BY JIM H. SMITH
In the tense days after Sept. 11, 2001, The New York Times reporter Gina Kolata introduced readers to a new and decidedly scary way to communicate. “The investigation of the terrorist attacks … is drawing new attention to a stealthy method of sending messages through the Internet,” Kolata wrote. “The method, called steganography, can hide messages in digital photographs or in music files but leave no outward trace that the files were altered.”
When Kolata’s article was published, Watson School Professor Jessica Fridrich, PhD '05, had just created for the United States Air Force a powerful tool that made it possible to search, with a high probability of success, for one of the most commonly used technologies used to conceal information. Fridrich, a professor in the Department of Electrical and Computer Engineering, developed complex algorithms that could analyze the numbers used to encode pixels — short for picture elements — the miniature dots of which visual images are composed.
Pixels are the smallest individual units of a picture that can be controlled in the reproduction process. They are so tiny that a single photo may contain thousands, or even millions, of them. So looking for messages hidden within a photo is daunting.
But Fridrich’s novel algorithms were up to the task. Designed to scan the numerical sequences of the pixels, they looked for anomalies. Those variations, which produce slightly altered images — too small for the naked eye to detect — reveal whether an image is clean or contains secret messages
Steganography, the process of burying messages within other messages, is not new. Concealing written information using strategies such as disappearing ink and codes has been around for centuries. Until recently those techniques were known, collectively, as cryptography.
What makes steganography (from the Greek for “hidden writing”) of special interest today, Fridrich says, is “digitization and use of the Internet to transmit messages. It’s essentially a brand new field.”
Among those who’ve contributed to the rapid advancement of that field, few have made a greater contribution than Fridrich. When her first-generation algorithms had trouble with JPEG images, she quickly retooled them. In the process she discovered a method for identifying digital cameras from their “fingerprints.” She was as amazed as everyone else when she learned that every digital camera leaves a unique, random-looking low-amplitude pattern on the images it produces.
A native of the Czech Republic, Fridrich earned her master’s in applied mathematics from Prague’s Czech Technical University in 1987. She completed her PhD in systems science at Binghamton in 1995 and was honored with Binghamton’s Distinguished Dissertation Award. Her work in steganography came about by serendipity, however.
“I’ve always liked puzzles,” she says, and, in fact, her Binghamton website features a “cubing page” that describes her unique method for solving Rubik’s Cube in a breathtaking 17 seconds! It was because of her interest in solving puzzles that she began researching chaos theory in the 1990s.
But, in 1998, she received her first research grant on steganography, from the Air Force. Since then she has devoted her research to the cat-andmouse business of inventing better and better technologies to embed information digitally — and find it. She has been awarded more than 20 grants by the U.S. Air Force, the National Science Foundation and industry, and she has published more than 100 papers, as well as a book, on the subject. Several of her students have gone on to positions in industry and academia, where they are applying and building upon her work.
Thanks to her, Binghamton is almost synonymous with hidden writing research. Just ask Electrical and Computer Engineering Assistant Professor Scott Craver.
The Next Generation of Covert Communications
As an undergraduate at Northern Illinois University, in 1997, Craver discovered a flaw in a digital watermarking security system. The ingenious method he invented to exploit the flaw made him famous. By the time he arrived at Princeton to work on his PhD in electrical engineering, the “Craver Attack” was already being cited in textbooks.
When Craver began to seriously contemplate a career in what he describes as “information hiding,” though, he encountered a problem. “Not many schools are working on information security,” he says. “Some have a single faculty member with an interest, but few have a department.”
The Watson School was the exception. He joined the faculty in 2004, and it has turned out to be a mutually beneficial relationship. A specialist in evading, breaking and reversing digital watermark security, Craver landed a three-year $300,000 grant through the Air Force’s Young Investigator Research Program two years after arriving in Binghamton. In 2009 he won the prestigious Presidential Early Career Award for Scientists and Engineers (PECASE), the highest honor bestowed by the U. S. government to researchers early in their careers.
PECASE provides Craver with $200,000 a year for five years, and with that money he and his students are researching the next generation of covert communications. “We work on counter-deception problems,” he says. “You’re trying to get away with something and I’m trying to catch you.” One of the best ways to do that, he says, is to break codes and information hiding screens. Consequently, he and his students strive to create increasingly difficult codes to break and ever-improving technologies for doing so.
Using computer simulations, they design proof-of-concept software to illustrate their techniques for concealing encrypted messages in a range of media. Recently they discovered a way to manipulate the imaging of a simple screen saver so as to conceal messages that evade detection by any pattern-based algorithms. If they can figure out what kinds of tools hackers will be using next, they can develop methods for countering them.
That task will require a new generation of covert communications experts to follow Fridrich and Craver’s lead. Toward that end, Binghamton established the Center for Advanced Information Technologies (CAIT), directed by Professor Victor Skormin, in July 2005. In addition to Craver — its deputy director — and Fridrich, the CAIT has seven other faculty members and three research assistants.
“The research being done by professors Fridrich and Craver and their colleagues is cutting-edge,” says Ron Miles, Watson School associate dean for research. “Few universities can match it. It’s a real point of distinction for Binghamton and the Watson School.”
When Binghamton’s new Engineering and Science Building opens in the fall, the Seymore Kunis Media Core will enhance both Fridrich’s and Craver’s work. Named with a gift from Gary Kunis ’73 in honor of his father, a strong supporter of public education, the lab will provide flexibility for physical experiments that Craver says are often constrained by the limits of his current lab space. And it will feature the advanced computing capacity needed to solve the powerful algorithms that (as prominent computer scientist Rafael Alonso once said in describing the potential of Fridrich’s work) shine “a flashlight in the sewers of the Web.”
In the game of steganalysis, learning is always about trying to up the ante. Is there a better, less detectable way to hide information? And, if so, what's it going to take to find the hidden information?
Those were the essential questions posed last year by the first major international scientific steganalysis contest. Three prominent steganalysis researchers –— Tomas Filler, PhD '11, from Binghamton University, Tomas Pevny, PhD '08, from the Technical University of Prague and Patrick Bas from the Centre National de Recherche Scientifique in Lille, France — collaborated on BOSS (Break Our Steganographic System). The team was well qualified to design a tough puzzle. Two of the three — Filler and Pevny — learned from the master, Jessica Fridrich, PhD '95, professor of electrical and computer engineering.
What the three researchers wanted to test was the security of HUGO (Highly Undetectable steGO), the steganographic technique they had developed for the contest. Last September they made 1,000 images available on the BOSS website. Teams from all over the world pitted their steganalysis skills against the challenge, trying to determine which of the images contained hidden messages and which did not.
So difficult was the task, in fact, that the HUGO images remained online for four months.
And that was barely enough time to complete the analysis, says Fridrich, whose team of students, the Hugobreakers, ultimately won the contest, discovering more than 80 percent of the hidden messages.
"I was somewhat skeptical about the educational value of the contest," Fridrich concedes. "It took a huge time commitment. But in the end I became a big believer and supporter of such competitions as the means for advancing the field. We all learned a lot from participating in it, and I think this sort of exercise can be a powerful learning tool."