Exploring electrospun nanofibers for physically unclonable functions: a scalable and robust method toward unique identifiers
Abstract Optical physical unclonable functions (PUFs) have great potentials in the security identification of Internet of Things. In this work, electrospun nanofibers are proposed as a candidate for a nanoscale, robust, stable and scalable PUF. The dark-field reflectance images of the polymer fibers are quantitatively analyzed by Hough transform. We find that the fiber length and orientation distribution reach an optimal point as the fiber density grows up over 850 in 400 x 400 pixels for a polyvinylpyrrolidone nanofiber based PUF device. Subsequently, we test the robustness and randomness of the PUF pattern by using the fiber amount as an encoding feature, generating a reconstruction success rate over 80% and simultaneously an entropy of 260 bits within a mean size of 4 cm2. A scale-invariant algorithm is adopted to identify the uniqueness of each pattern on a 256-sensor device. Furthermore, thermo-, moisture as well as photostability of the authentication process are systematically investigated by comparing polyacrylonitrile to polyvinylpyrrolidone system.