CPG-FS: A CPU Performance Graph Based Device Fingerprint Scheme for Devices Identification and Authentication

Device fingerprint extracting has emerged as a popular but debatable technique to identify users by recognizing their devices. Two main drawbacks significantly affect the accuracy and stability of the whole process: the spoofing and the unstable device hardware or software features. In this paper, we propose a fingerprint framework mainly based on the hardware system ability of the tested devices. We have made an experiment to prove that every computer or device has different CPU graph when running specific programs (challenges) from other computers. The experiment results show that using the CPU graphs with our proposed comparison mechanism can achieve the high uniqueness and stability of the extracted fingerprint comparing with other fingerprint techniques. We also design a prototype model based on the device fingerprint and biological features to fix the drift problem.