dan deblasio and brad mundt university of central florida

22
An Enhanced Steganographic An Enhanced Steganographic Method for Data Hiding in Method for Data Hiding in Microsoft Word Documents Microsoft Word Documents by a by a Change Tracking Technique Change Tracking Technique Dan DeBlasio and Brad Mundt Dan DeBlasio and Brad Mundt University of Central Florida University of Central Florida [email protected], [email protected], [email protected] [email protected]

Upload: zared

Post on 30-Jan-2016

35 views

Category:

Documents


0 download

DESCRIPTION

An Enhanced Steganographic Method for Data Hiding in Microsoft Word Documents by a Change Tracking Technique. Dan DeBlasio and Brad Mundt University of Central Florida [email protected], [email protected]. Steganography. What is Steg? - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Dan DeBlasio and Brad Mundt University of Central Florida

An Enhanced Steganographic Method An Enhanced Steganographic Method for Data Hiding in for Data Hiding in

Microsoft Word Documents Microsoft Word Documents by a by a

Change Tracking TechniqueChange Tracking Technique

Dan DeBlasio and Brad MundtDan DeBlasio and Brad MundtUniversity of Central FloridaUniversity of Central Florida

[email protected], [email protected] [email protected], [email protected]

Page 2: Dan DeBlasio and Brad Mundt University of Central Florida

SteganographySteganography

What is Steg?What is Steg?

The word steganography is of Greek origin The word steganography is of Greek origin and means "covered, or hidden writing". and means "covered, or hidden writing".

Modern implementation is embedding secret Modern implementation is embedding secret information into something innocentinformation into something innocent

Pictures, audio, videoPictures, audio, video

Page 3: Dan DeBlasio and Brad Mundt University of Central Florida

Picture StegPicture Steg

By removing all but the last 2 bits of each color component, an almost completely black image results. Making the resulting image 85 times brighter results in the image to the right.

Image extracted from tree image.

Image of a tree.

Page 4: Dan DeBlasio and Brad Mundt University of Central Florida

Our selectionOur selection

A New Steganographic Method for Data Hiding in Microsoft Word Documents by a Change Tracking Technique by Tsung-Yuan Liu and Wen-Hsiang Tsai

A new steg method to hide information via MS Word A new steg method to hide information via MS Word change trackingchange tracking

Collaborative document with correctionsCollaborative document with corrections

Deleted words are struck through and new words are Deleted words are struck through and new words are underlinedunderlined

Page 5: Dan DeBlasio and Brad Mundt University of Central Florida

OverviewOverview

Transform a regular document to one with Transform a regular document to one with misspellings or mistakesmisspellings or mistakes

Make changes, allow word to track them.Make changes, allow word to track them.

Using a Huffman tree on the original words in Using a Huffman tree on the original words in the file, create a binary string of the embedded the file, create a binary string of the embedded messagemessage

Decode the binary string to get the secret Decode the binary string to get the secret messagemessage

Page 6: Dan DeBlasio and Brad Mundt University of Central Florida

The quick brown fox jumped over the lazy dog.

Original sentenceOriginal sentence

Page 7: Dan DeBlasio and Brad Mundt University of Central Florida

The quick borwn fox jumped oevr the lazy dag.

Sentences with MisspellingsSentences with Misspellings

Page 8: Dan DeBlasio and Brad Mundt University of Central Florida

The quick borwnbrown fox jumped oevrover the lazy dagdog.

Sentence with changes trackedSentence with changes tracked

Page 9: Dan DeBlasio and Brad Mundt University of Central Florida

Huffman TreeHuffman Tree

Created from the frequency of misspellings

WordWord HuffmanHuffman

tehteh 1111

qwikqwik 101101

borwnborwn 1001010010

brounbroun 00110011

jumppedjumpped 000010000010

oevroevr 00011100001110

orverorver 100010100010

dagdag 0101

Page 10: Dan DeBlasio and Brad Mundt University of Central Florida

The quick borwnbrown fox jumped oevrover the lazy dagdog.

How to DecodeHow to Decode

WordWord HuffmanHuffman

tehteh 1111

qwikqwik 101101

borwnborwn 1001010010

brounbroun 00110011

jumppedjumpped 000010000010

oevroevr 00011100001110

orverorver 100010100010

dagdag 0101

1001000 0111001

0x72 0x105

Hi

Page 11: Dan DeBlasio and Brad Mundt University of Central Florida

Our implementationOur implementation

Recreated the method from scratchRecreated the method from scratch

Made improvementsMade improvements

Modular code for easy improvementsModular code for easy improvements

Encryption, simple XOR pad with key in the Encryption, simple XOR pad with key in the author fieldauthor field

Page 12: Dan DeBlasio and Brad Mundt University of Central Florida

Example next,Example next,then a live demothen a live demo

Page 13: Dan DeBlasio and Brad Mundt University of Central Florida

A Word documentA Word documentBefore steg encodingBefore steg encoding

Page 14: Dan DeBlasio and Brad Mundt University of Central Florida

Encoding in ProgressEncoding in Progress

Page 15: Dan DeBlasio and Brad Mundt University of Central Florida

Encoding CompleteEncoding Complete

Page 16: Dan DeBlasio and Brad Mundt University of Central Florida

The Word documentThe Word documentAfter steg encodingAfter steg encoding

Page 17: Dan DeBlasio and Brad Mundt University of Central Florida

Decoding in ProgressDecoding in Progress

Page 18: Dan DeBlasio and Brad Mundt University of Central Florida

Decoding completeDecoding complete

Page 19: Dan DeBlasio and Brad Mundt University of Central Florida

Live DemoLive Demo

Page 20: Dan DeBlasio and Brad Mundt University of Central Florida

Statistical ResultsStatistical Results

Changes Bits b/change Words % changed

Sent. Changes/Sent

Document 1 40.4 93.8 2.321 475 8.5% 29 1.393

Document 2 43.2 99.8 2.310 551 7.8% 31 1.394

Document 3 65.6 129.6 1.976 641 10.2% 51 1.286

Document 4 124.4 224.8 1.968 1268 9.8% 89 1.398

Document 5 37.6 78 2.074 481 7.8% 27 1.392

Page 21: Dan DeBlasio and Brad Mundt University of Central Florida

ImprovementsImprovements

Due to the code being modular, improvements Due to the code being modular, improvements will be easywill be easy

More advanced or alternate encryption algorithmsMore advanced or alternate encryption algorithms

Implement a Markov model instead of a Huffman tree Implement a Markov model instead of a Huffman tree for statistical attacksfor statistical attacks

Use synonyms instead of misspellings or a Use synonyms instead of misspellings or a combination of bothcombination of both

Page 22: Dan DeBlasio and Brad Mundt University of Central Florida

That’s all for now folks!That’s all for now folks!