International Journal of
Intelligent Systems and Applications in Engineering
ISSN:2147-67992147-6799
www.ijisae.org
Original Research Paper
New Approach in E-mail Based Text Steganography
Kemal Tutuncu *1, Abdikarim Abi Hassan
Accepted 26th May 2015
Abstract: In this study combination of lossless compression techniques and Vigenere cipher was used in the e-mail based text
steganography. It makes use of email addresses to be the keys to embed/to extract the secret message into/from the email text (cover
text). After selecting the cover text that has highest repetition pattern regarding to the secret message the distance matrix was formed.
The members of distance matrix were compressed by following lossless compression algorithms as in written sequence; Run Length
Encoding (RLE) + Burrows Wheeler Transform (BWT) + Move to Front (MTF) + Run Length Encoding (RLE) + Arithmetic Encoding
(ARI). Later on Latin Square was used to form stego key 1and then Vigenere cipher was used to increase complexity of extracting stego
key 1. Final step was to choose e-mail addresses by using stego key 1 (K1) and stego key 2 (K2) to embed secret message into forward email platform. The experimental results showed that proposed method has reasonable performance in terms of capacity and also higher
security in terms of complexity.
Keywords: Text steganography, Latin square, Vigenere chipher, Stego key, BWT, MTF, RLE, ARI
1. Introduction
The case of safety and security of information and especially
secret communications has led to the introduction of several
methods for secret message. Among these methods
steganography is a rather new method. Steganography, the art of
information hiding, has been used around for thousands of years,
with the earliest examples coming from as early as 440 B.C.
Steganography is the traditional method used in information
hiding. The use of systems like cryptography it makes the output
of these methods very conspicuous. The main reason behind this
is that the systems make messages unreadable to anyone except
the intended recipient and does not hid its existence. The design
of steganographic approaches is aimed at hiding the message or
(data) and ensuring that its existence is hidden from observers.
Modern steganography can be applied to text, images, audio and
video as shown in Fig.1. However, text steganography, has
received less interest recent years, primarily due to the difficulty
in finding redundant bits in text files and the lower capacity to
hide information than the other mediums. This should not be the
issue, as text Steganography has many advantages over the other
mediums which make it model for effective Steganography.
One advantage of text Steganography over images and audio is
that while they are both susceptible to compression due to their
use of redundant data, this is not an issue with text Steganography
as even though text contains redundancy, it cannot be removed or
compressed. In addition the advantage to prefer text
Steganography over images and audio is its smaller memory
occupation and simpler communication. Text is also still one of
the major forms of communication in the world, both in digital
and printed form, and there are not many people who do not have
access to text.
Text Steganography can be classified into three main groups:
Structural, random and statistical generational, and finally
linguistic. The main characteristics of structural text
Steganography is that it modifies the physical form of the text, for
instance through the appending white spaces and linespaces. On
the other hand, random and statistical generation entails
providing the cover text. This can be in a random manner or
_______________________________________________________________________________________________________________________________________________________________
1
Faculty of Technology, Selcuk University Campus,Konya,Turkey
* Corresponding Author: Email: ktutuncu@selcuk.edu.tr
depending on specified input. While linguistic Steganography
uses contents of natural language, for example verbs, nouns,
adjectives and so on. Linguistic Steganography are classified into
two: syntactic and semantic Syntactic text Steganography deals
with changing the format of the text without considerable
alteration to the meaning or tone.
Figure 1. Categories of steganography
In this study, an email based data hiding method which makes use
of combinatorial lossless compression to increase the hiding
capacity and Vigenere cipher to increase the complexity was
proposed. Run Length Encoding (RLE) + Burrows Wheeler
Transform (BWT) + Move to Forward (MTF) + Run Length
Encoding + Arithmetic Encoding (ARI) algorithms were used to
have better compression ratio. Additionally Vigenere cipher was
used to increase complexity of the system. The secret data was
embedded into the email ids of forward mail platform. The most
appropriate cover text is chosen from the text base by using
distance matrix. After the stego cover determination we make use
of distance matrix to find out email ids of a forward mail
platform. Then e-mails were chosen from the previously arranged
email address list. This email address list is used as a global stego
key that is shared between both the sender and the receiver in
advance. Capacity metric of the proposed method was evaluated
by using obtained e-mail addresses.
The rest of the paper is structured as follows. Section II provides
the related works. Section III and Section IV describe the
proposed method and the experimental result. Finally, the
International Journal of Intelligent Systems and Applications in Engineering
IJISAE, 2015, 3(2), 54-56 | 54
conclusion is drawn in Section V.
2. Related Works
In [1], [2], Satir and Isik used email to be the cover of the secret
data. Sending same e-mail to many recipients were the key idea
of their studies. There may not be a proper method to validate
whether the email address of each recipient is valid but sending
an email to the respective address. In those papers [1], [2], email
addresses are used to hide the stego keys, regardless they are
valid addresses or not [3]. In [3] the authors proposed
improvement of the study [2], such as the randomity of the
generated email addresses. They claimed that random generated
email addresses may increase the security level. In [4], the
authors proposed combination of BWT + MTF + LZW coding
algorithm to increase the hiding capacity and adding some
random characters just before the ‘@’ symbol of email ids to
increase the randomness.
In a chat application, Wang et al. [5] employ emoticons
(emotional icons) to hide a secret message. This is intended to
improve the performance of other chat-based steganography. The
proposed method is developed based on the assumption that the
use of emoticons in chatting is high, where many recent
applications using it. It is claimed that this method is able to raise
up the capacity of the secret data to be embedded as well as to
provide an easy to use application. This method requires both
sender and receiver have an exactly same emoticon table
containing some emoticon sets. In this case, the order of the
emoticon affects the meaning of the secret message. Therefore,
there must be a synchronization process before the
communication begins. Practically, this method requires the users
to input many emoticons in order to send the secret data. This
may have an effect on the user convenience [4]. Wayner [6,7]
discussed an important method using mimic functions. It applies
the inverse of Huffman Code having employed the randomly
distributed bits of input stream on itself. Other researches have
been carried out which employ varied communication medium,
such as in [8], [9]. Hiding the secret in the text is also
investigated by Desoky [10]. He proposes Listega, a method to
hide data which exploits the popularity of list of items, such as
song, food, drink and car. Another text steganography given by
Maher [11] which is popular as TEXTO is built to transform
uuencoded or PGP ASCII armored ASCII data into English
language sentences. It converts the secret data into English
words. To extend the work of [12], another important method is
synonyms-based approach [13-15]. Unlike [11], the method uses
legitimate words and sentences having appropriate preciseness.
Thus, the visual attack will have wee significance on these types
of methods.
3. Proposed Method
3.1. Embedding Phase
Let, S: secret message T: A text base of cover texts in which the
secret message will be embedded. K1: set of email addresses
having four characters before ‘@’ symbol. It is shared between
the sender and the receiver. A: set of the second parts of email
addresses such as outlook.com, gmail.com, etc.
Step1. Construct difference matrix D in order to select most
suitable or relevant text from the T. the D is calculated by having
difference of index of last matched symbol of S with current
index of match symbol in T iteratively except for the first symbol
of S as in case of first symbol, the last symbol index is 0 as in
[1,2,3,4]. Step2. Calculate vectors R and E using following
equations: R=D mod 26 E=D/26
55 | IJISAE, 2015, 3(2), 54-56
Step3. Estimate the maximum dual pattern repetition in R and
store in a column matrix P. Now, select the largest row of P and
denote it as Pmax. The corresponding rows of R and E vectors
are also selected and put in R* and E* vectors. The text from T
corresponding to Pmax is also chosen and put in T* as it is the
most suitable text.
Step4. Apply Run Length Encoding (RLE) + Burrows Wheeler
Transform (BWT) + Move to Forward (MTF) + Run Length
Encoding + Arithmetic Encoding (AE) in the written order for
compressing elements of R*
Step5. Represent each element of R in binary form and
concatenate them in order to obtain bit stream.
Step6. The bit stream is partitioned into groups of 12 bits, in each
group, the first 9 bits are called G1, next 3 bits are called G2. The
quotient and remainder of decimal representation of G1 with
respect to 26 are known as x and y respectively and decimal
representation of G2 is known as z.
Step7. Choose two characters as K1 by employing Latin square
on x and y. Select extensions of email addresses using z from A.
Step8. Use Vigenere cipher to alter two characters obtained from
Latin Square. Thus these two new characters will be used to
choose e-mail address. Choose email addresses that starts with
modified K1 before @ and ends with z after @.
Step9. Modify resultant email addresses in order to complete
construction of K2 set by using E*. Combine modified K1 and
K2 (obtained from z) to form e-mail addresses.
3.2. Extracting Phase
Before starting extraction phase both the sender and receiver
must have email address list, three bits corresponding of email
extension, Vigenere cipher key, RLE, MFT, BWT and ARI tables
in advance.
Step1. Get the stego-cover. Extract numeric elements of K2
before ‘@’ symbol to construct the vector E. If there is not any
numeric element then E will be 0.
Step2. Extract first two elements from K2 to obtain Vigenere
values of x and y by employing Vigenere Cipher.
Step 3. Extract the values of x and y by employing Latin Square
Also extract email address extension to obtain z.
Step 4. Calculate G1 and G2 for each group of 12 bits by using
the following equations: G1 = (x • 26+y) G2 = (z)
Step4. Concatenate G1 and G2 in same order to obtain
compressed bit stream.
Step5. Decompress the bit stream using Arithmetic Encoding
(AE) + Run Length Encoding (RLE) + Move to Forward (MTF)
+ Burrows Wheeler Transform (BWT) + Run Length decoding to
obtain original R*
Step7. Estimate original difference D using R* and E as follows:
D = R + (26 • E)
Step8. By using elements of D, extract the elements of S through
T*, in the stego cover.
4. Results
The proposed method was prepared by programming in
MATLAB. The computer used for running the program was Intel
Pentium7, with 8 GB RAM. Later on same computer was used to
analyse performance of proposed method or program. The
performances of the text steganography methods can be analysed
by capacity ratio or complexity. In this study both parameters are
investigated. In order to calculate the capacity ratio secret data
whose length is 196 characters is used. The secret message is as
follows:
‘behind using a cover text is to hide the presence of secret
International Journal of Intelligent Systems and Applications in Engineering
messages the presence of embedded messages in the resulting
stego text cannot be easily discovered by anyone except the
intended recipient’. Cover data set was ranging from 700
characters to 4000 characters. 35 cover data set was used. It has
been seen that the success of the embedding is directly related
with length of the secret message and cover text. The smaller size
the cover text the less successful the embedding is. Two factors
are important here; Firstly the length of cover to possibly include
all the characters of the secret message due to its length and
secondly the match between the content of the cover message
with the frequencies off English alphabet. If a cover text is
formed according to the frequency of English alphabet it will
cause better capacity ratio for text steganography.
Table 1. Applying text message (200 char) to different cover text
Cover
Cover
Length
(char)
Result
Number of e-mail
addresses
CT1
300
Fail
-
CT2
800
Fail
-
CT3
1500
Fail
-
CT4
3993
Success
1957
As can be seen from Table 1. CT1, CT2 and CT3 cover texts
resulted in not hiding secret message. No e-mail address is
produced to hide secret text. CT4 successfully hided the secret
text. Here the key point is how much repetition occurs when we
created distance matrix. If we have much more repeated number
in distance matrix then the compression ratio of the combined
lossless compression techniques will increase. This will cause
increase in capacity ratio, thus decrease number of e-mail
addresses. From this point of view it is quite important to choose
cover text to increase the capacity ratio in text steganography.
Briefly explaining not only the length of the cover data
determines the successfulness of the embedding process, but also
variation of the characters. Longer the cover data actually
increases the possibility of providing more varied characters.
Therefore, successfulness of the embedding process depends on
the appropriate selection of the cover data [3].
Complexity is one of the important aspects of the steganography.
It is directly related with extraction of hidden information by an
adversary in an easy way or not. In this study compression
algorithms and two combinatory based coding (Latin square and
Vigenere cipher) were used to increase the complexity.
Complexity analysis of the proposed system will be explained.
Before explanation one must keep in mind that global key and the
key of Vigenere cipher were shared between sender and receiver
in advance. Thus, if any adversary would like to break the system
he will not have these two keys. The step that an adversary will
follow to break the system as follows:
1. Adversary must extract compressed R*. In order to implement
this action he/she has to obtain G1from x, y and G2 from z. At his
point x and y can only be obtained by using the Vigenere cipher
that consists of two characters. Finding correct combination of
these two characters will require 26*26 combinations. Finding
correct combination will let adversary to find the Latin Square
corresponding of x and y. Afterwards x and y can be obtained
from Latin Square. The z can be obtained by the extension of email address. Adversary has to calculate 8 combination per each
e-mail. Thus, if the number of e-mail address is 8 then 8N
combination must be calculated. Later on each z is expressed in
binary form and then combined with x and y to obtain G1 and
G2. In order to find out correct bit stream R*, 8 N*262
combination must be formed.
2. R is obtained by solving R*. As it is known that D = R + (26 •
E) and each e-mail address is checked to see whether it has
number before @ sign or not. In case of including numbers
before @ sign these numbers are used for obtaining members of
E. These numbers can belong to e-mail address naturally or added
to e-mail address to hide members of E during embedding phase.
If we express number of numbers in an e-mail address as m then
the adversary will implement 2m combination to extract the
members of E.
3. Obtained each R and E must be tested to each other to see the
correct hidden message.
Thus complexity of the proposed system is
𝑚
𝐶𝑜𝑚𝑝_𝑝𝑟𝑜𝑝𝑜𝑠𝑒𝑑_𝑠𝑦𝑠𝑡𝑒𝑚 = 262 8𝑁 × ∏𝑁
1 2
Complexity of the system is calculated without regarding the
compression algorithms that are used to have combination of
RLE + BWT + MTF + RLE + AE. In the study of [2] the authors
didn’t make use of complexity of obtaining decompressed text. In
order to make comparison with this study only the complexities
contribution of k1, k2 and Vigenere cipher were used. In [2]
𝑚
complexity was followed as 8𝑁 × ∏𝑁
1 2 . Thus proposed method
increased the complexity as the amount of 262 .
5. Conclusion
Proposed method is extension of text-based steganography that
makes use of email environment to hide the secret text. Capacity
ratio and complexity of the proposed method are improvements
of the previous e-mail environment based text steganography
methods [1-4] in terms of complexity. Reasonable capacity was
obtained. By using combination of Run Length Encoding (RLE)
+ Burrows Wheeler Transform (BWT) + Move to Front (MTF) +
Run Length Encoding (RLE) + Arithmetic Encoding (ARI)
algorithms the size of the secret message was reduced. Vigenere
cipher added extra complexity or security to the system for
obtaining stego key (K1). The study also showed that the length
and the content (variation) of the cover text directly related with
the success of the embedding process. Short cover text with not
well distributed alphabet content (doesn’t match with the
frequency of the English alphabet) will result failure in
embedding process.
Other combination of lossless compression techniques can be
tried for increasing the capacity ratio of the email environment
based steganography. Additionally some other techniques that put
randomness to the email addresses can be used to increase the
complexity of the related system.
References
[1]
[2]
[3]
[4]
International Journal of Intelligent Systems and Applications in Engineering
E. Satir and H. Isik, A compression-based text
steganography method, The Journal of Systems and
Software Science Direct, vol. 85, issue 10, pp. 2385-2394,
2012.
E. Satir and H. Isik, A Huffman Compression based Text
Steganography Method, Multimedia Tools Appl, September
2012.
Tohari Ahmad, Melvin S. Z. Marbun, Hudan Studiawan,
Waskitho Wibisono, and Royyana M. Ijtihadie, A Novel
Random Email-Based Steganography International Journal
of e-Education, e-Business, e-Management and e-Learning,
Vol. 4, No. 2, April 2014
Rajeev Kumar , Satish Chand , and Samayveer Singh, An
IJISAE, 2015, 3(2), 54–56 | 56
Email based high capacity text steganography scheme using
combinatorial compression, 25th International ConferenceConfluence The Next Generation Information Technology
Summit (Confluence), 25-26 September 2014, India
[5] Z. Wang, T. Kieu, C. Chang, and M. Li, Emoticon-based
text steganography, in Proc. 2009 Asia-Pacific Conference
on Computational, Wuhan, China, 2009.
[6] P. Wayner, Mimic Functions, Cryptologia vol. 16(3), pp.
193-214, 1992.
[7] P. Wayner, Disappearing Cryptography, AP Professional,
Chestnut Hill, MA (1996).
[8] L. Por, K. Wong, and K. Chee, UniSpaCh: a text based data
hiding method using unicode space characters, Journal of
Systems and Software, vol. 85, issue 5, pp. 1075-1082,
2012.
[9] M. Topkara, U. Topkara, and M. Atallah, Information
hiding through errors: a confusing approach, in Proc. SPIE
International Conference on Security,Steganography, and
Watermarking of Multimedia Contents, San Jose, CA, 2007
[10] A. Desoky, Listega: list-based steganography methodology,
57 | IJISAE, 2015, 3(2), 54-56
[11]
[12]
[13]
[14]
[15]
International Journal of Information Security, vol. 8, no. 4,
pp. 247-261, 2009.
K. Maher, TEXTO. 1995.
ftp://ftp.funet.fi/pub/crypt/steganography/texto.tar.gz.
K. Winstein, “Lexical steganography through adaptive
modulation of the word choice hash”, Secondary education
at the Illinois Mathematics and Science Academy, January
1999.
H. Nakagawa, K. Sanpei, T. Matsumoto, T. Kashiwagi, S.
Kawaguchi, K. Makino and I. Murase, “Meaning
Preserving Information Hiding _Japanese text Case,” IPSJ
Journal, Vol.42, No.9, pp. 2339 - 2350, 2001. (In Japanese)
B. Murphy and C. Vogel, The syntax of concealment:
reliable methods for plain text information hiding, In
Proceedings of the SPIE Conference on Security,
Steganography, and Watermarking of Multimedia Contents,
San Jose, CA, vol. 65(05), 2007.
A. Desoky, Listega: List-Based Steganography
Methodology, International Journal of Information
Security, Springer-Verlag, vol. 8, pp. 247-261, April 2009.
International Journal of Intelligent Systems and Applications in Engineering