Public Hotspot, Public Vulnerability
By David Pescovitz, Thu Feb 24 08:00:00 GMT 2005

A new security protocol seeks to expose the evil twins and dodgy middlemen lurking in the shadows of wireless access points.

When you connect to a wireless hotspot at a cafe or airport, how do you know it's legit? Here's a scenario: You sit down with an espresso, fire up your laptop or Wi-Fi-equipped handheld, and the Web browser automatically goes to a familiar-looking login page for the wireless provider. You enter your username and password or, if you're a new user, your credit card number. After catching up on e-mail, you decide to check your stocks and pay some bills. At your bank's URL, you key in your account ID and password. What you don't realize is that the hotspot you logged into is actually an "evil twin," a rogue base station impersonating the real hotspot. Although the login screens look right, complete with "https://" in the address bar, they're actually "doppelganger windows." The moment you typed in your password or credit card number, you delivered them into the hands of the "phisher" patiently waiting to rip you off.

Phishers use technical spoofing and social engineering to trick potential victims into thinking that they're interacting with a legitimate Web site. For example, you've probably received at least a few e-mails purporting to be from PayPal and asking you to change your password because your account may have been the victim of a cyberattack. Of course, the reality is that the e-mail is itself an attack. Following the link takes you to a page that looks just like PayPal, but in reality is a phisher's net.

"Phishing exists in both wireless and wired settings," says cryptographer Markus Jakobsson, a professor at the Indiana University School of Informatics. "But it's a bit more difficult to protect against when you're using a public wireless access point and you can't be entirely sure of its identity."

While the the threat of evil twins may be overexaggerated by the mainstream media, Jakobsson is convinced that there is cause for concern. According to his bio, Jakobsson "teaches cryptography, security, protocol design, and likes to cheat." The combination of his professional practice and, well, appreciation for a good con helps him stay one step ahead of the phishers. His latest leap, announced at last week's American Association for the Advancement of Science conference in Washington, DC, is a cryptographic security protocol that he hopes will foil more than a few phishing trips.

Developed with professor Steve Myers, the Delayed Password Disclosure protocol fights identity theft by augmenting known authentication techniques. For example, existing "mutual authentication protocols," considered by many to be quite secure, work by determining if the user and server both know the same password without ever sharing the secret with each other. Authentication software acts as a third party, receiving the password that the user sends, checking it against the correct password that a bank or other Web service provider sends, and determining if access should be granted. In a wireless environment, the mutual authentication protocol is a relative easy target for an "evil twin" attack, Jakobsson says.

"The user thinks he and his bank are mutually authenticating, but he's really just entering his password in a doppelganger of the mutual authentication window that goes right to the attacker," Jakobsson explains. "The practical flaw of mutual authentication is that you don't know if you're running the right software. And so these kinds of attacks can be deployed both against novices and users who are fairly technically savvy."

Jakobsson and Meyer's approach adds a human component to the authentication process. When a customer signs up for an online service, Web bill pay for example, the password he selects is tied to a sequence of images. Each time he enters a letter in his password, the bank responds by displaying an image.

"If the user at any point does not recognize the images, then he stops entering the password," Jakobsson says.

If there were a million images to choose from during the registration, the odds are astronomical that a phisher operating a fake login screen would respond with the correct images in the right sequence.

The researchers have established a Web page that explains their patent-pending Delayed Password Disclosure protocol in depth. Within a few months, they hope to release a beta version of the protocol for PCs. The next step, Jakobsson says, is to tweak the code so that it will run on smartphones.

"Someone on a Symbian-based phone or the equivalent is as liable to be attacked with a fake browser window as someone on a traditional computer," Jakobsson says. "It's precisely the same problem but a phone has limited processing power so it requires a slightly different software optimization."

As major cities become blanketed with Wi-Fi, Jakobsson says that the key is to stay informed and bring your street smarts with you into the online world. After all, it's hard to spot a con artist in a crowded public space. Especially when they're disguised as your bank.

"I'm incredibly careful and so is my wife because I talk about this so much," he says. "But there have still been some attacks that almost caught us. If phishers can trick paranoid people, think about what can happen to the majority of people who are not at all suspicious."