A supercomputer to visualize global traffic on the Internet!

A supercomputer to visualize global traffic on the Internet!

Technology

US researchers have used a supercomputer with the ability to display real-time global Internet traffic. It is an AI-boosted tool that could be useful for network maintenance and cybersecurity, among others.

A new tool
As reported in a statement released by MIT on October 27, 2019, a team of researchers is creating a way to fully visualize Internet traffic. And yet, analyzing this incredible mass of data exchange is a daunting challenge. The purpose of the maneuver? Identify and resolve failures, optimize defenses against cyberattacks and possibly improve the efficiency and quality of the network.

MIT scientists say they have analyzed no less than 50 billion data packets. These have been collected since 2015 in the state of California, but also in Japan. However, this new tool for analyzing global internet traffic works thanks to a supercomputer.

A calculating monster
It should be noted that the researchers first applied a technique called Dynamic Distributed Dimensional Data Mode (D4M) whose details are visible in a folder (PDF in English / 4 pages). It was a question of treating data with more empty spaces than values ​​(data called “hypercreuses”). Subsequently, a neural network was created, the latter operating thanks to the MIT SuperCloud. It is a supercomputer – composed of 10,000 processors – specially developed for the development of artificial intelligences.

However, this monster of calculation made it possible to analyze in real time as well the data coming from the giants of the Web as those of more discrete platforms. This is an interesting advance as traditional traffic analysis techniques can only analyze small samples of packets exchanged. In addition, these analyzes are generally limited by the location of the sources.

The results were interpreted as a measure of the “background noise” of the Internet. In short, this makes it possible to identify a lot of things. Indeed, researchers are now able to spot malicious IP addresses, spam, the scope of cyber attacks and other anomalies such as traffic jams. It is also possible to obtain information about file sharing.