Fast and Robust Speckle Pattern Authentication by Scale Invariant Feature Transform algorithm in Physical Unclonable Functions

TL;DR

The paper addresses secure authentication of optical Physical Unclonable Functions (PUFs) using speckle patterns, where traditional post-processing is sensitive to transformations. It introduces Scale-Invariant Feature Transform (SIFT) to extract abundant, invariant features from speckle CRP responses, enabling robust identification. Experiments on PS-PUF, PDLC-PUF, and TiO2-PUF demonstrate that SIFT remains effective under rotation, scaling, and cropping, with a practical authentication threshold around 100 matches and varying feature counts based on scattering strength. On multi-core hardware, verification times reach microsecond-scale performance, indicating strong potential for real-time, high-security anti-counterfeiting applications across diverse optical PUFs.

Abstract

Nowadays, due to the growing phenomenon of forgery in many fields, the interest in developing new anti-counterfeiting device and cryptography keys, based on the Physical Unclonable Functions (PUFs) paradigm, is widely increased. PUFs are physical hardware with an intrinsic, irreproducible disorder that allows for on-demand cryptographic key extraction. Among them, optical PUF are characterized by a large number of degrees of freedom resulting in higher security and higher sensitivity to environmental conditions. While these promising features led to the growth of advanced fabrication strategies and materials for new PUF devices, their combination with robust recognition algorithm remains largely unexplored. In this work, we present a metric-independent authentication approach that leverages the Scale Invariant Feature Transform (SIFT) algorithm to extract unique and invariant features from the speckle patterns generated by optical Physical Unclonable Functions (PUFs). The application of SIFT to the challenge response pairs (CRPs) protocol allows us to correctly authenticate a client while denying any other fraudulent access. In this way, the authentication process is highly reliable even in presence of response rotation, zooming, and cropping that may occur in consecutive PUF interrogations and to which other postprocessing algorithm are highly sensitive. This characteristics together with the speed of the method (tens of microseconds for each operation) broaden the applicability and reliability of PUF to practical high-security authentication or merchandise anti-counterfeiting.

Figures (6)

• Figure 1: SEM images and schematic representation of a) PS-PUF constituted by a single layer of polystyrene nanoparticles, b) PDLC-PUF constituted Polymer Dispersed Liquid Crystals PDLC-PUF and c) TiO$_2$-PUF constituted by TiO$_2$ nanoparticles dispersed in a dense polymer matrix. d-f) photographs of the proposed three PUFs.
• Figure 2: a) Schematic representation of the experimental setup used to collect the CRPs. b) Real picture of the main part of the setup used to generate and project the challenges. c) The challenge on the DMD, d) the projected challenge on the PUFs.
• Figure 3: a–b) Example images showing fully recognized speckles (produced under the same challenge conditions) and cases with no or few recognized points (from different speckles). c) Maps showing recognized points for a comparison of 20 versus 20 speckles, with varying matching distance (Md) parameter value. The scale is the same for the three matrices with a maximum number of recognized point of 600. d) Comparison between 200 versus 200 speckles from the same dataset (PS1). In the histogram the speckles that found their match are reported in green while the other one with no match are reported in red. e) Comparison between 200 versus 200 speckles from two datasets collected at different times (PS1 and PS2). In the histogram the speckles that found their match are reported in blue while the other one with no match are reported in red. f) Fractional Hamming Distance (FHD) distribution calculated over the larger database, illustrating the 'ideal-like,' 'like,' and 'unlike' distributions.
• Figure 4: Comparison, using SIFT analysis, of 1 speckle into a database of 100 speckles which is a) inside, b) outside the database. c) Comparison of speckle patterns, using SIFT analysis, by rotating them at 0°, 15°, 30°, 45°, 60° and 90°
• Figure 5: a) Comparison, using SIFT analysis, of 1 speckle (tag number $\#$100) into a database of 500 ones when scaled-up by 1.5, scaled-down by 0.8, cropped of 10$\%$ along the frame, cropped of 10$\%$ from the Top-Left (TL) corner, or from the right side and finally cropped of 20$\%$ in the center. Zero and full recognized points from scale-up tag (b,c) and cropping the surrounding frame by 10$\%$ (d-e).