Low-Overhead Implementation of a Soft Decision Helper Data Algorithm for SRAM PUFs
Using a Physically Unclonable Function or PUF to extract a secret key from the unique submicron structure of a device, instead of storing it in non-volatile memory, provides interesting advantages like physical unclonability and tamper evidence. However, an additional Helper Data Algorithm (HDA) is required to deal with the fuzziness of the PUF’s responses. To provide a viable alternative to costly protected non-volatile memory, the PUF+HDA construction should have a very low overhead. In this work, we propose the first HDA design using soft-decision information providing an implementation that occupies 44.8% less resources than previous proposals. Moreover, the required size of the used PUF can be reduced upto 58.4% due to the smaller entropy loss.