A New Way to Beat Spam?

You've tried blacklisting. And whitelisting. And half a dozen content-filtering tools. But every day, you and your users still paw through multiple spam messages, trying to find the stuff that's real email.

There's gotta be a better way, you say. And a couple of researchers at Georgia Tech think they've found it.

At the Association of Computing Machinery's annual Special Interest Group on Data Communication (SIGCOMM) yesterday, a Georgia Tech assistant professor and his doctoral student presented a paper that indicates it may be possible to detect and eradicate spam at the network layer, without using content filters.

In their paper, "Understanding the Network-Level Behavior of Spammers," Prof. Nick Feamster and Anirudh Ramachandran revealed the results of a 17-month study of more than 10 million spam messages sent over the Internet. The primary finding: Spam generally travels a different path on the network than real email.

The study found that most span is sent from a few regions of IP address space, and that spammers are using transient "bots" that send only a few pieces of email over a very short period of time.

"These trends suggest that developing algorithms to identify botnet membership, filtering email messages based on network-level properties (which are less variable than email content), and improving the security of the Internet routing infrastructure may prove to be extremely effective for combating spam," the paper says.

Security managers, email administrators and end users are constantly frustrated by content-based spam filters, which frequently block desired messages while allowing junk mail to accumulate in the user's mailbox. The problem, experts say, is that content filters generally use words and phrases to flag spam, and spammers can circumvent them by simply rewording or misspelling their messages.

The Georgia Tech researchers, on the other hand, studied IP addresses, routing paths and packet breakdown of spam messages to identify spam. "More than 10 percent of spam [in the study] originated from mail relays in two attack sources," the paper says, "and 36 percent of all received spam originated form only 20 attack sources. With a few exceptions, the attack sources containing hosts responsible for sending large quantities of spam differ from those sending large quantities of legitimate email."

Although the authors concede that theirs is only one study, the results raise the hope that anti-spam vendors will eventually be able to filter spam, at least in part, using its network characteristics as well as its content. If the researchers' principles hold up, they may pave the way for vendors to develop new, more accurate anti-spam tools.

— Tim Wilson, Site Editor, Dark Reading