News Release

Chandramouli and partners use US Air Force STTR grant

Grant and Award Announcement

Stevens Institute of Technology

HOBOKEN, N.J. -- An Assistant Professor of Electrical and Computer Engineering at Stevens Institute of Technology, Dr. Rajarathnam Chandramouli, recently received a Small Business Technology Transfer (STTR) Phase-I grant from the US Air Force Research Lab to develop a commercial software product based on his research in steganalysis, or the detection of hidden digital information, to automatically scan e-mails and web downloads for hidden messages.

Chandramouli received the $100,000 STTR grant to pursue the development project jointly with David Wagner Associates of Pennsylvania. The grant applies to activities performed Sept. 2004-May 2005 for the production of an initial prototype and demonstration of the software's capabilities.

Steganography is the process of hiding information within a cover signal. Very commonly the cover signal is a large digital file such as a digital image. To hide the information, some features of the cover signal are slightly modified by an embedding technique. The features of the cover signal that are modified are typically chosen by using a secret key. Ideally, the modification to these features results in imperceptible changes to the cover signal, so that the presence of the hidden information as well as its content is not readily apparent.

"Recent media coverage has focused on the possibility of terrorists and other criminals using steganography to conceal information," said Chandramouli. "The advantage of such a technique is twofold; it allows messages to be passed undetected and it allows messages to be posted in public forums from which it is difficult or impossible to trace those who access the stego-object. In addition, an employee or contractor of a private company could send sensitive information without raising suspicion by using e-mail containing steganographic content. Naturally, law enforcement officials, government agencies, and corporate security officers would like to develop tools to identify (and extract) such stego-information. We are proposing to develop such a system."

There are several different steganographic techniques for hiding information and even more steganalytic techniques for discovering the presence of hidden information. Because of the ever-changing and developing nature of steganography and steganalysis, Chandramouli will design a software framework that will allow a plug-and-play approach to incorporate a variety of steganalytic tools and techniques.

"We would like the framework to facilitate the ability of the user to scan selected files, directories, or drives on a system, to scan emails as they are received, and to scan web pages as they are loaded into a browser," he said.

Such a system could be used in conjunction with a web crawler to detect publicly posted media steg-files.

"We propose a platform-independent Rapid Steganographic Detection Suite (RSDS)," said Chandramouli, "equipped with the following capabilities: to run as a background process above the mail server, filtering incoming mail and scan for steganographic content; to run as a layer above the internet browser, scanning browsed URLs for steganographic content; to scan selected files, directories or system drives for steganographic content. The user will select the types of files that are to be scanned." The RSDS will include an option to de-noise each media file (using wavelet analysis), create a corresponding hash entry, and determine if multiple versions of the same picture may be appearing. This functionality will allow the user to detect repeated appearances of altered images.

In each of the first three applications, the user will also have the ability to select the following operational parameters:

1. For each type of media (.jpg, .gif, .wav, etc), the specific steganalytic algorithms to be employed, as well as the acceptable false alarm rate for each algorithm.

2. Whether a dictionary attack is to be run against those media files testing positive and the lexicons that should be used in the dictionary attack.

"We also may wish to focus extra attention on those file types that are most likely to be used to embed steganographic content," Chandramouli said. "Examples would be grayscale images and suspiciously large files."

Based on the results at the end of the STTR Phase I grant, Chandramouli plans to apply with Wagner Associates for Phase-II funding, which would yield $500,000 for three years.

###

Established in 1870, Stevens offers baccalaureate, masters and doctoral degrees in engineering, science, computer science, management and technology management, as well as a baccalaureate in the humanities and liberal arts, and in business and technology. The university has enrollments of approximately 1,780 undergraduates and 2,700 graduate students, and a current enrollment of 2,250 online-learning students worldwide. Additional information may be obtained from its web page at www.Stevens.edu.

For the latest news about Stevens, please visit www.StevensNewsService.com.


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.