Beyond encryption: Data transformations with homeland security potential
Jim Mitchell Communications Services Oklahoma State University November 18, 2002
OSU Professor Rathindra Sarathy has developed processes for dealing with computer data which promise to help agencies enhance homeland security without greater threat to individual privacy.
A professor from OSU and his colleague have developed various ways to protect computer data that will frustrate even the best hackers while giving agencies a greater capability to share such data without threat to individual privacy. "Perhaps the biggest selling point for our research is that it allows us to reject the argument that Americans must be willing to give up more of their privacy in order for agencies to share data necessary for better homeland security," says Dr. Rathindra Sarathy, associate professor of Management Science and Information Systems (MSIS) in the College of Business Administration.
According to Sarathy, the various methods he and Dr. Krish Muralidhar from the University of Kentucky have developed are currently being patented for use. They are data transformation processes called "data perturbation" and "data shuffling."
"The message we want to send with these processes is that they do not require any tradeoff÷we can use them to protect individual privacy while giving agencies the data they need to make better decisions, including security decisions," insists Sarathy.
The researchers' procedures do not rely on encrypting or coding the data but instead use mathematical transformations that randomly "perturb" or "shuffle" the original data. According to Sarathy, there are some factors worth distinguishing between the two systems.
"Encryption allows only certain authorized people to view individual pieces of data, but theoretically, it's a process that can still be reversed and information can be retrieved that can violate privacy. In addition, you can't perform any analysis on encrypted data," notes Sarathy.
"On the other hand, our procedures make it possible to ensure that no one sees the specific details of an original data set on their computer screen while still being able to use the data for analysis."
Sarathy anticipates that these processes will be especially useful not only for security purposes but for social scientists, government agencies, or business analysts who routinely perform data mining.
The professor describes one of the procedures as similar to "shuffling cards in a deck."
"Let's say each card in the deck contains a certain number and if you add them up, you get a certain sum. By shuffling the cards, the numbers in a particular pile of cards don't change, they are just shuffled to the point that no one knows what their original order was. Even though the cards are now out of order, you can still add them up and get the same sum. That's the end result of the process--data you can still use."
For instance, according to Sarathy, the program can ensure that no one can match names or salary figures of individuals in a particular data set while still being able to provide average income figures for the group as a whole. In this way, the data is still usable without these specifics.
As for a homeland security application, Sarathy says his program could allow law enforcement officers greater use of data sets that may have been off limits because they contained certain information that would cause privacy concerns. "These perturbation and shuffling processes allow those concerns to be addressed up front, before the information is ever handed over to security authorities."
Sarathy and Muralidhar have applied for a patent and have already made presentations to the U.S. Census Bureau where they suspect the processes could be especially useful. Their work has also triggered some interest among data confidentiality researchers in Europe. They hope to be talking to officials from private industry as well.
For information about this page, send e-mail to
Jim Mitchell.
