Steganography is concerned with hiding information in redundant space of any
unremarkable cover medium and keeps the secret information undetectable without
destroying the cover medium integrity (Gutub and Fattani,
As a technique of protecting the secret information, steganography does not
focus on limiting or controlling the access to information, but protecting the
hidden information not to be detected or destroyed. Instead of how the encryption
protects data, steganography hides the very existence of the information (Provos
and Honeyman, 2003). Capacity, security and robustness, which are the three
main factors that influence steganography, are contending with each other. Capacity
regards to the number of secret information bits could be hidden in the cover
medium. Security refers to the possibilities to figuring out the hidden informationby
the eavesdropper. Robustness relates to the amount of modification the stego-medium
can withstand before the adversary destroys the hidden information (Chen
and Wornell, 2001). It should be seeking for the appropriate balance between
the three aspects according to the specific requirement.
In the past decade, a number of steganographic methods have been proposed,
however, the majority of them use cover medium such as pictures (Chandramouli
and Memon, 2001), video clips (Doerr and Dugelay, 2003)
and sounds (Gopalan, 2003). In spite of that, text documents
are the most prevalent and indispensable form of information nowadays and always
be used as a cover medium. For instance, Margaret Thatcher, former British Prime
Minister used to put certain number of white-spaces in documents associated
with each minister to identify the owner of the document in order to prevent
disclosure of government documents by the press (Moerland,
Most text steganography are based on the formats of TXT, MS Word, PDF, PPT, etc. However, few works have studied on steganographic methods of embedding data in MS Excel document. This paper considers the problem of steganography in MS Excel document for the purpose of hiding additional information in it.
Many works have been done on text steganography. Since mostly Excel documents
have little natural language text, the linguistic steganographic schemes (Murphy
and Vogel, 2007; Bennett, 2004) are inefficient. Hence,
we do not consider the linguistic steganography in this study. The list of several
different methods, which could be use in Excel documents, are as follow.
Open space method (Brassil et al., 1999):
Hiding information is done through embedding information by utilizing the space
characters in a plain text document in this method. These space characters are
placed at the end of paragraph lines or between the words. Open space method
could be used more widely for the Excel document. Most cells in a MS Excel document
are empty so that these can be used for embedding space without changing the
Character features methods (Rabah, 2004): Some
features of characters in a text are changed to embed information in these methods.
For example, it can change the font color slightly to embed the information
in the text. Steganography based on character features can hold a large quantity
of secret information without making normal readers aware of the existence of
such information in the text. Most steganography methods base on character feature
are suitable for Excel document, such as changing the text size, color, alignment,
etc. However, modifying a single character without affecting the whole string
in the cell is notpossible in an Excel document. Therefore, the steganographic
effect in an Excel document is not as good as that in the Word document.
Line and word shifting methods (Low et al., 1995):
Shifting text lines vertically and shifting words horizontally could embed information
in these methods. Security of this method depends on the availability of varying
the distances between words and lines to puzzle intruders. For example, some
lines are shifted 1/300 inch up or down in the text and information are hidden
by creating a hidden unique shape of the text. But these methods can not be
directly applied in Excel document. However, we could change the height of rows
or the width of columns slightly to embed information.
Abbreviation methods (Bender et al., 1996):
Another method for hiding information is using of abbreviations. Usually, there
are few words could be translated to abbreviations in a text, so very little
information can be hidden in the text, as in the case of Excel document.
File structure methods (Cantrell and Dampier, 2004;
Castiglionea et al., 2007): Many documents
contain readily available spaces that can be used inside their file structures.
In these methods, some unused space is used to embed information. Meta-data
is an example; it is ingrained in file structures but not visible to the user
without special tools. Some files also have unused space. In these spaces, bits
can be overwritten without any adverse or obvious effect on the file. These
spaces create an opportunity to hide information. In Excel file, there are 420
bytes continuous block below the header where can embed data easily.
Here, steganography scheme in MS Excel document based on text-rotation algorithm is presented. The proposed steganography scheme described below composes of the embedding and extracting process. This study proposes a novel steganographic method by slightly rotating the angle of the text in the cells according to the value of the corresponding secret bit.
The general algorithm: The proposed hiding Algorithm consists of three steps. In the first step, encoder encrypt the secret information by some encrypt methods such as RC2, RC4, DES etc. In the second step, the encoder first scans the Excel document to obtain the text in a cell, then judges whether the currently processing text is embeddable. Here, embeddable means that the length of currently processing text is less than a threshold, and it will be discussed in the next section. In the last step, the text-rotation algorithm is used for embedding the data, and outputs the stego-document. Figure 1 depicts the embedding process of the proposed algorithm. The basic idea in the third step is to make sure attackers cannotdistinguish the difference between the stego-document and the original document with the naked eye.
Text rotates in MS excel document: In general, HVS (Human Visual System) is hard to detect the minor changes in the texts angle. However, being more than 1° rotation, the changes in text is obvious, therefore only 1° rotation is considered. Table 1 shows the comparison between original text and 1° rotated text.
However, the Fig. 1 shows that the stego-text gets worse with the increase of texts length in a cell. After 1° rotation, the appearance of the first three characters almost have no changed and every five characters compose a group in the behind characters. Every character in a group is in the same horizontal line. Figure 2 shows that each group will raise or fall of a point comparing to the former group after rotation.
|| The embedding process
||Comparison between original text and 1° rotated text.
(The font style is MS reference sans serif and the font size is 14)
||Comparison between 1°rotated numeric text and-1° rotated
numeric text. (The font style is MS Reference Sans Serif and size is 14)
|| Groups relative height after rotation
Since the data type mostly is numeric in the Excel documents, numerical data
in the cells should be given more attention. As it was shown in Table
2, while rotating the numeric data by one degree, the alignment of the data
will be left-aligned into a right-aligned. Furthermore, because the default
alignment of a numeric data is left-aligned, the changes of appearance are magnificent.
Nevertheless, if rotating the numeric data by-1°, its alignment remain left-aligned.
The data in cells should be treated separately, 1° rotation for the text
data and-1° rotation for the numeric ones.
Table 1 and 2 have shown that when the texts length in a cell is less than four, the rotation is hard to detect.
Embedding algorithm based on text-rotationt technique: The proposed
algorithm consists of three steps. The first step is finding out the non-empty
cell and judge whether its length less than threshold p. The larger the threshold
p, the lower the imperceptibility; the imperceptibility is more important than
embedding rate in the steganography process, therefore, the threshold p can
not be too larger. Threshold p is suggested to be 4 in order to get the good
imperceptibility. The second step is to determine the contents of the cell are
numeric data or character data. The final step is to changes the angle of cell
contents according to the type of data. The details of the message embedding
process are presented in the algorithm:
||Embedding process based on text-rotation technique
Data extraction process: The extracting process essentially reverses
the embedding process. First, the extractor check the cell if its embeddable.
Then, exam the angle and data type from the selected cell to get the secret
data. Finally, decrypt the encrypted data and get the secret information.
RESULTS AND DISSCUSION
The proposed methods were implemented using Microsoft Excel 2003 and Microsoft Visual C++6.0 software and running on the Pentium Dual, 2.2 GHz CPU, and 1 GB RAM hardware platform.
In experiments, a student transcript was used as a cover which was shown in Table 3 and the font style of the text was Song and font size was 12. Assuming the secret message bits to be embedded is 10110011000010110010110000110001110000. RC4 was used to encrypt the secret message.
Table 3 compared an original document with the one in which the secret message had been embedded by algorithm1. Note that the difference between the stego-text and the cover text is virtually unnoticeable. In this experiment, the threshold p was 4.
Finally, several configurations for the threshold p of the algorithm1 were used. While changing the threshold p, the embedding rate mostly changed, and the imperceptibility would be changed as well. Figure 3 showed that the embedding rate would increase while the threshold p rose. But setting the threshold too large often leads to poor imperceptibility. Table 4 showed the effect of using large threshold p and the hidden effect was relatively poor.
The embedding rate was closely related to the cover document. Short length
text in cells in Excel could result in a high embedding rate. In extreme cases,
the embedding rate would close to 100%, while still maintaining good concealment
of the stego-document.
|| Comparison between cover document and stego-document
|| Effect of setting threshold p to 11, bits 11010011100101100001
|| Embedding rate wiyh difference threshold p
The contribution of this study is to develop a novel algorithm for embedding data in an Excel document, and the proposal of it is to use text-rotation technique for steganography. It benefits from the feature that HVS is hard to detect the minor changes in the texts angle. Since many cells in Excel document are very short in length, the embedding rate would be very high.
Although various data hiding methods which based on the text documents are focus on the formats of TXT, MS Word, PDF, PPT, etc., few have studied methods of embedding data in MS Excel document. Compared with other text formats, the expression form of Excel document data are quite different. Therefore, steganography based on Excel document deserves to be further investigations.
Many text-based steganographic methods can be used in Excel document, the proposed technique, which is characterized by high capacity and security, will be complementary to the steganographic methods for Excel document.
This study was partially or fully supported by National Basic Research Program of China (973 Program) under Grant No. 2009CB326202 (2009.4-2011.8, Hunan University) and 2010CB334706 (2010.4-2013.3, Hunan University); Key Program of National Natural Science Foundation of China under Grant No. 60736016 (2008.1-2011.12, Hunan University); National Natural Science Foundation of China under Grant No. 60973128 (2010.1-2012.12, Hunan University), 60973113 (2010.1-2012.12, Hunan University), 61073191(2011.1-2013.12, Hunan University) and 61073196 (2011.1-2013.12, Hunan University); Guangdong Province Natural Science Foundation under Grant No. 8151063101000040 (2008.5-2011.4, South China Normal University); the 3rd Guangdong Province 211 program for key subject development (2008.1-2012.12, South China Normal University).