While police use fingerprints to determine who stole a piece of property or handled a murder weapon, engineers may soon be able to use digital fingerprints to determine who pirated a multimedia file.
Negar Kiyavash, a UI doctoral student in electrical and computer engineering, is helping in the battle against multimedia piracy. Her recent paper, titled “On Optimal Collusion Strategies for Fingerprinting,” won the Best Student Paper Award in the multimedia signal processing area at the International Conference on Acoustics, Speech, and Signal Processing (ICASSP) in Toulouse, France.
Kiyavash won a $500 prize for her paper and is planning to continue her research on the subject for several more years.
Pierre Moulin, a professor of electrical and computer engineering and a researcher at the Beckman Institute for Advanced Science and Technology, is co-author of the paper. Moulin and Kiyavash have been researching theory and applications for digital fingerprinting. For instance, one could make each copy of a DVD or music file slightly different so that if it is tampered with, engineers can determine who the culprit was. A strong challenge to digital fingerprinting is mounted when two or more people collude to create a forgery that combines their individual copies, so that this forgery cannot be easily traced to the colluders involved.
“The first step is to find what the worst attack is,” Kiyavash said. “The best way for colluders to avoid being caught is to try to make the forgery as close to the original as possible, to reduce the visibility of the fingerprints.”
Kiyavash said that the more people collude, the harder it is to determine who colluded. According to Moulin, much research went toward identifying the maximum number of colluders that the fingerprinting system can cope with, i.e. the system can detect the fingerprint of at least one of the colluders.
“There are definite limits to what we could identify,” Moulin said. “For instance, depending on the nature of the media signal, we could be successful if there are as many as 100 colluders, but not 1,000.”
Kiyavash said that the more users there are, the less robust the fingerprinting system is, and the harder it is to identify a colluder. There is a tradeoff between the number of users and the number of colluders you can tolerate.
While Hollywood is already using the technology, especially in limited release films, Kiyavash said the technique is relatively new, and she is looking forward to improving it over the next few years.
“It’s very mathematical. Information theory, signal processing and coding theory, all play a central role and they are all very close to my heart,” Kiyavash said. She will finish her doctorate this fall and will continue her research as a postdoctorate fellow at the university in the spring.
“The applications go far beyond the movie piracy. Any multimedia content or form of intellectual property that you can imagine can be subjected to collusion attacks, and these attacks are facilitated by availability of peer-to-peer networks.”
