202 IJCSNS International Journal of Computer Science and Network Security, VOL.10 No.
7, July 2010
A Steganography Method Based on Hiding secrete data in
MPEG/Audio Layer III*
Dr Osamah Abdulgader Al-rababah,
College of Alkamel - Computer Department king Abdulaziz university-
110Alkamel 21931 Jedda, Saudia Arabia
In [7] Presented spread spectrum audio data hiding. This
Abstract method is different from the traditional way in that it is
Steganography is a method for hidden information in such a way based on dealing with the original sound by sub-and phase
that can only be detected by its intended recipient. shifting. The proposed scheme can provide the robust
Steganography in audio becomes a challenging discipline, since aerial data transmission compared to the traditional
the Human Auditory System (HAS) is highly sensitive. One of schemes.
the main obstacles of the data hiding in audio is to develop a
In [8] Proposed a method by compressing MP3 and
system which has the quality to include a big amount of data and
without affectively the quality of sound . modifying spectrum values of audio layers to embed secret
This paper proposes a novel information-hiding method to hide information into audios especially music files in [Link]
more information into audios media file (MP3). The bits of method can extract secret information without original
informatione will be hidden between frames (BF) in MP3 file. audio and it has characteristics that information hiding
A In the experimental results, we hide more characters into technique must be responsible for prerequisites.
audios and extract them In [9] Proposed data hiding algorithm based on Piecewise
Correctly. The audios with secret information are indiscernible to Linear Haar (PLHaar) transform. The PLHaar transform
human ears. maps an n-bit integer to another n-bit integer which don’t
Key words: guarantee under flow or over flow to occur. Secret
Steganography, MP3, Hiding
information is embedded into the Least Significant Bits
(LSB’s) of the PLHaar transform coefficients. Although
Introduction the proposed algorithm has a slight decline in the quality
of the stego audio compared with the algorithm based on
secret digital music MP3 files, are an important and Haar transform, it showed an ability of retrieve embedded
popular audio compression standard in the Internet. MP3 secret information correctly without any additional
uses the destructive compression technologies to achieve operation. The researcher recommended expanding upon
high compression rate and the original file is shrunk to a the PLHaar transform to reversible data hiding for audio.
very small size. [1] In [10-12] proposed a methods of audio steganography
MP3 compression [2, 3], is too time-consuming [4]. To based on modification of least significant bits (LSB).
protect digital media files, researchers propose and In [13] Proposed a new approach under the rubric of
improve many data-hiding algorithms, which are known as Wave-Based Steganography System (WBSS). This system
steganographic algorithms. consists of two techniques: Firstly the wave differential
steganography means covered writing. Steganography can technique (WDT). It is used to decrease the limitation and
be classified into three types: pure steganography, secret complexity of embedding huge amount of data (Text,
key steganography, public key steganography [5]. Image) into wave-cover media. The second technique is
All data-hiding applications require hiding algorithms on used to embed and extract wave into wave-cover media.
the sender side and a detector mechanism or algorithm on In [14] Presented three methods of Steganography for
the receiver side. The hiding hidden message or data in the hiding secret data in audio media file (WAV). The first
can be retrieved by authorized people only. proposed method is used to embed binary sequence with
The most crucial parameters of the data-hiding high data rate by modulating the amplitude of WAV file.
applications are: security, reliability, invisibility, The second and third suggested hiding methods are
complexity, and data-hiding capacity, these parameters are oriented to embed the secret data such that it is capable of
mostly related to each other [6]. surviving against modifications produced by MP3
In recent years various techniques for steganography in compression standard. Statistical and analytical
digital audio with various purposes have been developed investigations are performed to assess the variations which
[7-14]
Manuscript received July 5, 2010
Manuscript revised July 20, 2010
IJCSNS International Journal of Computer Science and Network Security, VOL.10 No.7, July 2010 203
may occur in the WAV audio when it is subjected to MP3 {
compression. If (I*NOB + NOB <= temp count + Offset) then
In this paper, we propose two new steganographic Store NOB in result
methods based on novel data-embedding algorithms. The Num saved + = NOB
main objective of the study is to embed data between }
MP3 frames. Else If num saved < n then store from (I-offset) NOB to
end of temp
}
The Proposed method Store frame I in result
Store remaining
This method is summed up by embedding data between
frames in the MP3 file . REPLACEDATAIN
START
The major requirements of this method: TEMPARRAYWITH
THEENCRYPTED
1. k-lite multi-media player program(windows media DATA
player classic). READMP3
FILEINTO
2. The MP3 file must be of CBR type only. BUFFER
The following algorithm demonstrates the way of
FINDALL
embedding and extracting encrypted text file LOCATIONIN
TEMP=255
SETPOS=0
Embeding
STORELOCATIONS
COUNTINTEMP
1) Read MP3 file into buffer BEFOREALLDATAIN
2) Pos=0 HEXADECIMAL
IFPOS<BUFFER YES
3) While (Pos < Buffer)
{
SEARCHFORNEXT
SETLOCATION
Search for next header (header position) TO0INTEMP
NO HEADER
Remove all data before the header
Calculate FS READREMAINING
I=0,OFFSET=2 BYTESANDSTOREINTO REMOVEALLDATA
Read FS byte and store into array frame NOOFBYTES=CEILING(TEMPCOUNT/ (FRAME ARRAYFRAME BEFORETHE
Add frame array to all frame array COUNT-2)) HEADER
NUMSAVED=0
Pos = Pos + Header Pos + FS
}
CONVERTSTRINGTEXTDATATO CALCULATE
4) Read remaining and store IFI< BINARYBYUSINGUTF8ENCODING FRAMESIZE( FS)
5) Convert string text data to binary by using UTF8 FRAMESCOUNT
encoding YES READFSBYTEAND
6) Generate RSA keys GENERATERSA STOREINTOARRAY
KEYS FRAME
{ IFI>OFFSET
Store keys length in temp array YES
Store keys in temp array STOREKEYS
ADDFRAME
} NO STORE#BYTES
LENGTHINTEMP
ARRAY
ARRAYTOALL
INTORESULTING
7) Encrypt (data string and keys length and keys) and store FILE
FRAMEARRAY
in temp
8) Find all location in temp=255 STOREKEYSIN
{ TEMPARRAY SETPOS=POS+
HEADERPOS+FS
Store locations count in temp before all data in STOREAFRAME
INTORESULTING
hexadecimal FILE,I=I+1
Store all location in hexadecimal ENCRYPT(DATASTRINGAND
} KEYSLENGTHANDKEYS)
STORE
9) I=0 , Offset=2 REMAINING
NOB = Ceiling(temp count/ (frame count -2))
Num Saved =0
Fig.1. shows the flow chart for Embedding text of (BF)
While I < All frames count
{
If I> Offset then
204 IJCSNS International Journal of Computer Science and Network Security, VOL.10 No.7, July 2010
Extracting Pos = Pos + Header Pos + FS
}
1) Read MP3 file into buffer 4) Read hexadecimal from no of location.
2) Pos=0 5) Read hexadecimal from all location.
3) While (Pos<Buffer) 6) Convert hexadecimal into 255.
{ 7) Read (key length and keys.)
Search for next header (header position) 8) Decrypt data.
Save all data before the header 9) Convert data to string using UTF8 encoding .
Calculate FS 10) Show text.
Read FS byte and store into array frame
Add frame array to all frame array
S T A R T
R E A D M P 3 F IL E IN T O B U F F E R
S E T P O S = 0
S E A R C H F O R F IR S T H E A D E R
S A V E A L L D A T A B E F O R E T H E H E A D E R IN T O
T E M P A R R A Y
R E P E A T T H A T F O R A L L F R A M E S
R E A D L O C A T IO N S C O U N T A N D L O C A T IO N S F R O M T E M P
IN H E X A D E C IM A L
S E T L O C A T IO N S T O 2 5 5 IN T E M P
R E A D K E Y S L E N G T H F R O M T E M P A R R A Y
R E A D K E Y S F R O M T E M P A R R A Y
D E C R Y P T (D A T A S T R IN G A N D K E Y S L E N G T H A N D K E Y S )
C O N V E R T B IN A R Y D A T A T O S T R IN G T E X T B Y U S IN G U T F 8 E N C O D IN G
Fig.2. Flow Chart for Extracting text of (BAF)
IJCSNS International Journal of Computer Science and Network Security, VOL.10 No.7, July 2010 205
Different criteria have selected to measure either of the
best previous methods. Three scales have been used to
Experimental results and analysis evaluate those methods.
1- Time of embedding and extracting.
The optimum evaluative criterion is embedding as much 2- Size of file embedded.
text as possible with least time and high audio quality. A 3- Audio quality.
number of tests has been conducted on the proposed
method and Befor All Frames method (BAF), which Table (1) shows the results of methods BAF and BF using
summed up by embedding all encrypted text before the RSA algorithm for embedding files from both time and
first frame in MP3 file. size point of views:
Table (1) Embedding with RSA algorithm.
size file embed time of embedding time of embedding
#byte #frame original size results file size
KB BF(Sec) BAF(Sec)
1KB 3175 12281 5138432 5136260 1.944 1.6719
2KB 5279 12281 5138432 5138364 1.94575 1.765625
4KB 9527 12281 5138432 5142612 1.946375 1.9015
6KB 13703 12281 5138432 5146788 1.984375 1.923846875
8KB 18047 12281 5138432 5151132 1.99945
10KB 22103 12281 5138432 5155188 2.01145
20K 43343 12281 5138432 5176428 2.02112
40K 85367 12281 5138432 5218452 2.029923
80K 170543 12281 5138432 5303628 2.0337432
100K 212687 12281 5138432 5345772 2.109365
200K 424535 12281 5138432 5557620 2.568749
400K 847567 12281 5138432 5980652 3.118749
800K 1693999 12281 5138432 6827084 4.243749
1024KB 2166935 12281 5138432 7300020 4.88749
2048KB 4331847 12281 5138432 9464932 8.046873
4096KB 8666335 12281 5138432 13799420 14.26561
40960KB 86609191 12281 5138432 91742276 134.2
• From table (1) it is clearly seen that the embedded file Table (2) shows the results of methods BAF and BF using
size will affect the cover file size since the embedding RSA algorithm for extracting files time.
processes is to insert the hidden text between frames. Figure (2)shows the analysis results of table(2)
Although this method allow embedding large files within The figure above shows that the time consumed in
cover file, but it is recommended to embed relatively small extracting increases with the size of the file. BF proved its
files (with respect to cover size) in order not to make the ability in extracting files in a short time.
stego file suspicious.
Figure (1) shows the results of the analysis as shown in
table(1)
BF can embed 40960 KB in a short time and maintain the
same sound quality where as BAF did not exceed 6 KB .
206 IJCSNS International Journal of Computer Science and Network Security, VOL.10 No.7, July 2010
15
) 10
c
e
S
( 5
E
M
IT 0
SIZE OF FILE (KB)
time of embe ding BF tim e of embeding BAF
Figure (1) Embedding with RSA algorithm.
Table (2) Extracting with RSA algorithm.
time of extracting time of extracting
size file extract KB #byte #frame Original file size Result file size
BF(Sec) BAF(Sec)
1KB 3175 12281 5136260 5138432 1.284375 1.225
2KB 5279 12281 5138364 5138432 1.334375 1.328125
4KB 9527 12281 5142612 5138432 1.465625 1.4125
6KB 13703 12281 5146788 5138432 1.578125 1.528125
8KB 18047 12281 5151132 5138432 1.7
10KB 22103 12281 5155188 5138432 1.809375
20K 43343 12281 5176428 5138432 2.4313
40K 85367 12281 5218452 5138432 3.699999
80K 170543 12281 5303628 5138432 6.1375
100K 212687 12281 5345772 5138432 7.346875
200K 424535 12281 5557620 5138432 13.609375
400K 847567 12281 5980652 5138432 25.94688
800K 1693999 12281 6827084 5138432 50.63125
1024KB 2166935 12281 7300020 5138432 64.9
2048KB 4331847 12281 9464932 5138432 128.5
4096KB 8666335 12281 13799420 5138432 255
40960KB 86609191 12281 91742276 5138432 2618.2
15
) 10
c
e
S
( 5
E
M
I
T 0
SIZE OF FILE(KB)
time of extracting BF time of extracting BAF
Figure (2) Extracting with RSA algorithm.
IJCSNS International Journal of Computer Science and Network Security, VOL.10 No.7, July 2010 207
Information Hiding and Multimedia Signal Processing
(IIHMSP' 06), pp. 3-6, Dec. 2006 , Pasadena, CA, USA.
7. Conclusion: [8] B. Deng and J. Tan and B. Yang and X. Li, "A Novel
Steganography Method Based on Modifying Quantized
Steganography is really an interesting subject and is Spectrum Values of MPEG/Audio Layer III” , Proceedings
outside the mainstream of cryptography and system of the 7th WSEAS International Conference on Applied
Informatics and Communications, Athens, Greece, IEEE,
administration that most of us deal with day after day. But
pp.325-330,August 24-26, 2007.
it is also quite real; this is not just something that is used [9] Diqun Yan and Rangding Wang,” Reversible Data Hiding
in the lab or an arcane subject of study in academia. for Audio Based on Prediction Error Expansion”, CKC
In general, the suggested method satisfy the capacity and Software Laboratory, University of Ningbo, Ningbo 315211,
complexity of steganography properties in addition to that, China, International Conference on Intelligent Information
the suggested method for hiding are robust against (noise Hiding and Multimedia Signal Processing, IEEE, pp.249-
only). It is also considered highly secure since data is 252,2008.
encrypted using RSA algorithm before embedding data [10] N. Cvejic, T. Seppanen, “Increasing robustness of LSB
which makes the system secure especially agents passive audio steganography using a novel embedding method,” In
Proc. IEEE Int. Conf. Info. Tech. : Coding and Computing,
attack.
Vol. 2, pp. 533-537, April 2004.
[11] N. Cvejic, T. Seppanen, “Increasing the capacity of
To overcome the shortcomings of the proposed method, LSBbased audio steganography.” IEEE Workshop on
suggested :Before All Frame (BAF). Multimedia Signal Processing, pp. 336-338, 2002.
The merits of Before All Frames (BAF):
extracting the text is faster than other methods for limited
size, because the whole text lies in the beginning .
The song can be played on all multi-media audio players.
High audio quality.
A relatively good text file size.
It can be played whether the MP3 file encodes VBR or
CBR .
Taking in consideration that the major issue is the number
of frames in the cover file not the file size (as number of
frames will not actually depends on audio file size only
but also on frame size).
8. References
[1] W. Jonker, J.-P. Linnartz, "Digital Rights Management in
Consumer Electronics Products", IEEE Signal Processing
Magazine, Vol. 21, No. 2, pp.82-91, 2004.
[2] D. Pan, “A tutorial on MPEG/Audio compression”, IEEE
Multimedia, 2(2), pp. 60-74, 1995.
[3] E. Ambikairajah, AG Davis and W. Wong, “Auditory
masking and MPEG-1 audio compression”, IEE Electronics
& Communication Engineering Journal, 9(4), pp.165-175,
1997.
[4] Cheng-Te Wang, Tung-Shou Chen and Wen-Hung Chao,
“A new audio watermarking based on modified discrete
cosine transform of MPEG/Audio Layer III”, Proceedings
of the 2004 IEEE International Conference on Networking,
Sensing & Control, pp.984-989,2004.
[5] Cheng-Te Wang, Tung-Shou Chen and Wen-Hung Chao,
“A new audio watermarking based on modified discrete
cosine transform of MPEG/Audio Layer III”, Proceedings
of the 2004 IEEE International Conference on Networking,
Sensing & Control, pp.984-989,2004.
[6] Wang H, Wang S. Cyber warfare: steganography vs.
steganalysis. Communications of the ACM 2004;47(10).
[7] H. Matsuka, “Spread Spectrum Audio Steganography using
Sub-band Phase Shifting,” IEEE Int. conf. Intelligent
