Bivio Networks supplied its deep packet inspection (DPI) platform for an information leak prevention research project commissioned by the Ministry of Internal Affairs and Communications of Japan (MIC). Hitachi served as the representative research center and the project lead on behalf of MIC.
Japan’s MIC commissioned the information leakage research project in January 2010 to study network information transfers initiated by P2P file-sharing software such as Winny. The project’s scope included the development of an engine and networking solution to detect and extract only the P2P file-sharing software from traffic. The project studied the hardware feasibility of such a networking system, the development of technology for traffic control, and completion of a large-scale field trial to verify the system.
Bivio Networks said P2P file-sharing software is often a primary driver of information leakage, which has become a serious concern for individuals, companies and government organizations alike.
The Hitachi Systems Development Laboratory oversaw the entire project’s design and management, including the development of the technology to detect P2P software-initiated communications. Hitachi conducted the large-scale field trial of the system, using the large-scale Internet simulator provided by the National Institute of Information and Communications Technology’s Hokuriku Research Center.
At the project’s outset, seven types of P2P file-sharing software — Winny, LimeWire, WinMX, BitTorrent, Share, Perfect Dark, and Winnyp — were selected based on research indicating that they jointly controlled more than 90% of the P2P file-sharing software market. Then, the detection module for the software was deployed on the Bivio DPI Application Platform, which was connected to the controlling devices.
The January 2010 trial achieved the target performance of 10 Gbps. Moreover, with two interconnected Bivio DPI Application Platforms, traffic processing speed reached 20 Gbps.
Bivio said its DPI Application Platform and software detection module achieved a detection rate between 98.87% and 100% of P2P file-sharing software communications. In addition, the system demonstrated a zero percent error rate in discriminating between non-P2P and P2P communications.
http://www.bivio.net
