the imperceptibility,
robustness, high payload as well as high level of security that cannot be
seen in some encryption systems.
77
CHAPTER VI
DISCUSSION AND CONCLUSIONS
Main contributions of the dissertation including achievements
regarding imperceptibility, robustness, security and payload enhancement
are summarized in this chapter. Some future directions have also been
pointed out.
1. SUMMARY OF RESULTS
In digital steganography, a secret message is concealed in the
digital content. In order to make this hidden information secure, robust and
imperceptible, the secret information is embedded in secret positions of the
digital cover media. Improving the security aspect of steganography system
is one of the challenges in the steganography domain. The steganography
schemes presented in this dissertation comprised of two parts. In the first
part, a time domain steganography scheme is presented in which the secret
data can be hide in slope of the cover audio while in second part, two
frequency domain audio steganography schemes are presented in which
secret message can be hide in wavelet coefficients.
In first part, a time domain scheme named “Changing Slope Method”
presented. This scheme contain the hidden bits in the slope of the line
joining the two consecutive amplitudes of a given digital 16 bit audio cover.
As a 16 bit audio can contain a maximum of 55 different levels and
78
a slight change in audio slope cannot be detected easily by human auditory
system (HAS). So, this method was highly imperceptible. Another advantage
of the scheme is that change in amplitude can result only in loss of leading
or trailing information chunk of hidden information and this error will not
propagate further. Secondly the PESQ score is 4.496 for k =1, 4.494 for k =2
and 4.489 for k= 3 which is very much close to 4.5 the maximum value.
While PESQ for existing networks is near about 3.8. Thirdly, the cross
correlation is just like the auto correlation of the original and stego signal
which also reflect that both voices are very close and cannot be segregated
by the human ear. Fourthly, the size of the stego file and the original file
cannot be changed even a single bit and this file memory minimize the
probability that no data is present in the file. This thing will not alert the
intruder which is provide another level of security. Fifthly, the graphs of
original and stego audios are also very much close and cannot be seen
without applying a high level zooming. This also saves the secure data from
intruder.
In the second part, two frequency domain schemes are presented to
solve the contradictory parameters of digital audio steganography i.e.
imperceptibility, robustness and payload.
In first scheme, a novel technique for digital audio steganography
using a weighted pattern matching approach in LWT is proposed. The
technique exhibits a very high level of robustness against common and well
known attacks like AWGN attack etc. In case of AWGN the signal can
survive at a very poor level of SNR that is -50dB while most of the schemes
in the literature afford same performance at 35dB or higher. That is why, in
79
our comparison text messages are taken that are highly sensitive to the
attacks and the scheme was able to recover them even in very hostile
conditions (below -50dB). Moreover, upon investigation through MATLAB
simulations it is clear that the scheme contains a high level of imperceptibility
as well. However, the maximum achievable capacity in the proposed
scheme is 60kbps, which makes is still suitable for many voice
communication channels like voice modems that support upto 56kbps etc.
The main focus of the scheme was to make the digital audio steganography
more robust and imperceptible, and these goals are achieved well with a fair
capacity payload.
In second scheme, the compressed sensing is applied to the
message image prior to embedding in the cover audio by means of LWT.
Here, compressed sensing provides high level security because sensing
matrix is only known at transmitter and receiver and to estimate
the real among the huge number of combinations is nearly
impossible. The simulation results show that the proposed scheme promises
a great enhancement in payload, robustness, security and imperceptibility.
Summary of the main contribution of the dissertation can be
highlighted and given below. The proposed scheme is;
Higher level of security due to compressed sensing technique which
cannot be seen even in encryption system while keeping the audio
imperceptibility.
Higher payloads which can be seen by higher compressibility ratio
due to compressed sensing.
Highly imperceptible and achieves a high level PESQ score.
80
Highly robust against AWGN attack so that text message can be
transmitted.
Higher normalized correlation between original and stego audios at
low SNR.
Having the same size of original and stego audios.
2. FUTURE DIRECTIONS
In future, a lot of directions can be explored for audio, images and
videos in the field of digital steganography. Few of these directions are given
as below
In the first part of dissertation, the scheme presented named
changing slope method investigated in time domain, this scheme can be
seen in frequency domain. For such scheme, many different types of
frequency domains can be investigated and their comparison can be
presented in case of different attacks. PESQ score can be also be visited for
little change.
In the Second part of dissertation, integer based lifting wavelets
used to transform used to transform the signal in frequency domain. Other
transformation techniques can also be investigated to improve performance
regarding NC or complexity.
On the other hand, we did not use any error correcting codes to
make our information more secure. Different error correcting codes can be
used to make the technique more robust.
81
It’s very hard to transmit a video as a secret message. In video
steganography, secure video in video cover can also be investigated by
applying some sort of higher level of compression techniques on payload.
82
REFERENCES
Al Azawi, A. F. and Fadhil, M. A., “An Arabic Text Steganography Technique Using Zwj And Zwnj Regular Expressions,” International Journal Of Academic Research Vol. 3. No. 3. May 2011.
Ali, A. H., Mokhtar, M. R. and George, L. E., "Recent Approaches for VoIP
Steganography." Indian Journal of Science and Technology” Vol 9, No. 38, 2016.
Anil, S., Mythili, T., and Hazra, S., "A Novel Rgb Based Steganography
Using Prime Component Alteration Technique." IIOAB JOURNAL Vol. 7, Issue 5, pp 58-73, 2016.
Asad, M., Gilani, J., Khalid, A., “An Enhanced Least Significant Bit
Modification Technique for Audio Steganography”.2011. Avval, A. H. A. and Mohanna, S., "A New Robust Audio Signal
Steganography Based On Multiple Chaotic Algorithms With Edge Detection Technique." Advances in Computer Science: an International Journal, Vol. 3, Issue 2, No. 8, 2014.
Banerjee, I., Bhattacharyya, S. and Sanyal, G., "Robust image
steganography with pixel factor mapping (PFM) technique." Computing for Sustainable Global Development (INDIACom), International Conference on 5-7 March. pp 692-698, 2014.
Bandyopadhyay S. K., Banik, B. G., “LSB Modification and Phase Encoding
Technique of Audio Steganography Revisited ”. International Journal of Advanced Research in Computer and Communication Engineering. Vol. 1, Issue 4, June 2012.
Bender W., Gruhl D., Morimoto N. and Lu A., “Techniques for data hiding.”,
IBM systems journal, Vol. 35, No. 3.4, pp 313-336, 1996. Bennett, K., "Linguistic Steganography: Survey, Analysis, and Robustness
Concerns for Hiding Information in Text", Purdue University, CERIAS Tech. Report, 2004.
Bhattacharyya, S., et al.“Audio Steganography Using Mod 4 Method(M4M)”.
Journal of Computing, New York ,USA, Vol. 3, Issue 8, August 2011. Bhattacharyya, S., Banerjee, I. and Sanyal, G. "A novel approach of secure
text based steganography model using word mapping method (WMM)." International Journal of Computer and Information Engineering, Vol. 4, No. 2, pp 96-103, 2010.
Bhowal, K., Sarkar, D., Biswas, S., and Sarkar P. P., “Secured Image
transmission with GA based Audio Steganography”. IEEE annual conference india(INDICON). pp 1-4, 2011.
83
Candes, E., and Romberg, J. Sparsity and incoherence in compressive
sampling. Inverse problems, Vol. 23, Issue 3, pp 969, 2007. Candès, E. J., and Wakin, M. B. An introduction to compressive sampling.
IEEE signal processing magazine, Vol. 25, Issue 2, pp 21-30, 2008. Candès, E. J. and Tao, T., “Near optimal signal recovery from random
Deepak D. et al.,“Efficient Method to Increase Robustness in Audio
Steganography”. International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering. Vol. 1, Issue 6, December 2012.
Delforouzi, A. and Pooyan, M. “Adaptive digital audio steganography based
on integer wavelet transform.” Circuits, Systems & Signal Processing, Vol. 27, Issue 2, pp 247-259, 2006.
Deshmukh, P. R. and Rahangdale, B.,” Hash Based Least Significant Bit
Technique For Video Steganography”, International Journal of
84
Engineering Research and Applications ,ISSN : 2248-9622, Vol. 4, Issue 1( Version 3), pp.44-49, January 2014.
Djebbar, F. et al.,“Comperitive study of digital audio steganography
technique”. EURASIP journal on audio, speech and music processing. 2012
Djebbar, F., Ayad, B., Hassmam, H., and Abed-Meraim, K. “A view on latest
audio steganography techniques”, International Conference on Innovations in Information Technology (IIT), IEEE, 2011.
Donoho, D. L., "Compressed Sensing," IEEE Transactions on Information
Theory, Vol. 52, pp.1289-1306, 2006. Duarte, M. F., Davenport, M. A., Wakin, M. B. and Baraniuk, R. G., 2006,
May. “Sparse signal detection from incoherent projections. In Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings”. 2006 IEEE International Conference on, Vol. 3, pp. 3 2006.
Ercelebi, E. and Batakc L., “Audio watermarking scheme based on
embedding strategy in low frequency components with a binary image.”, Dig. Signal Process., Vol. 19, Issue 2, pp 265-277, 2009.
Ghasemi, E., Shanbehzadeh, J., and Fassihi, N., "High capacity image
steganography using wavelet transform and genetic algorithm."Proceedings of the international multiconference of engineers and computer scientists. Vol. 1. 2011.
Hemalatha, S., Acharya, U. D. and Priya, R. K., "A secure color image
steganography in transform domain." International journal on cryptography and information security, Vol. 3, No. 1, pp 17-24, 2013.
Hemalatha, S., Acharya, D., U., Renuka, A. and Kamath, P., R., "An Integer
Wavelet Transform Based Steganography Technique for Color Images." International Journal of Information & Computation Technology. Vol. 3, Issue 1, pp 13-24, 2013.
Huang, X., Abe, Y., and Echizen I., “Capacity adaptive synchronized
acoustic steganography scheme.” J. Inform. Hiding Multimedia Signal Process., Vol. 1, Issue 2, pp 72-90, 2010.
Hussain, H. S., et al., "A novel hybrid fuzzy-SVM image steganographic
model." Information Technology (ITSim), 2010 International Symposium in. Vol. 1. pp 1-6, 2010.
ITU-T Recommendation P.862.3, "Application guide for objective quality
measurement based on Recommendation P.862, P.862.1 and P.862.2", November 2005.
85
Jagdale, B., N., Bedi, R., K., and Desai, S., "Securing MMS with high performance elliptic curve cryptography." International Journal of Computer Applications. Vol. 7, pp 17-20, 2010.
Jain, M. and Kumar, A., 2016. “RGB channel based decision tree grey-alpha
medical image steganography with RSA cryptosystem”. International Journal of Machine Learning and Cybernetics, pp.1-11, 2016.
Jain, N., Meshram, S. and Dubey, S., 2012. “Image steganography using
LSB and edge–detection technique”. International Journal of Soft Computing and Engineering (IJSCE) ISSN, 223.
Johannes Trithemius, Steganographia, 1499. Juneja, M., Sandhu, P. S. and Walia, E., 2009. “Application of LSB based
steganographic technique for 8-bit color images”. World Academy of Science, Engineering and Technology, Vol. 50, pp 423-425, 2009
Kahn, D., “The Code Breakers, The Story of Secret Writing”, The Macmillan
Company. (ISBN 0-684-83130-9) New York, NY 1967. Kaur, R., Singh, G., & Singh, S., “An Overview Of Data Hiding Technique:-
Steganography”. International Journal Of Data & Network Security, Vol. 1, Issue 2 , pp 34-36, 2012.
Kaur, S., Bansal, S., and Bansal, R. K., "An Efficient Adaptive Data Hiding
Scheme for Image Steganography." Proceedings of the International Congress on Information and Communication Technology. Springer Singapore, 2016.
Kelash, H. M., Wahab, O. F. A., Elshakankiry, O. A. and El-Sayed, H. S.,
2014. Utilization of steganographic techniques in video sequences. International Journal of Computing and Network Technology, Vol. 2, Issue 1, pp 17-24, 2014.
Kiah, M. L. M., et al.“A review of audio based Steganography and digital
watermarking” International journal of the physical sciences, Vol. 6, Issue 16, pp 3837-3850, August, 2011.
Kumar, A., and Sharma, R., "A secure image steganography based on RSA
algorithm and hash-LSB Technique." International Journal of Advanced Research in Computer Science and Software Engineering. Vol. 3, Issue 7, 2013.
Majumder, A., and Changder, S. “A Novel Approach for Text Steganography:
Generating Text Summary Using Reflection Symmetry”. Procedia Technology, Vol. 10, pp 112-120, 2013.
Mane, A., Galsetwar, G. and Jeyakumar, A. “Data hiding technique: Audio
Steganography LSB technique”. International Journal of Engineering Research and Applications. Vol. 2, Issue 3, pp. 1123-1125, 2012.
Mondal, A., and Pujari, S. "A Novel Approach of Image Based
Steganography Using Pseudorandom Sequence Generator Function and DCT Coefficients." International Journal of Computer Network and Information Security, Vol. 7,Issue 3, 2015.
Nag, A., et al. "A novel technique for image steganography based on DWT
and Huffman encoding." International Journal of Computer Science and Security, (IJCSS), Vol. 4, Issue 6, pp 497-610, 2011.
Nagaraj, V., Vijayalakshmi, V., & Zayaraz, G. (2013). “Color Image
Steganography based on Pixel Value Modification Method Using Modulus Function”. IERI Procedia, Vol. 4, pp 17-24.
Nagaraj, V., Vijayalakshmi, V., and Zayaraz, G., "Modulo based Image
Steganography Technique against Statistical and Histogram Analysis." IJCA Special Issue on “Network Security and Cryptography” NSC, Vol. 4, pp 34-39, 2011.
Nithyanandam, P., Ravichandran, T. , Priyadharshini, E., and Santron, N. M.
"A Image Steganography Technique on Spatial Domain Using Matrix and LSB Embedding based on Huffman Encoding." i-Manager's Journal on Future Engineering and Technology, Vol. 6, No. 3, pp 25 2011.
Nithyanandam, P., Ravichandran, T., Priyadharshini, E., and Santron, N. M.
"An image steganography for colour images using lossless compression technique." International Journal of Computational Science and Engineering, Vol. 7, No. 3, pp 194-205 2012.
Patil, B. A. and Vrishali, A. C., “Review of an improved Audio Steganography
Technique over LSB through Random Based Approach”. IOSR journal of computer engineerin. Vol. 9, Issue 1, 2013.
Pawar, S. S., Kakde, V., "REVIEW ON STEGANOGRAPHY FOR HIDING
DATA." International Journal of Computer Science and Mobile Computing, Vol. 3, Issue. 4, pp.225 – 229, 2014.
Perceptual Evaluation of Speech Quality (PESQ), an Objective Method for
End-To-End Speech Quality Assessment of Narrowband Telephone Networks and Speech Codecs, February 2001.
Poonia, S., Nokhwal, M., and Shankar, A., "A secure image based
steganography and cryptography with watermarking." International Journal of Emerging Science and Engineering (IJESE), Vol.1, Issue 8, 2013.
Pradhan, A., et al., "Performance evaluation parameters of image
steganography techniques." Research Advances in Integrated Navigation Systems (RAINS), International Conference on. IEEE, 2016.
87
Pujari, S., and Mukhopadhyay, S., "An Image based Steganography
Scheme Implying Pseudo-Random Mapping of Text Segments to Logical Region of Cover Image using a New Block Mapping Function and Randomization Technique." International Journal of Computer Applications, Vol. 50, Issue 2, 2012.
Qin, C., Chang, C.C., Huang, Y.H. and Liao, L.T., 2013. An inpainting-
assisted reversible steganographic scheme using a histogram shifting mechanism. IEEE Transactions on Circuits and Systems for Video Technology, Vol.23, Issue 7, pp 1109-1118.
Rabah, K., "Steganography-The Art of Hiding Data", Information Technology
Journal, Vol. 3, Issue 3, 2004, pp. 245-269. Rajput, V., Tiwari, S. K., and Gupta, R. "An enhanced image security using
improved RSA cryptography and spatial orientation tree compression method." Signal Processing, Communication, Power and Embedded System (SCOPES), 2016 International Conference on. IEEE, 2016.
Rameshkumar, P., Monisha, M. and Santhi, B., 2014. “Enhancement of
information hiding in audio signals with efficient LSB based methods”. Indian Journal of Science and Technology, Vol.7 No. S5, pp.80-85.
Reddy, H. S., Manjunatha, N., Sathisha, A. K., and Raja, K. B., "Secure
steganography using hybrid domain technique." In Computing communication & networking technologies (ICCCNT), 2012 third international conference on, pp. 1-11. IEEE, 2012.
Reddy, H. S. and Raja, K. B.,"Wavelet based secure steganography with
scrambled payload." International Journal of Innovative Technology and Exploring Engineering (IJITEE), Vol. 1, Issue 2, pp 121-9, 2012.
Reddy, H. S. and Raja, K. B. "Hybrid Domain based Steganography using
BPS, LSB and IWT." International Journal of Computer Applications Vol. 54, Issue 3, 2012.
Roy, S., and Venkateswaran P., "A text based steganography technique with
indian root." Procedia Technology, Vol. 10, pp 167-171, 2013. Salah S. A., et al., “Watermarking in WAV files Based on Phase Coding”.
Saraswathi, V., and Mrs Kingslin, S., "Different Approaches to Text
Steganography: A Comparison." Int. J. Res. Manag. Technol 9359, No.11, pp 124-127, 2014.
88
Seyyedi, S. A., Sadau, V., and Ivanov, N., "A Secure Steganography Method Based on Integer Lifting Wavelet Transform." IJ Network Security Vol. 18, Issue 1, pp 124-132, 2016.
Shahadi, H. I. and Jidin, R., “High capacity and in audibility Audio
steganography scheme.” Proceeding of 7th International Conference on Information Assurance and Security (IAS), pp: 104-109, 2011.
Shahadi, H., I., Jidin, R. and Way, W., H., 2014. A novel and high capacity
audio steganography algorithm based on adaptive data embedding positions. Research Journal of Applied Sciences, Engineering and Technology, Vol. 7, Issue 11, pp 2311-2323, 2014.
Shahreza, S., and Shalmani, M., “Adaptive wavelet domain audio
steganography with high capacity and low error rate.” Proceeding of IEEE International Conference on Information and Emerging Technologies (ICIET), pp: 25-29, 2007.
Shahreza, S. and Shalmani, M., “High capacity error free wavelet domain
speech steganography.” Proceeding of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp: 1729-1732, 2008.
Shirali-Shahreza, M. H., Shirali-Shahreza, M., “Arabic/Persian Text
Steganography Utilizing Similar Letters With Different Codes,” The Arabian Journal for Science and Engineering, Vol. 35, Number 1B, April 2010
Shirali-Shahreza, M. H., Shirali-Shahreza, M.,"A New Approach to
Persian/Arabic Text Steganography", Proceedings of the 5th IEEE/ACIS International Conference on Computer and Information Science (ICIS 2006), Honolulu, HI, USA, 10-12 July, 2006, pp. 310-315.
Singh, P. K., Aggrawal, R. K., “ Enhancement of LSB based Steganography
for Hiding Image in Audio.” International journal on Computer Science and Engineering, Vol. 2, No. 5, pp 1652-P1658, (2010).
Singh, A. K., Dave, M., and Mohan, A., "Hybrid technique for robust and
imperceptible image watermarking in DWT–DCT–SVD domain." National Academy Science Letters Vol. 37, No. 4, pp 351-358, 2014.
Singh, S., “Hiding Image to Video” International Journal of engineering
science & technology Vol. 2, Issue 12 , pp 6999-7003, 2010. Sharma, S., and Kumari, U., "A High Capacity Data-Hiding Technique Using
Steganography." International Journal of Emerging Trends & Technology in Computer Science (IJETTCS), Vol. 2, Issue 3, 2013.
89
Shirali-Shahreza, M., & Shirali-Shahreza, M., H. (2007, November). Text steganography in SMS. In Convergence Information Technology, 2007. International Conference on (pp. 2260-2265). IEEE.
Shirali-Shahreza, M. (2008). A New Persian/Arabic Text Steganography
Using “La” Word. In Advances in Computer and Information Sciences and Engineering (pp. 339-342). Springer Netherlands.
The PESQ Algorithm as the Solution for Speech Quality Evaluation on 2.5G
and 3G Networks, 2009 The PESQ Algorithm as the Solution for Speech Quality Evaluation on 2.5G
and 3G Networks, 2009. Vetterli, M., Marziliano, P., and Blu, T., “Sampling Signals with Finite Rate of
Innovation,” IEEE Transaction on Signal Process, Vol. 50, no. 6, pp.1417–1428, 2002.
Yadav, P., Mishra, N., and Sharma, S., "A secure video steganography with
encryption based on LSB technique." Computational Intelligence and Computing Research (ICCIC), International Conference on. IEEE, pp 1-5, 2013.
Zaidan, A. A., Zaidan, B. B., Taqa, Y. A., Sami M. K., Alam, G. M. and Jalab
A. H., "Novel multi-cover steganography using remote sensing image and general recursion neural cryptosystem." International Journal of Physical Sciences, Vol. 5, No. 11: pp 1776-1786. 2010.
Zhang, H. J., Tang, H. J. “A Novel Image Steganography Algorithm Against
Statistical Analysis”, Proceedings of The Sixth International Conference on Machine Learning and Cybernetics, IEEE, Hong Kong, Vol. 7, pp. 3884-3888, 2007.
Zamani, M., Manaf, A. A., Ahmad, R. B., Zeki, A. M. and Abdullah, S., "A
genetic-algorithm-based approach for audio steganography." World Academy of Science, Engineering and Techqnology, Vol 54, 360-363, 2009.