05.06.2013 Views

PNNL-13501 - Pacific Northwest National Laboratory

PNNL-13501 - Pacific Northwest National Laboratory

PNNL-13501 - Pacific Northwest National Laboratory

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

The second goal was to research ways to reduce the<br />

massive data logs that need to be reviewed by experts for<br />

detecting unusual events or compromises for which there<br />

is currently no signature. New concepts for log<br />

presentations will be created using multidimensional<br />

representations. Network packet data will be vectorized<br />

by mapping the packet meta data into three dimensions<br />

with source, destination intellectual property, and port<br />

numbers and time constituting the major axis. A unique<br />

clustering algorithm, which had already been developed,<br />

will dynamically convert and map Internet protocol<br />

addresses to “x” and “y” coordinate pairs while<br />

preserving network topological relationships. This<br />

approach will allow the display of current and historical<br />

network traffic data. Most intrusion detection sensors<br />

systems today work only on single packets or a small<br />

window of packets. The interactive network traffic<br />

modeling model will collect and preserve session data<br />

(collections of packets) that will allow the opportunity to<br />

detect the “low and slow” attacks which confound most<br />

intrusion detection sensors systems that are available<br />

today.<br />

Last, the analysis techniques used for providing visual<br />

representation will also serve as a provider of updated<br />

threshold data for the sniffer sensors described above.<br />

Analysts will be able to make changes manually and<br />

programatically through the interactive network traffic<br />

modeling API in the continually updated visual model<br />

that subsequently will alter data collection agents<br />

(sniffers) allowing real-time detection and adaptation to<br />

changing events.<br />

Results and Accomplishments<br />

The original scope of this project was redefined as a<br />

proof-of-concept functional prototype. (See Figure 1.)<br />

Networked<br />

resource<br />

*<br />

1<br />

Intrusion Detection System<br />

(IDS)<br />

hub / router<br />

sniffer<br />

1<br />

*<br />

Broadcast Network Segment<br />

sniffer<br />

Summary and Conclusions<br />

Other Analysis Engines<br />

Interactive Network Traffic Model (INTM)<br />

sniffer<br />

Switched Nodes<br />

Figure 1. Interactive network traffic model architecture<br />

Existing and potential DOE clients have expressed strong<br />

interest in interactive network traffic model technology.<br />

Our strategy for pursuing this technology in the Intrusion<br />

Detection field is to develop a proof-of-concept prototype<br />

to demonstrate capability and implementation feasibility.<br />

Demonstration of true, real-time intrusion detection at<br />

high bandwidth is a necessary prerequisite in this research<br />

domain.<br />

Computer Science and Information Technology 161<br />

1<br />

1<br />

1<br />

1

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!