Zero secrecy leakage for multiple enrollments of physical unclonable functions

We use physical unclonable functions (PUFs) to generate secret keys. We analyze the performance of the helper data scheme when the enrollment process is repeated multiple times. We show that codes exist such that the scheme remains secure after two enrollments, when all PUF observations are performed over the same channel. Furthermore, we show that a fuzzy commitment scheme remains secure after any number of enrollments, for PUF sources that meet a certain symmetry condition. We show that the temperature dependent model for SRAM-PUF meets this symmetry condition. Furthermore, we argue that many source-channel model pairs exist that meet the symmetry condition, and give some examples.