New Threat: Network Eavesdropping

LAS VEGAS -- Black Hat USA -- Attackers don't have to be able to read your data in order to hack your network, researchers said in a presentation here today.

A smart hacker can learn a great deal about your network and its vulnerabilities simply by analyzing its traffic patterns via the Internet, according to a quartet of researchers who presented their findings.

The presentation, entitled "Traffic Analysis: The Most Powerful and Least Understood Attack Method," was given by Jon Callas, CTO and CSO of PGP; Raven Alder, an independent security researcher; Riccardo Bettati, an associate professor in computer science at Texas A&M University; and Nick Mathewson, one of the developers of the Tor privacy network.

The researchers showed how detailed analysis of timestamps on packets can enable an attacker to identify how and what a particular user is typing. Attackers can also use traffic data to identify the operating system used on a particular server, or even what a password might be.

"It's even worse in wireless networks, where this sort of analysis can make it possible not only to find out who the sender is, but where the receiver might be," Callas said.

Some vendors and enterprises have tried to obfuscate their traffic patterns by embedding real traffic in a constant stream of dummy traffic. But this strategy, known as "link padding," is relatively easy to defeat, and it is possible for attackers to separate the real traffic from the fake stuff, the researchers said.

"The greatest problem with this sort of exploit is that it is very difficult, if not impossible, to defend against," Callas said. The mathematics required to mask the patterns of a particular traffic stream is much more complex than the math required for attackers to analyze traffic in nefarious ways, the researchers said.

The good news is that traffic analysis can also be used by the good guys to root out botnets and other attacks, the researchers observed. "By looking at ping response times, it's possible to tell the difference between users and botnets," Callas said. And once a botnet node is identified, it is possible to use traffic analysis to track down the other nodes on a hop-by-hop basis, he said.

In the future, traffic analysis might be used to identify other types of attacks, such as spam or click fraud, the researchers said. "We're just bringing out the techniques and issues now -- we're hoping that others will apply these findings to other areas," Callas said.

— Tim Wilson, Site Editor, Dark Reading