SURVEY OF STEGANOGRAPHY With an emphasis on audio techniques. By: Eric Hernandez DISCUSSION OVERVIEW What is steganography? Definition Dos/Donts Steganographic techniques Images Audio Other
applications Steganalysis Overview Specific audio example Examples of use in the real world WHAT IS STEGANOGRAPHY? Literally, covered writing Invisible inks Character arrangement (non-cryptographic)
Wax coverings Formal Definition: Steganography is the technique of concealing information within seemingly innocuous carriers. It is the art and science of writing hidden messages in such a way that no one, apart from the sender and intended recipient, suspects the existence of the message. RULES OF STEGANOGRAPHY 1. Strength is not stealth Stegonography =/= Cryptography Not exclusive 2.
Assume the hiding method is known 3. 4. Known as Kerchkoffs principle in crypto Never divulge side information Dont add noise to a where it doesnt belong Cover images Audio STEGANOGRAPHIC TECHNIQUES STEGANOGRAPHIC TECHNIQUES: IMAGES
Common approaches: LSB modification Masking Filtering Transformations via algorithms Not all approaches created equal Size restrictions Vulnerability to manipulation STEGANOGRAPHIC TECHNIQUES: IMAGES, LSB Composed of hiding information inside of the LSB of an image Leads to a few requirements:
Image must have suitable noise Image must be of sufficient size Must be able to hide in plain sight Advantages/Disadvantages Easy to implement Does not stand up to compression Scalability STEGANOGRAPHIC TECHNIQUES: IMAGES, MASKING AND FILTERING Most commonly seen as watermarks as a
mean to protect copyright Not traditional steganography in that information is being extended, not hidden Accomplished not by hiding in noise, but by manipulating luminescence Changes visible properties of an image, but can be still made to be undetectable to the human eye Advantages/Disadvantages More
suitable for lossy compression than LSB Not traditional stego STEGANOGRAPHIC TECHNIQUES: IMAGES, ALGOS AND TRANS Based on transformation via formula Discrete Cosine Transformation (DCT) Fast Fourier Transform Advantages/Disadvantages Can be visually undetectable Can scatter secret information across image Can be used together with encoding and similar watermarking (luminescence techniques)
Not as susceptible to cropping, compression, etc. Complex STEGANOGRAPHIC TECHNIQUES: AUDIO Human Auditory System (HAS) has a large dynamic range that it can listen through Perceives over a range of power > 1,000,000:1 Range of frequencies > 1,000:1 Makes it hard to add remove data from original sources HAS has a weakness, though: sound differentiation Loud
vs. Soft Several ways to exploit this weakness STEGANOGRAPHIC TECHNIQUES: AUDIO 1. Low-bit encoding Similar to previously discussed LSB stego Easy to use, weak to any changes in audio 2. Phase encoding Substitutes the phase of an initial audio segment with a reference phase that represents the hidden data
Inaudible in terms of signal-to-perceived noise ratio Low data transmission rate Best if used for watermarks or similar small data STEGANOGRAPHIC TECHNIQUES: AUDIO 3. Spread spectrum Spreads secret information across frequency spectrum Two schemes
High level of robustness against removal Possibility of introducing noise 4. Direct Sequence Frequency-hopping Echo Hiding Blends information by introducing echo Vary amplitude, decay rate, offset (delay time) of echo High level of robustness
Echoes mimic environmental conditions (lossy compress.) STEGANOGRAPHIC TECHNIQUES: AUDIO, LSB Can be augmented with pseudorandom number generator to scatter information over audio file. STEGANOGRAPHIC TECHNIQUES: AUDIO, PHASE ENCODING Encodes message as phase shifts in the phase spectrum of a digital signal.
Relies on the fact that phase components of sound are not as perceptible as noise is Steps: 1. Break signal into smaller segments 2. Discrete Fourier Transform (DFT) 3. Phase diff. between adjacent segments 4. Insert message into phase vector of first signal segment only
5. New phase matrix created from new phase of first segment and the original phase difference 6. Use inverse DFT and concatenate segments back together STEGANOGRAPHIC TECHNIQUES: AUDIO, SPREAD SPECTRUM Spreads secret information across the audio signals frequency spectrum Direct-sequence SS Message is spread out by a
constant called the chip rate. Modulated with a pseudorandom signal Interleaved with cover Frequency-hopping SS Audio files frequency spectrum is altered so that it hops randomly between frequencies STEGANOGRAPHIC TECHNIQUES: AUDIO, ECHO HIDING Vary 3 characteristics of
introduced echo Amplitude Decay rate Offset (delay time) Different vary times refer to different bit input So small that humans dont notice STEGANOGRAPHIC TECHNIQUES: AUDIO, ECHO HIDING Break audio into segments Message to be
inserted is Hey After code runs, recombine segments into final signal Creates a noticeable mix of echoes, increasing chance of detection init(Block blocks[]) { for (int i=0; i < blocks.length; i++) { if (blocks[i].echoValue() == 0) blocks[i] = offset0(blocks[i]); else blocks[i] = offset1(blocks[i]); } }Block offset0(Block block) { return (block + (block - OFFSET_0)); }
Block offset1(Block block) { return (block + (block - OFFSET_1)); } STEGANOGRAPHIC TECHNIQUES: AUDIO, ECHO HIDING Create two mixer signals Zero mixer One mixer Less abrupt than before Signals are compliments of each other Ramp ups and downs help with smooth transitions
between echoes Recovery Break signal into same block sequence Autocorrelation function of the signals cepstrum Reveals a spike at each echo time offset STEGANOGRAPHIC TECHNIQUES: OTHER APPLICATIONS Text steganography Little
noise in text to hide information in, but possibilities still exist: Since everyone can read, encoding text in neutral sentences is doubtfully effective. Line number/character number Vertical letter shifts Font changes (make some bs, ds, ps longer than others) Video steganography Combines ideas of both image and audio stego Open systems environment steganography TCP/IP
Containers STEGANALYSIS STEGANALYSIS: OVERVIEW Detect Difficult Need to understand technique in use Involves statistical analysis Especially difficult when Kerckhoffs principle is applied Defeat
Easy No understanding required No statistical analysis required As techniques evolve, will become more difficult STEGANALYSIS: TEXT Detect Defeat Patterns being disturbed
Inappropriate (odd) use of language Line height and white space Difficult when secret key (outside text) is used Rewrite Change spacing, punctuation Modify line heights or layout Add or remove words
Does not work on public cover source Key must be intercepted STEGANALYSIS: IMAGES Detect Defeat Changes in: size File format Last modified time stamp
Color palette File Statistical analysis Examine the LSB Compress using lossy compression Destroys LSB modification
Alter image: Resize Modify color palette Conversion to different file format Luminescence STEGANALYSIS: AUDIO Detect Statistical analysis Similar to images, look at LSB
Examine inaudible frequencies Patterns in background noise Distortions Defeat Similar to images Compression special note: Removes frequencies that cannot be heard (a hideout for some stego)
Lower bitrate Less available space to store stego = at least some data loss STEGANALYSIS: HAUSDORFF DISTANCE Calculation of Hausdorff distance between lines X and Y USES IN THE REAL WORLD STEGANOGRAPHY: USES Copyright Watermarks
Covert military operations Keys STEGANOGRAPHY: USES STEGANOGRAPHY: USES STEGANOGRAPHY: USES STEGANOGRAPHY: USES STEGANOGRAPHY: USES RESOURCES
A Detailed look at Steganographic Techniques and their use in an Open-Systems Environment http://www.sans.org/reading_room/whitepapers/covert/de tailed-steganographic-techniques-open-systems-environ ment_677 Exploring Steganography: Seeing the Unseen by Neil F. Johnson, Sushil Jajodia http://www.jjtc.com/pub/r2026.pdf Steganography and Steganalysis by J.R. Krenn http://www.krenn.nl/univ/cry/steg/article.pdf
Steganalysis: The Investigation of Hidden Information by Neil F. Johnson, Sushil Jajodia http://www.simovits.com/archive/it98jjgmu.pdf Echo Hiding by Daniel Gruhl, Anthony Lu, Walter Bender http://www.almaden.ibm.com/cs/people/dgruhl/edh2.pdf RESOURCES A Novel Audio Steganalysis Based on HighOrder Statistics of a Distortion Measure with Hausdorff Distance by Yali Liu, others http://www.ece.ucdavis.edu/~yliu/pub/papers/Tra cy_ISC08.pdf
Methods of Audio Steganography http://www.snotmonkey.com/work/school/405/me thods.html#phase Secret Code in Color Printers Lets Government Track You http://www.eff.org/press/archives/2005/10/16 Hausdorff Distance http://en.wikipedia.org/wiki/Hausdorff_distance
