Cryptographers have discovered a way to hack Bluetooth-enabled devices even when security features are switched on. The discovery may make it even easier for hackers to eavesdrop on conversations and charge their own calls to someone else’s cellphone.

Bluetooth is a protocol that allows different devices including phones, laptops, headsets and printers to communicate wirelessly over short ranges - typically between 10 and 100 metres.

Over the past few years security experts have devised many ways of hacking into Bluetooth communications, but most require the Bluetooth security features to be switched off.

In April 2004, UK-based Ollie Whitehouse, at that time working for security firm @Stake, showed that even Bluetooth devices in secure mode could be attacked. His method allowed someone to hijack the phone, giving them the power to make calls as if it were in their own hands.

Pairing up

But this technique did not pose a serious risk because it could be performed only if the hacker happened to catch two Bluetooth devices just before their first communication, during a process known as “pairing”.

Before two Bluetooth devices can communicate they must establish a secret key via this pairing process. But as long as the two devices paired up in a private place there was no risk of attack, explains Chris McNab of the UK security firm TrustMatta.

Now Avishai Wool and Yaniv Shaked of Tel Aviv University in Israel have worked out how to force devices to pair whenever they want. “Our attack makes it possible to crack every communication between two Bluetooth devices, and not only if it is the first communication between those devices,” says Shaked.

“Pairing allows you to seize control,” says Bruce Schneier, a security expert based in Mountain View, California. “You can sit on the train and make phone calls on someone else’s phone.”

Sniffing the airwaves

During pairing, two Bluetooth devices establish the 128-bit secret “link key” that they then store and use to encrypt all further communication. The first step requires the legitimate users to type the same secret, four-digit PIN into both devices. The two devices then use this PIN in a complex process to arrive at the common link key.

Whitehouse showed in 2004 that a hacker could arrive at this link key without knowing the PIN using a piece of equipment called a Bluetooth sniffer. This can record the exchanged messages being used to derive the link key and feed the recordings to software that knows the Bluetooth algorithms and can cycle through all 10,000 possibilities of the PIN. Once a hacker knows the link keys, Whitehouse reasoned they could hijack the device.

But pairing only occurs the first time two devices communicate. Wool and Shaked have managed to force pairing by pretending to be one of the two devices and sending a message to the other claiming to have forgotten the link key. This prompts the other device to discard the link key and the two then begin a new pairing session, which the hacker can then use.

Surprisingly easy

In order to send a “forget” message, the hacker must simply spoof one of the devices personal IDs, which can be done because all Bluetooth devices broadcast this automatically to any Bluetooth device within range.

“Having it done so easily is surprising,” says Schneier. He is also impressed by the fact that Wool and Shaked have actually implemented Whitehouse’s idea in real devices.

They show that once an attacker has forced two devices to pair, they can work out the link key in just 0.06 seconds on a Pentium IV-enabled computer, and 0.3 seconds on a Pentium-III. “This is not just a theoretical break, it’s practical,” says Schneier.

Shaked and Wool will present their findings at the MobiSys conference next Monday in Seattle, Washington, US.