Proof of Authenticity Statistics From Multiple Perceptual Hash Comparisons: A Case Study

Abstract The COVID-19 global pandemic created an optimal environment for counterfeiters to exploit vulnerabilities in the manufacturing industry. The decentralized and global nature of additive manufacturing (AM) systems created new attack vectors for counterfeiting due to ease of compromise of product and process information. To solve this challenge, innovative technologies and scientifically reliable methods for predicting authenticity are in great demand. In this work, a framework for differentiating between authentic and counterfeit AM automotive and aerospace components is proposed. Extant literature was reviewed and current anti-counterfeit technologies analyzed, informing the basis of the framework. The process was validated with a castle nut printed via selective laser-melting of a stainless steel M18 castle nut slightly modified with a Cantor dust fractal. The castle nut was then inspected with 225 and 450 kV X-ray tomography to assess the fidelity of fractal structure. A model trust anchor image sequence was developed and analyzed with a perceptual hash, Hamming distance computations, distribution functions, and null hypothesis for proof of authenticity. For authentic parts, a blockchain was updated with provenance. Future work will explore wider implementation with the goal of reducing prevalence of counterfeit parts in AM manufacturing systems.