Security researchers have developed a technique for remotely fingerprinting an electronic device using clock skews - small, microscopic deviations in device hardware. In a paper, Remote physical device fingerprinting, Tadayoshi Kohno, lead author and PhD student from the University of California San Diego, explains how the technique could be developed to track hardware wherever it is on the net, or in applications such as computer forensics.

Bruce Schneier, the noted cryptographer, described the approach as "nice work"; but the technique has already generated a fierce debate about its reliability. Mostly, the cryptographic community is discussing the ability of remote device fingerprinting to yield evidence good enough for a court of law - are clock skews stable on a particular PC? But there are wider concerns about device fingerprinting: taken at face value, the technology could end anonymous net access, a cause for celebration for repressive governments everywhere.

Kohno's approach is OS independent and relies on TCP Timestamps in TCP headers to "estimate a device's clock skew [drift] and thereby fingerprint a physical device". At 2000 packets per hour, this skew value has greater than six bits of entropy or enough to uniquely identify one computer in a million, according to Kohno's paper. TCP Timestamps can be turned off in Linux or BSD, but otherwise much data can be obtained from a great distance and without any need to modify a fingerprinted device.

"Our techniques report consistent measurements when the measurer is thousands of miles, multiple hops, and tens of milliseconds away from the fingerprinted device, and when the fingerprinted device is connected to the Internet from different locations and via different access technologies. Further, one can apply our passive and semi-passive techniques when the fingerprinted device is behind a NAT or firewall, and also when the device's system time is maintained via NTP or SNTP," the paper explains.

Kohno developed the research with Andre Broido and KC Claffy at the San Diego Supercomputer Center and presents their findings at the IEEE Symposium on Security and Privacy in California in May. ®

Related stories

Fingerprints as ID - good, bad, ugly?
Biometric sensors beaten senseless in tests
Forensic computing uncloaks industrial espionage
FBI retires Carnivore