Abstract :
This paper describes software-based techniques for generating fingerprints that are capable of reliably differentiating, with a high level of confidence, identically configured personal computer (PC) hardware. These fingerprints are based on the characteristics of the quartz crystals that control the frequencies of various clock signals within the PC, specifically, the minute, unit-to-unit variations between the precise frequencies at which these crystals oscillate. A multi-factor fingerprint, based on measurements of the frequency differences between two or more independent pairs of crystals, was developed. The use of multiple independent sets of measurements increases the ability of these fingerprints to differentiate between nominally identical machines. The techniques, the proof-of-concept software that implements these techniques, and the experiments that are described here demonstrate the feasibility of developing reliable, highly discriminating, multi-factor fingerprints. These fingerprints could strengthen software protection measures that attempt to control the machines on which software may execute, as well as provide another forensics tool that will help pierce efforts to preserve anonymity.
