A Deep Learning Modeling Attack Method for MISR-APUF Protection Structures
Wei Ge, Ji Quan Huang, Bo Liu, Min Zhu, Yuan Cao · 2018
With the continuous research of Arbiter Physical Unclonable Function (APUF), attack techniques for APUF are also emerging. Various protection structures have been proposed to increase the anti-modeling attack capability of APUF, due to its own linear delay network structure. The MISR-APUF structure as an effective anti-modeling attack structure can effectively resist the existing Machine Learning and Deep Learning attacks. This paper proposed a Package Prediction based Deep Learning (PPDL) modeling attack on MISR-APUF based on FPGA platform for the first time. The experimental results from the proposed PPDL method show the its high prediction accuracy, up to 98.76%.