MA-GRNN: A high-efficient modeling attack approach utilizing generalized regression neural network for XOR arbiter physical unclonable functions

In this paper, we propose a novel modeling attack approach to predict the responses of XOR arbiter physical unclonable functions (XOR APUFs), which improves prediction accuracy and reduces computational time. The high-dimensional mathematical model APUF is established its weakness analyzed. Furthermore, based on generalized regression neural network (MA-GRNN) introduced approximate APUFs. As proof-of-concept, four popular machine learning algorithms are utilized evaluate efficacy 3-XOR, 4-XOR, 5-XOR 6-XOR schemes. Experimental results show that MA-GRNN achieves high compared other three approaches while requiring less time simultaneously.