Two security researchers say they have discovered a technique for taking control of Bluetooth-enabled mobile phones, even when the handsets have security features switched on.
The technique is a practical implementation of a technique described by Ollie Whitehouse of security firm @Stake last year, which allows an attacker with specialised eqiupment to connect to a Bluetooth handset without authorisation. Once the connection is established, the attacker could make calls on the target's handset, siphon off data or listen in on data transfers between the device and, for example, a PC. Some security firms recommend financial traders avoid Bluetooth handsets because of the potential attack.
The original method required an attacker to listen in on the initial connection procedure between two Bluetooth devices - called "pairing" - which occurs only rarely. The new attack however allows an attacker to force two devices to repeat the pairing procedure, allowing the attacker to listen in and determine the identification code (PIN) used to protect the connection.
The researchers, senior lecturer Avishai Wool and graduate student Yaniv Shaked of Tel Aviv University's School of Electrical Engineering Systems, will present their paper, "Cracking the Bluetooth PIN", on Monday afternoon at the MobiSys conference in Seattle.
Various security holes have already appeared in Bluetooth, which is becoming widely used in mobile phones and high-end "smartphones". However, most require a poor implementation of Bluetooth's security features, or for the device to be left in "discoverable" mode. Whitehouse's attack, by contrast, could be used against a handset with security features switched on.
Whitehouse's attack is difficult to implement, because it requires the attacker to pick up some information during the pairing process. From this data, an attacker could determine the PIN for the connection, with the length of time depending on the number of digits in the PIN - under a second for four-digit PINs, which are standard on most devices.
Wool and Shaked's attack goes a step further, describing three methods for forcing a repeat of the pairing process. Using the information from this exchange, the researchers were able to determine the PIN in 0.06-.3 seconds for a 4-digit PIN, according to the paper.
For example, a user could be asked to re-enter the PIN number for connecting to his or her wireless headset, according to the paper. Once the two devices re-connected, the attacker would easily be able to crack the PIN in most cases. Many users could be fooled by this, since such re-pairing is built into the Bluetooth specification; in fact many devices have a mode requiring the user to re-enter their PIN each time a connection is made, the researchers said.
"Taken together, this is an impressive result," said security expert Bruce Schneier in a Weblog post.
Wool and Shaked recommend users refrain from entering their pairing PINs as much as possible, particularly in public places. Using longer PINs can also make a big difference, they said - even a six-digit PIN would take 10 seconds to crack, while a 10-digit number would require weeks, according to Whitehouse. Users may find they don't have a choice, since many devices only allow four-digit PINs.